portelli/Grid
1
0
Fork 0
mirror of https://github.com/paboyle/Grid.git synced 2024-11-10 07:55:35 +00:00
Code Releases Activity

Debugged the copy constructor of the Lattice class

Browse Source
This commit is contained in:
Guido Cossu 2016-07-06 15:31:00 +01:00
parent e3d5319470
commit e87182cf98
4 changed files with 888 additions and 849 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

51
lib/lattice/Lattice_base.h
View File

@ -1,32 +1,33 @@
/*************************************************************************************
/*************************************************************************************
Grid physics library, www.github.com/paboyle/Grid
Grid physics library, www.github.com/paboyle/Grid
Source file: ./lib/lattice/Lattice_base.h
Source file: ./lib/lattice/Lattice_base.h
Copyright (C) 2015
Copyright (C) 2015
Author: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Author: Peter Boyle <paboyle@ph.ed.ac.uk>
Author: paboyle <paboyle@ph.ed.ac.uk>
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation; either version 2 of the License, or
(at your option) any later version.
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation; either version 2 of the License, or
(at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License along
with this program; if not, write to the Free Software Foundation, Inc.,
51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
You should have received a copy of the GNU General Public License along
with this program; if not, write to the Free Software Foundation, Inc.,
51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
See the full license in the file "LICENSE" in the top level distribution directory
*************************************************************************************/
/* END LEGAL */
See the full license in the file "LICENSE" in the top level distribution
directory
*************************************************************************************/
/* END LEGAL */
#ifndef GRID_LATTICE_BASE_H
#define GRID_LATTICE_BASE_H
@ -255,6 +256,18 @@ PARALLEL_FOR_LOOP
checkerboard=0;
}
Lattice(const Lattice& r){ // copy constructor
_grid = r._grid;
checkerboard = r.checkerboard;
_odata.resize(_grid->oSites());// essential
PARALLEL_FOR_LOOP
for(int ss=0;ss<_grid->oSites();ss++){
_odata[ss]=r._odata[ss];
}
}
virtual ~Lattice(void) = default;
template<class sobj> strong_inline Lattice<vobj> & operator = (const sobj & r){
@ -267,7 +280,7 @@ PARALLEL_FOR_LOOP
template<class robj> strong_inline Lattice<vobj> & operator = (const Lattice<robj> & r){
this->checkerboard = r.checkerboard;
conformable(*this,r);
std::cout<<GridLogMessage<<"Lattice operator ="<<std::endl;
PARALLEL_FOR_LOOP
for(int ss=0;ss<_grid->oSites();ss++){
this->_odata[ss]=r._odata[ss];

250
lib/qcd/smearing/GaugeConfiguration.h
View File

@ -6,11 +6,11 @@
#ifndef GAUGE_CONFIG_
#define GAUGE_CONFIG_
namespace Grid {
namespace Grid {
namespace QCD {
namespace QCD {
/*!
/*!
@brief Smeared configuration container
It will behave like a configuration from the point of view of
@ -20,11 +20,11 @@
it, like smearing.
It stores a list of smeared configurations.
*/
template <class Gimpl>
class SmearedConfiguration {
*/
template <class Gimpl>
class SmearedConfiguration {
public:
INHERIT_GIMPL_TYPES(Gimpl) ;
INHERIT_GIMPL_TYPES(Gimpl);
private:
const unsigned int smearingLevels;
@ -33,34 +33,35 @@
// Member functions
//====================================================================
void fill_smearedSet(GaugeField& U){
ThinLinks = &U; //attach the smearing routine to the field U
void fill_smearedSet(GaugeField& U) {
ThinLinks = &U; // attach the smearing routine to the field U
//check the pointer is not null
if (ThinLinks==NULL)
std::cout << GridLogError << "[SmearedConfiguration] Error in ThinLinks pointer\n";
// check the pointer is not null
if (ThinLinks == NULL)
std::cout << GridLogError
<< "[SmearedConfiguration] Error in ThinLinks pointer\n";
if (smearingLevels > 0){
std::cout<< GridLogDebug << "[SmearedConfiguration] Filling SmearedSet\n";
if (smearingLevels > 0) {
std::cout << GridLogDebug
<< "[SmearedConfiguration] Filling SmearedSet\n";
GaugeField previous_u(ThinLinks->_grid);
previous_u = *ThinLinks;
for(int smearLvl = 0; smearLvl < smearingLevels; ++smearLvl){
StoutSmearing.smear(SmearedSet[smearLvl],previous_u);
for (int smearLvl = 0; smearLvl < smearingLevels; ++smearLvl) {
StoutSmearing.smear(SmearedSet[smearLvl], previous_u);
previous_u = SmearedSet[smearLvl];
// For debug purposes
RealD impl_plaq = WilsonLoops<Gimpl>::avgPlaquette(previous_u);
std::cout<< GridLogDebug << "[SmearedConfiguration] Plaq: " << impl_plaq<< std::endl;
std::cout << GridLogDebug
<< "[SmearedConfiguration] Plaq: " << impl_plaq << std::endl;
}
}
}
//====================================================================
GaugeField AnalyticSmearedForce(const GaugeField& SigmaKPrime,
const GaugeField& GaugeK) const{
GridBase *grid = GaugeK._grid;
}
//====================================================================
GaugeField AnalyticSmearedForce(const GaugeField& SigmaKPrime,
const GaugeField& GaugeK) const {
GridBase* grid = GaugeK._grid;
GaugeField C(grid), SigmaK(grid), iLambda(grid);
GaugeLinkField iLambda_mu(grid);
GaugeLinkField iQ(grid), e_iQ(grid);
@ -71,37 +72,33 @@ GaugeField AnalyticSmearedForce(const GaugeField& SigmaKPrime,
SigmaK = zero;
iLambda = zero;
for (int mu = 0; mu < Nd; mu++){
Cmu = peekLorentz( C,mu);
GaugeKmu = peekLorentz(GaugeK,mu);
SigmaKPrime_mu = peekLorentz(SigmaKPrime,mu);
iQ = Ta(Cmu*adj(GaugeKmu));
for (int mu = 0; mu < Nd; mu++) {
Cmu = peekLorentz(C, mu);
GaugeKmu = peekLorentz(GaugeK, mu);
SigmaKPrime_mu = peekLorentz(SigmaKPrime, mu);
