portelli/Grid
1
0
Fork 0
mirror of https://github.com/paboyle/Grid.git synced 2024-11-14 01:35:36 +00:00
Code Releases Activity
67ccb043f1
Grid/lib/qcd/action/fermion/CayleyFermion5D.h

78 lines
2.8 KiB
C
Raw Normal View History

Domain wall fermions now invert ; have the basis set up for Tanh/Zolo * (Cayley/PartFrac/ContFrac) * (Mobius/Shamir/Wilson) Approx Representation Kernel. All are done with space-time taking part in checkerboarding, Ls uncheckerboarded Have only so far tested the Domain Wall limit of mobius, and at that only checked that it i) Inverts ii) 5dim DW == Ls copies of 4dim D2 iii) MeeInv Mee == 1 iv) Meo+Mee+Moe+Moo == M unprec. v) MpcDagMpc is hermitan vi) Mdag is the adjoint of M between stochastic vectors. That said, the RB schur solve, RB MpcDagMpc solve, Unprec solve all converge and the true residual becomes small; so pretty good tests.
2015-06-02 16:57:12 +01:00
#ifndef GRID_QCD_CAYLEY_FERMION_H
#define GRID_QCD_CAYLEY_FERMION_H
namespace Grid {
namespace QCD {
Rework/global edit to enforce type templating of fermion operators. Allows multi-precision work and paves the way for alternate BC's and such like allowing for example G-parity which is important for K pipi programme. In particular, can drive an extra flavour index into the fermion fields using template types.
2015-08-10 20:47:44 +01:00
template<class Impl>
class CayleyFermion5D : public WilsonFermion5D<Impl>
Domain wall fermions now invert ; have the basis set up for Tanh/Zolo * (Cayley/PartFrac/ContFrac) * (Mobius/Shamir/Wilson) Approx Representation Kernel. All are done with space-time taking part in checkerboarding, Ls uncheckerboarded Have only so far tested the Domain Wall limit of mobius, and at that only checked that it i) Inverts ii) 5dim DW == Ls copies of 4dim D2 iii) MeeInv Mee == 1 iv) Meo+Mee+Moe+Moo == M unprec. v) MpcDagMpc is hermitan vi) Mdag is the adjoint of M between stochastic vectors. That said, the RB schur solve, RB MpcDagMpc solve, Unprec solve all converge and the true residual becomes small; so pretty good tests.
2015-06-02 16:57:12 +01:00
{
public:
* Finished the template/policy style introduction of gparity, except the gparity force terms. So valence sector looks ok. FermionOperatorImpl.h provides the policy classes. Expect HMC will introduce a smearing policy and a fermion representation change policy template param. Will also probably need multi-precision work. * HMC is running even-odd and non-checkerboarded (checked 4^4 wilson fermion/wilson gauge). There appears to be a bug in the multi-level integrator -- <e-dH> passes with single level but not with multi-level. In any case there looks to be quite a bit to clean up. This is the "const det" style implementation that is not appropriate yet for clover since it assumes that Mee is indept of the gauge fields. Easily fixed in future.
2015-08-15 23:25:49 +01:00
INHERIT_IMPL_TYPES(Impl);
Rework/global edit to enforce type templating of fermion operators. Allows multi-precision work and paves the way for alternate BC's and such like allowing for example G-parity which is important for K pipi programme. In particular, can drive an extra flavour index into the fermion fields using template types.
2015-08-10 20:47:44 +01:00
public:
Domain wall fermions now invert ; have the basis set up for Tanh/Zolo * (Cayley/PartFrac/ContFrac) * (Mobius/Shamir/Wilson) Approx Representation Kernel. All are done with space-time taking part in checkerboarding, Ls uncheckerboarded Have only so far tested the Domain Wall limit of mobius, and at that only checked that it i) Inverts ii) 5dim DW == Ls copies of 4dim D2 iii) MeeInv Mee == 1 iv) Meo+Mee+Moe+Moo == M unprec. v) MpcDagMpc is hermitan vi) Mdag is the adjoint of M between stochastic vectors. That said, the RB schur solve, RB MpcDagMpc solve, Unprec solve all converge and the true residual becomes small; so pretty good tests.