iQ = Ta(Cmu * adj(GaugeKmu));
set_iLambda(iLambda_mu, e_iQ, iQ, SigmaKPrime_mu, GaugeKmu);
pokeLorentz(SigmaK, SigmaKPrime_mu*e_iQ + adj(Cmu)*iLambda_mu, mu);
pokeLorentz(SigmaK, SigmaKPrime_mu * e_iQ + adj(Cmu) * iLambda_mu, mu);
pokeLorentz(iLambda, iLambda_mu, mu);
}
StoutSmearing.derivative(SigmaK, iLambda, GaugeK);// derivative of SmearBase
StoutSmearing.derivative(SigmaK, iLambda,
GaugeK); // derivative of SmearBase
return SigmaK;
}
}
/*! @brief Returns smeared configuration at level 'Level' */
const GaugeField& get_smeared_conf(int Level) const{
/*! @brief Returns smeared configuration at level 'Level' */
const GaugeField& get_smeared_conf(int Level) const {
return SmearedSet[Level];
}
}
//====================================================================
void set_iLambda(GaugeLinkField& iLambda,
GaugeLinkField& e_iQ,
const GaugeLinkField& iQ,
const GaugeLinkField& Sigmap,
const GaugeLinkField& GaugeK)const{
GridBase *grid = iQ._grid;
//====================================================================
void set_iLambda(GaugeLinkField& iLambda, GaugeLinkField& e_iQ,
const GaugeLinkField& iQ, const GaugeLinkField& Sigmap,
const GaugeLinkField& GaugeK) const {
GridBase* grid = iQ._grid;
GaugeLinkField iQ2(grid), iQ3(grid), B1(grid), B2(grid), USigmap(grid);
GaugeLinkField unity(grid);
unity=1.0;
unity = 1.0;
LatticeComplex u(grid), w(grid);
LatticeComplex f0(grid), f1(grid), f2(grid);
@ -110,7 +107,8 @@ void set_iLambda(GaugeLinkField& iLambda,
LatticeComplex emiu(grid), e2iu(grid), qt(grid), fden(grid);
LatticeComplex r01(grid), r11(grid), r21(grid), r02(grid), r12(grid);
LatticeComplex r22(grid), tr1(grid), tr2(grid);
LatticeComplex b10(grid), b11(grid), b12(grid), b20(grid), b21(grid), b22(grid);
LatticeComplex b10(grid), b11(grid), b12(grid), b20(grid), b21(grid),
b22(grid);
LatticeComplex LatticeUnitComplex(grid);
LatticeUnitComplex = 1.0;
@ -118,9 +116,9 @@ void set_iLambda(GaugeLinkField& iLambda,
// Exponential
iQ2 = iQ * iQ;
iQ3 = iQ * iQ2;
StoutSmearing.set_uw(u,w,iQ2,iQ3);
StoutSmearing.set_fj(f0,f1,f2,u,w);
e_iQ = f0*unity + timesMinusI(f1) * iQ - f2 * iQ2;
StoutSmearing.set_uw(u, w, iQ2, iQ3);
StoutSmearing.set_fj(f0, f1, f2, u, w);
e_iQ = f0 * unity + timesMinusI(f1) * iQ - f2 * iQ2;
// Getting B1, B2, Gamma and Lambda
// simplify this part, reduntant calculations in set_fj
@ -131,36 +129,36 @@ void set_iLambda(GaugeLinkField& iLambda,
cosw = cos(w);
emiu = cos(u) - timesI(sin(u));
e2iu = cos(2.0*u) + timesI(sin(2.0*u));
e2iu = cos(2.0 * u) + timesI(sin(2.0 * u));
r01 = (2.0*u + timesI(2.0*(u2-w2))) * e2iu
+ emiu * ((16.0*u*cosw + 2.0*u*(3.0*u2+w2)*xi0) +
timesI(-8.0*u2*cosw + 2.0*(9.0*u2+w2)*xi0));
r01 = (2.0 * u + timesI(2.0 * (u2 - w2))) * e2iu +
emiu * ((16.0 * u * cosw + 2.0 * u * (3.0 * u2 + w2) * xi0) +
timesI(-8.0 * u2 * cosw + 2.0 * (9.0 * u2 + w2) * xi0));
r11 = (2.0*LatticeUnitComplex + timesI(4.0*u))* e2iu
+ emiu * ((-2.0*cosw + (3.0*u2-w2)*xi0) +
timesI((2.0*u*cosw + 6.0*u*xi0)));
r11 = (2.0 * LatticeUnitComplex + timesI(4.0 * u)) * e2iu +
emiu * ((-2.0 * cosw + (3.0 * u2 - w2) * xi0) +
timesI((2.0 * u * cosw + 6.0 * u * xi0)));
r21 = 2.0*timesI(e2iu)
+ emiu * (-3.0*u*xi0 + timesI(cosw - 3.0*xi0));
r21 =
2.0 * timesI(e2iu) + emiu * (-3.0 * u * xi0 + timesI(cosw - 3.0 * xi0));
r02 = -2.0 * e2iu +
emiu * (-8.0 * u2 * xi0 +
timesI(2.0 * u * (cosw + xi0 + 3.0 * u2 * xi1)));
r02 = -2.0 * e2iu + emiu * (-8.0*u2*xi0 +
timesI(2.0*u*(cosw + xi0 + 3.0*u2*xi1)));
r12 = emiu * (2.0 * u * xi0 + timesI(-cosw - xi0 + 3.0 * u2 * xi1));
r12 = emiu * (2.0*u*xi0 + timesI(-cosw - xi0 + 3.0*u2*xi1));
r22 = emiu * (xi0 - timesI(3.0 * u * xi1));
r22 = emiu * (xi0 - timesI(3.0*u*xi1));
fden = LatticeUnitComplex / (2.0 * (9.0 * u2 - w2) * (9.0 * u2 - w2));
fden = LatticeUnitComplex/(2.0*(9.0*u2-w2)*(9.0*u2-w2));
b10 = 2.0 * u * r01 + (3.0 * u2 - w2) * r02 - (30.0 * u2 + 2.0 * w2) * f0;
b11 = 2.0 * u * r11 + (3.0 * u2 - w2) * r12 - (30.0 * u2 + 2.0 * w2) * f1;
b12 = 2.0 * u * r21 + (3.0 * u2 - w2) * r22 - (30.0 * u2 + 2.0 * w2) * f2;
b10 = 2.0 * u * r01 + (3.0* u2 - w2)*r02 - (30.0 * u2 + 2.0 * w2)*f0;
b11 = 2.0 * u * r11 + (3.0* u2 - w2)*r12 - (30.0 * u2 + 2.0 * w2)*f1;
b12 = 2.0 * u * r21 + (3.0* u2 - w2)*r22 - (30.0 * u2 + 2.0 * w2)*f2;
b20 = r01 - (3.0*u)*r02 - (24.0*u)*f0;
b21 = r11 - (3.0*u)*r12 - (24.0*u)*f1;
b22 = r21 - (3.0*u)*r22 - (24.0*u)*f2;
b20 = r01 - (3.0 * u) * r02 - (24.0 * u) * f0;
b21 = r11 - (3.0 * u) * r12 - (24.0 * u) * f1;
b22 = r21 - (3.0 * u) * r22 - (24.0 * u) * f2;
b10 *= fden;
b11 *= fden;
@ -169,13 +167,12 @@ void set_iLambda(GaugeLinkField& iLambda,
b21 *= fden;
b22 *= fden;
B1 = b10*unity + timesMinusI(b11) * iQ - b12 * iQ2;
B2 = b20*unity + timesMinusI(b21) * iQ - b22 * iQ2;
B1 = b10 * unity + timesMinusI(b11) * iQ - b12 * iQ2;
B2 = b20 * unity + timesMinusI(b21) * iQ - b22 * iQ2;
USigmap = GaugeK * Sigmap;
tr1 = trace(USigmap*B1);
tr2 = trace(USigmap*B2);
tr1 = trace(USigmap * B1);
tr2 = trace(USigmap * B2);
GaugeLinkField QUS = iQ * USigmap;
GaugeLinkField USQ = USigmap * iQ;
@ -184,99 +181,82 @@ void set_iLambda(GaugeLinkField& iLambda,
timesI(f1) * USigmap + f2 * QUS + f2 * USQ;
iLambda = Ta(iGamma);
}
}
//====================================================================
public:
GaugeField* ThinLinks; /*!< @brief Pointer to the thin
//====================================================================
public:
GaugeField*
ThinLinks; /*!< @brief Pointer to the thin
links configuration */
/*! @brief Standard constructor */
SmearedConfiguration(GridCartesian * UGrid,
unsigned int Nsmear,
Smear_Stout<Gimpl>& Stout):
smearingLevels(Nsmear),
StoutSmearing(Stout),
ThinLinks(NULL){
for (unsigned int i=0; i< smearingLevels; ++i)
SmearedConfiguration(GridCartesian* UGrid, unsigned int Nsmear,
Smear_Stout<Gimpl>& Stout)
: smearingLevels(Nsmear), StoutSmearing(Stout), ThinLinks(NULL) {
for (unsigned int i = 0; i < smearingLevels; ++i)
SmearedSet.push_back(*(new GaugeField(UGrid)));
}
/*! For just thin links */
SmearedConfiguration():
smearingLevels(0),
StoutSmearing(),
SmearedSet(),
ThinLinks(NULL){}
SmearedConfiguration()
: smearingLevels(0), StoutSmearing(), SmearedSet(), ThinLinks(NULL) {}
// attach the smeared routines to the thin links U and fill the smeared set
void set_GaugeField(GaugeField& U){ fill_smearedSet(U);}
void set_GaugeField(GaugeField& U) { fill_smearedSet(U); }
//====================================================================
void smeared_force(GaugeField& SigmaTilde) const{
if (smearingLevels > 0){
GaugeField force(SigmaTilde._grid);
//====================================================================
void smeared_force(GaugeField& SigmaTilde) const {
if (smearingLevels > 0) {
GaugeField force = SigmaTilde; // actually = U*SigmaTilde
GaugeLinkField tmp_mu(SigmaTilde._grid);
force = SigmaTilde;//actually = U*SigmaTilde
for (int mu = 0; mu < Nd; mu++){
for (int mu = 0; mu < Nd; mu++) {
// to get just SigmaTilde
tmp_mu = adj(peekLorentz(SmearedSet[smearingLevels-1], mu)) * peekLorentz(force,mu);
tmp_mu = adj(peekLorentz(SmearedSet[smearingLevels - 1], mu)) *
peekLorentz(force, mu);
pokeLorentz(force, tmp_mu, mu);
}
for(int ismr = smearingLevels - 1; ismr > 0; --ismr)
force = AnalyticSmearedForce(force,get_smeared_conf(ismr-1));
for (int ismr = smearingLevels - 1; ismr > 0; --ismr)
force = AnalyticSmearedForce(force, get_smeared_conf(ismr - 1));
force = AnalyticSmearedForce(force,*ThinLinks);
force = AnalyticSmearedForce(force, *ThinLinks);
for (int mu = 0; mu < Nd; mu++){
for (int mu = 0; mu < Nd; mu++) {
tmp_mu = peekLorentz(*ThinLinks, mu) * peekLorentz(force, mu);
pokeLorentz(SigmaTilde, tmp_mu, mu);
}
}// if smearingLevels = 0 do nothing
}
} // if smearingLevels = 0 do nothing
}
//====================================================================
GaugeField& get_SmearedU() { return SmearedSet[smearingLevels - 1]; }
GaugeField& get_SmearedU(){
return SmearedSet[smearingLevels-1];
}
GaugeField& get_U(bool smeared=false) {
GaugeField& get_U(bool smeared = false) {
// get the config, thin links by default
if (smeared){
if (smearingLevels){
RealD impl_plaq = WilsonLoops<Gimpl>::avgPlaquette(SmearedSet[smearingLevels-1]);
std::cout<< GridLogDebug << "getting Usmr Plaq: " << impl_plaq<< std::endl;
if (smeared) {
if (smearingLevels) {
RealD impl_plaq =
WilsonLoops<Gimpl>::avgPlaquette(SmearedSet[smearingLevels - 1]);
std::cout << GridLogDebug << "getting Usmr Plaq: " << impl_plaq
<< std::endl;
return get_SmearedU();
}
else {
} else {
RealD impl_plaq = WilsonLoops<Gimpl>::avgPlaquette(*ThinLinks);
std::cout<< GridLogDebug << "getting Thin Plaq: " << impl_plaq<< std::endl;
std::cout << GridLogDebug << "getting Thin Plaq: " << impl_plaq
<< std::endl;
return *ThinLinks;
}
} else {
RealD impl_plaq = WilsonLoops<Gimpl>::avgPlaquette(*ThinLinks);
std::cout << GridLogDebug << "getting Thin Plaq: " << impl_plaq
<< std::endl;
return *ThinLinks;
}
}
else{
RealD impl_plaq = WilsonLoops<Gimpl>::avgPlaquette(*ThinLinks);
std::cout<< GridLogDebug << "getting Thin Plaq: " << impl_plaq<< std::endl;
return *ThinLinks;}
}
};
}
}
#endif

36
scripts/copyright
View File

@ -5,13 +5,13 @@ while (( "$#" )); do
echo $1
cat > message <<EOF
/*************************************************************************************
/*************************************************************************************
Grid physics library, www.github.com/paboyle/Grid
Grid physics library, www.github.com/paboyle/Grid
Source file: $1
Source file: $1
Copyright (C) 2015
Copyright (C) 2015
EOF
@ -19,23 +19,23 @@ git log $1 | grep Author | sort -u >> message
cat >> message <<EOF
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation; either version 2 of the License, or
(at your option) any later version.
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation; either version 2 of the License, or
(at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License along
with this program; if not, write to the Free Software Foundation, Inc.,
51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
You should have received a copy of the GNU General Public License along
with this program; if not, write to the Free Software Foundation, Inc.,
51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
See the full license in the file "LICENSE" in the top level distribution directory
*************************************************************************************/
/* END LEGAL */
See the full license in the file "LICENSE" in the top level distribution directory
*************************************************************************************/
/* END LEGAL */
EOF
cat message > tmp.fil

708
tests/Test_main.cc
View File

@ -1,89 +1,91 @@
/*************************************************************************************
/*************************************************************************************
Grid physics library, www.github.com/paboyle/Grid
Grid physics library, www.github.com/paboyle/Grid
Source file: ./tests/Test_main.cc
Source file: ./tests/Test_main.cc
Copyright (C) 2015
Copyright (C) 2015
Author: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Author: Peter Boyle <paboyle@ph.ed.ac.uk>
Author: neo <cossu@post.kek.jp>
Author: paboyle <paboyle@ph.ed.ac.uk>
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation; either version 2 of the License, or
(at your option) any later version.
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation; either version 2 of the License, or
(at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License along
with this program; if not, write to the Free Software Foundation, Inc.,
51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
You should have received a copy of the GNU General Public License along
with this program; if not, write to the Free Software Foundation, Inc.,
51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
See the full license in the file "LICENSE" in the top level distribution directory
*************************************************************************************/
/* END LEGAL */
See the full license in the file "LICENSE" in the top level distribution
directory
*************************************************************************************/
/* END LEGAL */
#include "Grid.h"
using namespace std;
using namespace Grid;
using namespace Grid::QCD;
/*
Grid_main.cc(232): error: no suitable user-defined conversion from "Grid::iScalar<Grid::iMatrix<Grid::iScalar<Grid::Complex>, 4>>" to "const Grid::iScalar<Grid::iScalar<Grid::iMatrix<Grid::Complex, 3>>>" exists
Grid_main.cc(232): error: no suitable user-defined conversion from
"Grid::iScalar<Grid::iMatrix<Grid::iScalar<Grid::Complex>, 4>>" to "const
Grid::iScalar<Grid::iScalar<Grid::iMatrix<Grid::Complex, 3>>>" exists
c_m = peekIdiot<SpinColourMatrix>(scm,1,2);
*/
template<class vobj> auto peekIdiot(const vobj &rhs,int i,int j) -> decltype(peekIndex<2>(rhs,0,0))
{
return peekIndex<2>(rhs,i,j);
template <class vobj>
auto peekIdiot(const vobj &rhs, int i, int j)
-> decltype(peekIndex<2>(rhs, 0, 0)) {
return peekIndex<2>(rhs, i, j);
}
template<class vobj> auto peekDumKopf(const vobj &rhs,int i,int j) -> decltype(peekIndex<3>(rhs,0,0))
{
return peekIndex<3>(rhs,i,j);
template <class vobj>
auto peekDumKopf(const vobj &rhs, int i, int j)
-> decltype(peekIndex<3>(rhs, 0, 0)) {
return peekIndex<3>(rhs, i, j);
}
template<class vobj> auto peekDumKopf(const vobj &rhs,int i) -> decltype(peekIndex<3>(rhs,0))
{
return peekIndex<3>(rhs,i);
template <class vobj>
auto peekDumKopf(const vobj &rhs, int i) -> decltype(peekIndex<3>(rhs, 0)) {
return peekIndex<3>(rhs, i);
}
int main (int argc, char ** argv)
{
Grid_init(&argc,&argv);
int main(int argc, char **argv) {
Grid_init(&argc, &argv);
std::vector<int> latt_size = GridDefaultLatt();
std::vector<int> simd_layout = GridDefaultSimd(4,vComplex::Nsimd());
std::vector<int> simd_layout = GridDefaultSimd(4, vComplex::Nsimd());
std::vector<int> mpi_layout = GridDefaultMpi();
latt_size.resize(4);
#ifdef AVX512
for(int omp=128;omp<236;omp+=16){
for (int omp = 128; omp < 236; omp += 16) {
#else
for(int omp=1;omp<2;omp*=20){
for (int omp = 1; omp < 2; omp *= 20) {
#endif
#ifdef OMP
omp_set_num_threads(omp);
#endif
for(int lat=8;lat<=16;lat+=40){
std::cout << "Lat "<<lat<<std::endl;
for (int lat = 8; lat <= 16; lat += 40) {
std::cout << "Lat " << lat << std::endl;
latt_size[0] = lat;
latt_size[1] = lat;
latt_size[2] = lat;
latt_size[3] = lat;
double volume = latt_size[0]*latt_size[1]*latt_size[2]*latt_size[3];
double volume = latt_size[0] * latt_size[1] * latt_size[2] * latt_size[3];
GridCartesian Fine(latt_size,simd_layout,mpi_layout);
GridRedBlackCartesian rbFine(latt_size,simd_layout,mpi_layout);
GridCartesian Fine(latt_size, simd_layout, mpi_layout);
GridRedBlackCartesian rbFine(latt_size, simd_layout, mpi_layout);
GridParallelRNG FineRNG(&Fine);
GridSerialRNG SerialRNG;
GridSerialRNG SerialRNG1;
@ -91,19 +93,19 @@ int main (int argc, char ** argv)
FineRNG.SeedRandomDevice();
SerialRNG.SeedRandomDevice();
std::cout <<"SerialRNG" << SerialRNG._generators[0] <<std::endl;
std::cout << "SerialRNG" << SerialRNG._generators[0] << std::endl;
std::vector<typename GridSerialRNG::RngStateType> saved;
SerialRNG.GetState(saved,0);
SerialRNG1.SetState(saved,0);
SerialRNG.GetState(saved, 0);
SerialRNG1.SetState(saved, 0);
RealD dd1,dd2;
RealD dd1, dd2;
std::cout << "Testing RNG state save restore"<<std::endl;
for(int i=0;i<10;i++){
random(SerialRNG,dd1);
random(SerialRNG1,dd2);
std::cout << "i "<<i<<" "<<dd1<< " " <<dd2<<std::endl;
std::cout << "Testing RNG state save restore" << std::endl;
for (int i = 0; i < 10; i++) {
random(SerialRNG, dd1);
random(SerialRNG1, dd2);
std::cout << "i " << i << " " << dd1 << " " << dd2 << std::endl;
}
LatticeColourMatrix Foo(&Fine);
LatticeColourMatrix Bar(&Fine);
@ -132,7 +134,6 @@ int main (int argc, char ** argv)
LatticeLorentzColourMatrix lcMat(&Fine);
LatticeComplex scalar(&Fine);
LatticeReal rscalar(&Fine);
LatticeReal iscalar(&Fine);
@ -141,82 +142,96 @@ int main (int argc, char ** argv)
iSpinMatrix<vComplex> iGammaFive;
ColourMatrix cmat;
random(FineRNG,Foo);
gaussian(FineRNG,Bar);
random(FineRNG,scFoo);
random(FineRNG,scBar);
random(FineRNG,cMat);
random(FineRNG,sMat);
random(FineRNG,scMat);
random(FineRNG,lcMat);
random(FineRNG,cVec);
random(FineRNG,sVec);
random(FineRNG,scVec);
random(FineRNG, Foo);
gaussian(FineRNG, Bar);
random(FineRNG, scFoo);
random(FineRNG, scBar);
random(FineRNG, cMat);
random(FineRNG, sMat);
random(FineRNG, scMat);
random(FineRNG, lcMat);
random(FineRNG, cVec);
random(FineRNG, sVec);
random(FineRNG, scVec);
fflush(stdout);
LatticeColourMatrix newFoo = Foo;
// confirm correctness of copy constructor
Bar = Foo - newFoo;
std::cout << "Copy constructor diff check: ";
double test_cc = norm2(Bar);
if (test_cc < 1e-5){
std::cout << "OK\n";
}
else{
std::cout << "fail\n";
abort();
}
TComplex tr = trace(cmat);
cVec = cMat * cVec; // LatticeColourVector = LatticeColourMatrix
// * LatticeColourVector
sVec = sMat * sVec; // LatticeSpinVector = LatticeSpinMatrix
// * LatticeSpinVector
scVec = scMat * scVec; // LatticeSpinColourVector =
// LatticeSpinColourMatrix *
// LatticeSpinColourVector
scVec = cMat * scVec; // LatticeSpinColourVector = LatticeColourMatrix
// * LatticeSpinColourVector
scVec = sMat * scVec; // LatticeSpinColourVector = LatticeSpinMatrix
// * LatticeSpinColourVector
cVec = cMat * cVec; // LatticeColourVector = LatticeColourMatrix * LatticeColourVector
sVec = sMat * sVec; // LatticeSpinVector = LatticeSpinMatrix * LatticeSpinVector
scVec= scMat * scVec;// LatticeSpinColourVector = LatticeSpinColourMatrix * LatticeSpinColourVector
scVec= cMat * scVec; // LatticeSpinColourVector = LatticeColourMatrix * LatticeSpinColourVector
scVec= sMat * scVec; // LatticeSpinColourVector = LatticeSpinMatrix * LatticeSpinColourVector
cMat = outerProduct(cVec,cVec);
scalar = localInnerProduct(cVec,cVec);
cMat = Ta(cMat); //traceless antihermitian
cMat = outerProduct(cVec, cVec);
scalar = localInnerProduct(cVec, cVec);
cMat = Ta(cMat); // traceless antihermitian
scalar += scalar;
scalar -= scalar;
scalar *= scalar;
add(scalar,scalar,scalar);
sub(scalar,scalar,scalar);
mult(scalar,scalar,scalar);
add(scalar, scalar, scalar);
sub(scalar, scalar, scalar);
mult(scalar, scalar, scalar);
mac(scalar,scalar,scalar);
scalar = scalar+scalar;
scalar = scalar-scalar;
scalar = scalar*scalar;
mac(scalar, scalar, scalar);
scalar = scalar + scalar;
scalar = scalar - scalar;
scalar = scalar * scalar;
scalar=outerProduct(scalar,scalar);
scalar = outerProduct(scalar, scalar);
scalar=adj(scalar);
scalar = adj(scalar);
// rscalar=real(scalar);
// iscalar=imag(scalar);
// scalar =cmplx(rscalar,iscalar);
PokeIndex<ColourIndex>(cVec,scalar,1);
PokeIndex<ColourIndex>(cVec, scalar, 1);
scalar = transpose(scalar);
scalar = TransposeIndex<ColourIndex>(scalar);
scalar = TraceIndex<SpinIndex>(scalar);
scalar = PeekIndex<ColourIndex>(cVec, 0);
scalar=transpose(scalar);
scalar=TransposeIndex<ColourIndex>(scalar);
scalar=TraceIndex<SpinIndex>(scalar);
scalar=PeekIndex<ColourIndex>(cVec,0);
scalar = trace(scalar);
scalar = localInnerProduct(cVec, cVec);
scalar = localNorm2(cVec);
scalar=trace(scalar);
scalar=localInnerProduct(cVec,cVec);
scalar=localNorm2(cVec);
// -=,+=,*=,()
// add,+,sub,-,mult,mac,*
// adj,conjugate
// real,imag
// transpose,transposeIndex
// trace,traceIndex
// peekIndex
// innerProduct,outerProduct,
// localNorm2
// localInnerProduct
scMat = sMat*scMat; // LatticeSpinColourMatrix = LatticeSpinMatrix * LatticeSpinColourMatrix
// -=,+=,*=,()
// add,+,sub,-,mult,mac,*
// adj,conjugate
// real,imag
// transpose,transposeIndex
// trace,traceIndex
// peekIndex
// innerProduct,outerProduct,
// localNorm2
// localInnerProduct
scMat = sMat * scMat; // LatticeSpinColourMatrix = LatticeSpinMatrix
// * LatticeSpinColourMatrix
///////////////////////
// Non-lattice (const objects) * Lattice
@ -224,103 +239,103 @@ int main (int argc, char ** argv)
SpinColourMatrix scm;
vSpinColourMatrix vscm;
Complex cplx(1.0);
Integer myint=1;
double mydouble=1.0;
Integer myint = 1;
double mydouble = 1.0;
// vSpinColourMatrix vscm;
scMat = cMat*scMat;
scm = cm * scm; // SpinColourMatrix = ColourMatrix * SpinColourMatrix
scm = scm *cm; // SpinColourMatrix = SpinColourMartix * ColourMatrix
scm = GammaFive * scm ; // SpinColourMatrix = SpinMatrix * SpinColourMatrix
scm = scm* GammaFive ; // SpinColourMatrix = SpinColourMatrix * SpinMatrix
scMat = cMat * scMat;
scm =
cm * scm; // SpinColourMatrix = ColourMatrix * SpinColourMatrix
scm = scm * cm; // SpinColourMatrix = SpinColourMartix * ColourMatrix
scm = GammaFive *
scm; // SpinColourMatrix = SpinMatrix * SpinColourMatrix
scm =
scm * GammaFive; // SpinColourMatrix = SpinColourMatrix * SpinMatrix
scm = scm*cplx;
vscm = vscm*cplx;
scMat = scMat*cplx;
scm = scm * cplx;
vscm = vscm * cplx;
scMat = scMat * cplx;
scm = cplx*scm;
vscm = cplx*vscm;
scMat = cplx*scMat;
scm = myint*scm;
vscm = myint*vscm;
scMat = scMat*myint;
scm = cplx * scm;
vscm = cplx * vscm;
scMat = cplx * scMat;
scm = myint * scm;
vscm = myint * vscm;
scMat = scMat * myint;
scm = scm*mydouble;
vscm = vscm*mydouble;
scMat = scMat*mydouble;
scMat = mydouble*scMat;
cMat = mydouble*cMat;
scm = scm * mydouble;
vscm = vscm * mydouble;
scMat = scMat * mydouble;
scMat = mydouble * scMat;
cMat = mydouble * cMat;
sMat = adj(sMat); // LatticeSpinMatrix adjoint
sMat = iGammaFive*sMat; // SpinMatrix * LatticeSpinMatrix
sMat = GammaFive*sMat; // SpinMatrix * LatticeSpinMatrix
scMat= adj(scMat);
cMat= adj(cMat);
cm=adj(cm);
scm=adj(scm);
scm=transpose(scm);
scm=transposeIndex<1>(scm);
sMat = iGammaFive * sMat; // SpinMatrix * LatticeSpinMatrix
sMat = GammaFive * sMat; // SpinMatrix * LatticeSpinMatrix
scMat = adj(scMat);
cMat = adj(cMat);
cm = adj(cm);
scm = adj(scm);
scm = transpose(scm);
scm = transposeIndex<1>(scm);
random(SerialRNG, cm);
std::cout<<GridLogMessage << cm << std::endl;
std::cout << GridLogMessage << cm << std::endl;
cm = Ta(cm);
TComplex tracecm= trace(cm);
std::cout<<GridLogMessage << cm << std::endl;
TComplex tracecm = trace(cm);
std::cout << GridLogMessage << cm << std::endl;
cm = Exponentiate(cm, 2.0, 12);
std::cout<<GridLogMessage << cm << " " << std::endl;
std::cout << GridLogMessage << cm << " " << std::endl;
Complex det = Determinant(cm);
std::cout<<GridLogMessage << "determinant: " << det << std::endl;
std::cout<<GridLogMessage << "norm: " << norm2(cm) << std::endl;
std::cout << GridLogMessage << "determinant: " << det << std::endl;
std::cout << GridLogMessage << "norm: " << norm2(cm) << std::endl;
cm = ProjectOnGroup(cm);
std::cout<<GridLogMessage << cm << " " << std::endl;
std::cout<<GridLogMessage << "norm: " << norm2(cm) << std::endl;
std::cout << GridLogMessage << cm << " " << std::endl;
std::cout << GridLogMessage << "norm: " << norm2(cm) << std::endl;
cm = ProjectOnGroup(cm);
std::cout<<GridLogMessage << cm << " " << std::endl;
std::cout<<GridLogMessage << "norm: " << norm2(cm) << std::endl;
std::cout << GridLogMessage << cm << " " << std::endl;
std::cout << GridLogMessage << "norm: " << norm2(cm) << std::endl;
// det = Determinant(cm);
// std::cout<<GridLogMessage << "determinant: " << det << std::endl;
// Foo = Foo+scalar; // LatticeColourMatrix+Scalar
// Foo = Foo*scalar; // LatticeColourMatrix*Scalar
// Foo = Foo-scalar; // LatticeColourMatrix-Scalar
// Foo = scalar*Foo; // Scalar*LatticeColourMatrix
// Foo = scalar+Foo; // Scalar+LatticeColourMatrix
// Foo = scalar-Foo; // Scalar-LatticeColourMatrix
// Foo = Foo+scalar; // LatticeColourMatrix+Scalar
// Foo = Foo*scalar; // LatticeColourMatrix*Scalar
// Foo = Foo-scalar; // LatticeColourMatrix-Scalar
// Foo = scalar*Foo; // Scalar*LatticeColourMatrix
// Foo = scalar+Foo; // Scalar+LatticeColourMatrix
// Foo = scalar-Foo; // Scalar-LatticeColourMatrix
LatticeComplex trscMat(&Fine);
trscMat = trace(scMat); // Trace
// Exponentiate test
std::vector<int> mysite {0,0,0,0};
random(FineRNG,cMat);
std::vector<int> mysite{0, 0, 0, 0};
random(FineRNG, cMat);
cMat = Ta(cMat);
peekSite(cm, cMat, mysite);
std::cout<<GridLogMessage << cm << " " << std::endl;
std::cout << GridLogMessage << cm << " " << std::endl;
cm = Exponentiate(cm, 1.0, 12);
std::cout<<GridLogMessage << cm << " " << std::endl;
std::cout<<GridLogMessage << "norm: " << norm2(cm) << std::endl;
std::cout << GridLogMessage << cm << " " << std::endl;
std::cout << GridLogMessage << "norm: " << norm2(cm) << std::endl;
std::cout<<GridLogMessage << "norm cMmat : " << norm2(cMat) << std::endl;
std::cout << GridLogMessage << "norm cMmat : " << norm2(cMat)
<< std::endl;
cMat = expMat(cMat, ComplexD(1.0, 0.0));
std::cout<<GridLogMessage << "norm expMat: " << norm2(cMat) << std::endl;
std::cout << GridLogMessage << "norm expMat: " << norm2(cMat)
<< std::endl;
peekSite(cm, cMat, mysite);
std::cout<<GridLogMessage << cm << " " << std::endl;
std::cout<<GridLogMessage << "determinant: " << Determinant(cm) << std::endl;
std::cout<<GridLogMessage << "norm: " << norm2(cm) << std::endl;
std::cout << GridLogMessage << cm << " " << std::endl;
std::cout << GridLogMessage << "determinant: " << Determinant(cm)
<< std::endl;
std::cout << GridLogMessage << "norm: " << norm2(cm) << std::endl;
// LatticeComplex trlcMat(&Fine);
// trlcMat = trace(lcMat); // Trace involving iVector - now generates error
// trlcMat = trace(lcMat); // Trace involving iVector - now generates
// error
{ // Peek-ology and Poke-ology, with a little app-ology
Complex c;
@ -328,25 +343,29 @@ int main (int argc, char ** argv)
SpinMatrix s_m;
SpinColourMatrix sc_m;
s_m = TensorIndexRecursion<ColourIndex>::traceIndex(sc_m); // Map to traceColour
c_m = TensorIndexRecursion<SpinIndex>::traceIndex(sc_m); // map to traceSpin
s_m = TensorIndexRecursion<ColourIndex>::traceIndex(
sc_m); // Map to traceColour
c_m = TensorIndexRecursion<SpinIndex>::traceIndex(
sc_m); // map to traceSpin
c = TensorIndexRecursion<SpinIndex>::traceIndex(s_m);
c = TensorIndexRecursion<ColourIndex>::traceIndex(c_m);
s_m = TensorIndexRecursion<ColourIndex>::peekIndex(scm,0,0);
c_m = TensorIndexRecursion<SpinIndex>::peekIndex(scm,1,2);
s_m = TensorIndexRecursion<ColourIndex>::peekIndex(scm, 0, 0);
c_m = TensorIndexRecursion<SpinIndex>::peekIndex(scm, 1, 2);
// c_m = peekSpin<SpinColourMatrix>(scm,1,2);
// c_m = peekIdiot<SpinColourMatrix>(scm,1,2);
printf("c. Level %d\n",c_m.TensorLevel);
printf("c. Level %d\n",c_m().TensorLevel);
printf("c. Level %d\n",c_m()().TensorLevel);
printf("c. Level %d\n", c_m.TensorLevel);
printf("c. Level %d\n", c_m().TensorLevel);
printf("c. Level %d\n", c_m()().TensorLevel);
c_m()() = scm()(0,0); //ColourComponents of CM <= ColourComponents of SpinColourMatrix
scm()(1,1) = cm()(); //ColourComponents of CM <= ColourComponents of SpinColourMatrix
c = scm()(1,1)(1,2);
scm()(1,1)(2,1) = c;
c_m()() = scm()(0, 0); // ColourComponents of CM <= ColourComponents of
// SpinColourMatrix
scm()(1, 1) = cm()(); // ColourComponents of CM <= ColourComponents of
// SpinColourMatrix
c = scm()(1, 1)(1, 2);
scm()(1, 1)(2, 1) = c;
// pokeIndex<ColourIndex> (c_m,c,0,0);
}
@ -364,47 +383,49 @@ int main (int argc, char ** argv)
*/
lex_sites(Foo);
Integer mm[4];
mm[0]=1;
mm[1]=Fine._rdimensions[0];
mm[2]=Fine._ldimensions[0]*Fine._ldimensions[1];
mm[3]=Fine._ldimensions[0]*Fine._ldimensions[1]*Fine._ldimensions[2];
mm[0] = 1;
mm[1] = Fine._rdimensions[0];
mm[2] = Fine._ldimensions[0] * Fine._ldimensions[1];
mm[3] =
Fine._ldimensions[0] * Fine._ldimensions[1] * Fine._ldimensions[2];
LatticeInteger lex(&Fine);
lex=zero;
for(int d=0;d<4;d++){
lex = zero;
for (int d = 0; d < 4; d++) {
LatticeInteger coor(&Fine);
LatticeCoordinate(coor,d);
lex = lex + coor*mm[d];
LatticeCoordinate(coor, d);
lex = lex + coor * mm[d];
}
// Bar = zero;
// Bar = where(lex<Integer(10),Foo,Bar);
cout << "peeking sites..\n";
{
std::vector<int> coor(4);
for(coor[3]=0;coor[3]<latt_size[3]/mpi_layout[3];coor[3]++){
for(coor[2]=0;coor[2]<latt_size[2]/mpi_layout[2];coor[2]++){
for(coor[1]=0;coor[1]<latt_size[1]/mpi_layout[1];coor[1]++){
for(coor[0]=0;coor[0]<latt_size[0]/mpi_layout[0];coor[0]++){
for (coor[3] = 0; coor[3] < latt_size[3] / mpi_layout[3]; coor[3]++) {
for (coor[2] = 0; coor[2] < latt_size[2] / mpi_layout[2]; coor[2]++) {
for (coor[1] = 0; coor[1] < latt_size[1] / mpi_layout[1];
coor[1]++) {
for (coor[0] = 0; coor[0] < latt_size[0] / mpi_layout[0];
coor[0]++) {
ColourMatrix bar;
peekSite(bar,Bar,coor);
for(int r=0;r<3;r++){
for(int c=0;c<3;c++){
// cout<<"bar "<<coor[0]<<coor[1]<<coor[2]<<coor[3] <<" "<<bar()()(r,c)<<std::endl;
}}
}}}}
peekSite(bar, Bar, coor);
for (int r = 0; r < 3; r++) {
for (int c = 0; c < 3; c++) {
// cout<<"bar "<<coor[0]<<coor[1]<<coor[2]<<coor[3] <<"
// "<<bar()()(r,c)<<std::endl;
}
}
}
}
}
}
}
//setCheckerboard(ShiftedCheck,rFoo);
//setCheckerboard(ShiftedCheck,bFoo);
// setCheckerboard(ShiftedCheck,rFoo);
// setCheckerboard(ShiftedCheck,bFoo);
// Lattice SU(3) x SU(3)
Fine.Barrier();
@ -414,128 +435,137 @@ int main (int argc, char ** argv)
scFooBar = scFoo * scBar;
// Benchmark some simple operations LatticeSU3 * Lattice SU3.
double t0,t1,flops;
double t0, t1, flops;
double bytes;
int ncall=5000;
int ncall = 5000;
int Nc = Grid::QCD::Nc;
LatticeGaugeField U(&Fine);
// LatticeColourMatrix Uy = peekLorentz(U,1);
// LatticeColourMatrix Uy = peekDumKopf(U,1);
flops = ncall*1.0*volume*(8*Nc*Nc*Nc);
bytes = ncall*1.0*volume*Nc*Nc *2*3*sizeof(Grid::Real);
if ( Fine.IsBoss() ) {
printf("%f flop and %f bytes\n",flops,bytes/ncall);
flops = ncall * 1.0 * volume * (8 * Nc * Nc * Nc);
bytes = ncall * 1.0 * volume * Nc * Nc * 2 * 3 * sizeof(Grid::Real);
if (Fine.IsBoss()) {
printf("%f flop and %f bytes\n", flops, bytes / ncall);
}
FooBar = Foo * Bar;
Fine.Barrier();
t0=usecond();
for(int i=0;i<ncall;i++){
t0 = usecond();
for (int i = 0; i < ncall; i++) {
Fine.Barrier();
mult(FooBar,Foo,Bar); // this is better
mult(FooBar, Foo, Bar); // this is better
}
t1=usecond();
t1 = usecond();
Fine.Barrier();
if ( Fine.IsBoss() ) {
if (Fine.IsBoss()) {
#ifdef OMP
printf("mult NumThread %d , Lattice size %d , %f us per call\n",omp_get_max_threads(),lat,(t1-t0)/ncall);
printf("mult NumThread %d , Lattice size %d , %f us per call\n",
omp_get_max_threads(), lat, (t1 - t0) / ncall);
#endif
printf("mult NumThread %d , Lattice size %d , %f us per call\n",omp,lat,(t1-t0)/ncall);
printf("mult NumThread %d , Lattice size %d , %f Mflop/s\n",omp,lat,flops/(t1-t0));
printf("mult NumThread %d , Lattice size %d , %f MB/s\n",omp,lat,bytes/(t1-t0));
printf("mult NumThread %d , Lattice size %d , %f us per call\n", omp,
lat, (t1 - t0) / ncall);
printf("mult NumThread %d , Lattice size %d , %f Mflop/s\n", omp, lat,
flops / (t1 - t0));
printf("mult NumThread %d , Lattice size %d , %f MB/s\n", omp, lat,
bytes / (t1 - t0));
}
mult(FooBar,Foo,Bar);
mult(FooBar, Foo, Bar);
FooBar = Foo * Bar;
bytes = ncall*1.0*volume*Nc*Nc *2*5*sizeof(Grid::Real);
bytes = ncall * 1.0 * volume * Nc * Nc * 2 * 5 * sizeof(Grid::Real);
Fine.Barrier();
t0=usecond();
for(int i=0;i<ncall;i++){
t0 = usecond();
for (int i = 0; i < ncall; i++) {
Fine.Barrier();
mult(FooBar,Foo,Cshift(Bar,1,-1));
//mult(FooBar,Foo,Bar);
//FooBar = Foo * Bar; // this is bad
mult(FooBar, Foo, Cshift(Bar, 1, -1));
// mult(FooBar,Foo,Bar);
// FooBar = Foo * Bar; // this is bad
}
t1=usecond();
t1 = usecond();
Fine.Barrier();
FooBar = Foo * Bar;
if ( Fine.IsBoss() ) {
printf("Cshift Mult: NumThread %d , Lattice size %d , %f us per call\n",omp,lat,(t1-t0)/ncall);
printf("Cshift Mult: NumThread %d , Lattice size %d , %f Mflop/s\n",omp,lat,flops/(t1-t0));
printf("Cshift Mult: NumThread %d , Lattice size %d , %f MB/s\n",omp,lat,bytes/(t1-t0));
if (Fine.IsBoss()) {
printf("Cshift Mult: NumThread %d , Lattice size %d , %f us per call\n",
omp, lat, (t1 - t0) / ncall);
printf("Cshift Mult: NumThread %d , Lattice size %d , %f Mflop/s\n",
omp, lat, flops / (t1 - t0));
printf("Cshift Mult: NumThread %d , Lattice size %d , %f MB/s\n", omp,
lat, bytes / (t1 - t0));
}
// pickCheckerboard(0,rFoo,FooBar);
// pickCheckerboard(1,bFoo,FooBar);
// setCheckerboard(FooBar,rFoo);
// setCheckerboard(FooBar,bFoo);
double nrm=0;
double nrm = 0;
LatticeColourMatrix deriv(&Fine);
double half=0.5;
deriv = 0.5*Cshift(Foo,0,1) - 0.5*Cshift(Foo,0,-1);
double half = 0.5;
deriv = 0.5 * Cshift(Foo, 0, 1) - 0.5 * Cshift(Foo, 0, -1);
for (int dir = 0; dir < 4; dir++) {
for (int shift = 0; shift < latt_size[dir]; shift++) {
pickCheckerboard(0, rFoo,
Foo); // Pick out red or black checkerboards
pickCheckerboard(1, bFoo, Foo);
for(int dir=0;dir<4;dir++){
for(int shift=0;shift<latt_size[dir];shift++){
pickCheckerboard(0,rFoo,Foo); // Pick out red or black checkerboards
pickCheckerboard(1,bFoo,Foo);
if ( Fine.IsBoss() ) {
std::cout<<GridLogMessage << "Shifting both parities by "<< shift <<" direction "<< dir <<std::endl;
if (Fine.IsBoss()) {
std::cout << GridLogMessage << "Shifting both parities by " << shift
<< " direction " << dir << std::endl;
}
Shifted = Cshift(Foo,dir,shift); // Shift everything
Shifted = Cshift(Foo, dir, shift); // Shift everything
bShifted = Cshift(rFoo,dir,shift); // Shift red->black
rShifted = Cshift(bFoo,dir,shift); // Shift black->red
bShifted = Cshift(rFoo, dir, shift); // Shift red->black
rShifted = Cshift(bFoo, dir, shift); // Shift black->red
ShiftedCheck=zero;
setCheckerboard(ShiftedCheck,bShifted); // Put them all together
setCheckerboard(ShiftedCheck,rShifted); // and check the results (later)
ShiftedCheck = zero;
setCheckerboard(ShiftedCheck, bShifted); // Put them all together
setCheckerboard(ShiftedCheck,
rShifted); // and check the results (later)
// Check results
std::vector<int> coor(4);
for(coor[3]=0;coor[3]<latt_size[3]/mpi_layout[3];coor[3]++){
for(coor[2]=0;coor[2]<latt_size[2]/mpi_layout[2];coor[2]++){
for(coor[1]=0;coor[1]<latt_size[1]/mpi_layout[1];coor[1]++){
for(coor[0]=0;coor[0]<latt_size[0]/mpi_layout[0];coor[0]++){
for (coor[3] = 0; coor[3] < latt_size[3] / mpi_layout[3]; coor[3]++) {
for (coor[2] = 0; coor[2] < latt_size[2] / mpi_layout[2];
coor[2]++) {
for (coor[1] = 0; coor[1] < latt_size[1] / mpi_layout[1];
coor[1]++) {
for (coor[0] = 0; coor[0] < latt_size[0] / mpi_layout[0];
coor[0]++) {
std::complex<Grid::Real> diff;
std::vector<int> shiftcoor = coor;
shiftcoor[dir]=(shiftcoor[dir]+shift+latt_size[dir])%(latt_size[dir]/mpi_layout[dir]);
shiftcoor[dir] = (shiftcoor[dir] + shift + latt_size[dir]) %
(latt_size[dir] / mpi_layout[dir]);
std::vector<int> rl(4);
for(int dd=0;dd<4;dd++){
rl[dd] = latt_size[dd]/simd_layout[dd]/mpi_layout[dd];
for (int dd = 0; dd < 4; dd++) {
rl[dd] = latt_size[dd] / simd_layout[dd] / mpi_layout[dd];
}
int lex = coor[0]%rl[0]
+ (coor[1]%rl[1])*rl[0]
+ (coor[2]%rl[2])*rl[0]*rl[1]
+ (coor[3]%rl[3])*rl[0]*rl[1]*rl[2];
lex +=
+1000*(coor[0]/rl[0])
+1000*(coor[1]/rl[1])*simd_layout[0]
+1000*(coor[2]/rl[2])*simd_layout[0]*simd_layout[1]
+1000*(coor[3]/rl[3])*simd_layout[0]*simd_layout[1]*simd_layout[2];
int lex = coor[0] % rl[0] + (coor[1] % rl[1]) * rl[0] +
(coor[2] % rl[2]) * rl[0] * rl[1] +
(coor[3] % rl[3]) * rl[0] * rl[1] * rl[2];
lex += +1000 * (coor[0] / rl[0]) +
1000 * (coor[1] / rl[1]) * simd_layout[0] +
1000 * (coor[2] / rl[2]) * simd_layout[0] *
simd_layout[1] +
1000 * (coor[3] / rl[3]) * simd_layout[0] *
simd_layout[1] * simd_layout[2];
int lex_coor = shiftcoor[0]%rl[0]
+ (shiftcoor[1]%rl[1])*rl[0]
+ (shiftcoor[2]%rl[2])*rl[0]*rl[1]
+ (shiftcoor[3]%rl[3])*rl[0]*rl[1]*rl[2];
lex_coor +=
+1000*(shiftcoor[0]/rl[0])
+1000*(shiftcoor[1]/rl[1])*simd_layout[0]
+1000*(shiftcoor[2]/rl[2])*simd_layout[0]*simd_layout[1]
+1000*(shiftcoor[3]/rl[3])*simd_layout[0]*simd_layout[1]*simd_layout[2];
int lex_coor = shiftcoor[0] % rl[0] +
(shiftcoor[1] % rl[1]) * rl[0] +
(shiftcoor[2] % rl[2]) * rl[0] * rl[1] +
(shiftcoor[3] % rl[3]) * rl[0] * rl[1] * rl[2];
lex_coor += +1000 * (shiftcoor[0] / rl[0]) +
1000 * (shiftcoor[1] / rl[1]) * simd_layout[0] +
1000 * (shiftcoor[2] / rl[2]) * simd_layout[0] *
simd_layout[1] +
1000 * (shiftcoor[3] / rl[3]) * simd_layout[0] *
simd_layout[1] * simd_layout[2];
ColourMatrix foo;
ColourMatrix bar;
@ -544,69 +574,85 @@ int main (int argc, char ** argv)
ColourMatrix shifted3;
ColourMatrix foobar1;
ColourMatrix foobar2;
ColourMatrix mdiff,amdiff;
ColourMatrix mdiff, amdiff;
peekSite(shifted1,Shifted,coor);
peekSite(shifted2,Foo,shiftcoor);
peekSite(shifted3,ShiftedCheck,coor);
peekSite(foo,Foo,coor);
peekSite(shifted1, Shifted, coor);
peekSite(shifted2, Foo, shiftcoor);
peekSite(shifted3, ShiftedCheck, coor);
peekSite(foo, Foo, coor);
mdiff = shifted1-shifted2;
amdiff=adj(mdiff);
ColourMatrix prod = amdiff*mdiff;
mdiff = shifted1 - shifted2;
amdiff = adj(mdiff);
ColourMatrix prod = amdiff * mdiff;
Complex trprod = trace(prod);
Real Ttr=real(trprod);
double nn=Ttr;
if ( nn > 0 )
cout<<"Shift real trace fail "<<coor[0]<<coor[1]<<coor[2]<<coor[3] <<endl;
Real Ttr = real(trprod);
double nn = Ttr;
if (nn > 0)
cout << "Shift real trace fail " << coor[0] << coor[1]
<< coor[2] << coor[3] << endl;
for(int r=0;r<3;r++){
for(int c=0;c<3;c++){
diff =shifted1()()(r,c)-shifted2()()(r,c);
nn=real(conjugate(diff)*diff);
if ( nn > 0 )
cout<<"Shift fail (shifted1/shifted2-ref) "<<coor[0]<<coor[1]<<coor[2]<<coor[3] <<" "
<<shifted1()()(r,c)<<" "<<shifted2()()(r,c)
<< " "<< foo()()(r,c)<< " lex expect " << lex_coor << " lex "<<lex<<endl;
else if(0)
cout<<"Shift pass 1vs2 "<<coor[0]<<coor[1]<<coor[2]<<coor[3] <<" "
<<shifted1()()(r,c)<<" "<<shifted2()()(r,c)
<< " "<< foo()()(r,c)<< " lex expect " << lex_coor << " lex "<<lex<<endl;
}}
for(int r=0;r<3;r++){
for(int c=0;c<3;c++){
diff =shifted3()()(r,c)-shifted2()()(r,c);
nn=real(conjugate(diff)*diff);
if ( nn > 0 )
cout<<"Shift rb fail (shifted3/shifted2-ref) "<<coor[0]<<coor[1]<<coor[2]<<coor[3] <<" "
<<shifted3()()(r,c)<<" "<<shifted2()()(r,c)
<< " "<< foo()()(r,c)<< " lex expect " << lex_coor << " lex "<<lex<<endl;
else if(0)
cout<<"Shift rb pass 3vs2 "<<coor[0]<<coor[1]<<coor[2]<<coor[3] <<" "
<<shifted3()()(r,c)<<" "<<shifted2()()(r,c)
<< " "<< foo()()(r,c)<< " lex expect " << lex_coor << " lex "<<lex<<endl;
}}
peekSite(bar,Bar,coor);
peekSite(foobar1,FooBar,coor);
foobar2 = foo*bar;
for(int r=0;r<Nc;r++){
for(int c=0;c<Nc;c++){
diff =foobar2()()(r,c)-foobar1()()(r,c);
nrm = nrm + real(conjugate(diff)*diff);
}}
}}}}
if( Fine.IsBoss() ){
std::cout<<GridLogMessage << "LatticeColorMatrix * LatticeColorMatrix nrm diff = "<<nrm<<std::endl;
for (int r = 0; r < 3; r++) {
for (int c = 0; c < 3; c++) {
diff = shifted1()()(r, c) - shifted2()()(r, c);
nn = real(conjugate(diff) * diff);
if (nn > 0)
cout << "Shift fail (shifted1/shifted2-ref) " << coor[0]
<< coor[1] << coor[2] << coor[3] << " "
<< shifted1()()(r, c) << " " << shifted2()()(r, c)
<< " " << foo()()(r, c) << " lex expect "
<< lex_coor << " lex " << lex << endl;
else if (0)
cout << "Shift pass 1vs2 " << coor[0] << coor[1]
<< coor[2] << coor[3] << " " << shifted1()()(r, c)
<< " " << shifted2()()(r, c) << " "
<< foo()()(r, c) << " lex expect " << lex_coor
<< " lex " << lex << endl;
}
}
for (int r = 0; r < 3; r++) {
for (int c = 0; c < 3; c++) {
diff = shifted3()()(r, c) - shifted2()()(r, c);
nn = real(conjugate(diff) * diff);
if (nn > 0)
cout << "Shift rb fail (shifted3/shifted2-ref) "
<< coor[0] << coor[1] << coor[2] << coor[3] << " "
<< shifted3()()(r, c) << " " << shifted2()()(r, c)
<< " " << foo()()(r, c) << " lex expect "
<< lex_coor << " lex " << lex << endl;
else if (0)
cout << "Shift rb pass 3vs2 " << coor[0] << coor[1]
<< coor[2] << coor[3] << " " << shifted3()()(r, c)
<< " " << shifted2()()(r, c) << " "
<< foo()()(r, c) << " lex expect " << lex_coor
<< " lex " << lex << endl;
}
}
peekSite(bar, Bar, coor);
peekSite(foobar1, FooBar, coor);
foobar2 = foo * bar;
for (int r = 0; r < Nc; r++) {
for (int c = 0; c < Nc; c++) {
diff = foobar2()()(r, c) - foobar1()()(r, c);
nrm = nrm + real(conjugate(diff) * diff);
}
}
}
}
}
}
if (Fine.IsBoss()) {
std::cout << GridLogMessage
<< "LatticeColorMatrix * LatticeColorMatrix nrm diff = "
<< nrm << std::endl;
}
}
}
}}
} // loop for lat
} // loop for omp
/*
// Testing Smearing routine compilation, separate in a different file
GridCartesian Fine(latt_size,simd_layout,mpi_layout);