2015-06-02 16:57:12 +01:00
// override multiply
Rework/global edit to enforce type templating of fermion operators. Allows multi-precision work and paves the way for alternate BC's and such like allowing for example G-parity which is important for K pipi programme. In particular, can drive an extra flavour index into the fermion fields using template types.
2015-08-10 20:47:44 +01:00
virtual RealD M (const FermionField &in, FermionField &out);
virtual RealD Mdag (const FermionField &in, FermionField &out);
Domain wall fermions now invert ; have the basis set up for Tanh/Zolo * (Cayley/PartFrac/ContFrac) * (Mobius/Shamir/Wilson) Approx Representation Kernel. All are done with space-time taking part in checkerboarding, Ls uncheckerboarded Have only so far tested the Domain Wall limit of mobius, and at that only checked that it i) Inverts ii) 5dim DW == Ls copies of 4dim D2 iii) MeeInv Mee == 1 iv) Meo+Mee+Moe+Moo == M unprec. v) MpcDagMpc is hermitan vi) Mdag is the adjoint of M between stochastic vectors. That said, the RB schur solve, RB MpcDagMpc solve, Unprec solve all converge and the true residual becomes small; so pretty good tests.
2015-06-02 16:57:12 +01:00
// half checkerboard operations
Rework/global edit to enforce type templating of fermion operators. Allows multi-precision work and paves the way for alternate BC's and such like allowing for example G-parity which is important for K pipi programme. In particular, can drive an extra flavour index into the fermion fields using template types.
2015-08-10 20:47:44 +01:00
virtual void Meooe (const FermionField &in, FermionField &out);
virtual void MeooeDag (const FermionField &in, FermionField &out);
virtual void Mooee (const FermionField &in, FermionField &out);
virtual void MooeeDag (const FermionField &in, FermionField &out);
virtual void MooeeInv (const FermionField &in, FermionField &out);
virtual void MooeeInvDag (const FermionField &in, FermionField &out);
First pass at continued fraction; solver and even odd decomposition tests pass. Have to make ContFrac class virtual and derive end non-abstract actions for the particular cases.
2015-06-04 00:00:45 +01:00
virtual void Instantiatable(void)=0;
Conjugate residual algorithm; some more unary functions
2015-06-08 12:04:59 +01:00
Jackson smoothed chebyshev and (untested) completion of force terms for Cayley, Partial and Cont fraction dwf and overlap. have even odd and unprec forces.
2015-07-31 21:58:35 +01:00
// force terms; five routines; default to Dhop on diagonal
Rework/global edit to enforce type templating of fermion operators. Allows multi-precision work and paves the way for alternate BC's and such like allowing for example G-parity which is important for K pipi programme. In particular, can drive an extra flavour index into the fermion fields using template types.
2015-08-10 20:47:44 +01:00
virtual void MDeriv (GaugeField &mat,const FermionField &U,const FermionField &V,int dag);
virtual void MoeDeriv(GaugeField &mat,const FermionField &U,const FermionField &V,int dag);
virtual void MeoDeriv(GaugeField &mat,const FermionField &U,const FermionField &V,int dag);
Jackson smoothed chebyshev and (untested) completion of force terms for Cayley, Partial and Cont fraction dwf and overlap. have even odd and unprec forces.
2015-07-31 21:58:35 +01:00
Conjugate residual algorithm; some more unary functions
2015-06-08 12:04:59 +01:00
// Efficient support for multigrid coarsening
Rework/global edit to enforce type templating of fermion operators. Allows multi-precision work and paves the way for alternate BC's and such like allowing for example G-parity which is important for K pipi programme. In particular, can drive an extra flavour index into the fermion fields using template types.
2015-08-10 20:47:44 +01:00
virtual void Mdir (const FermionField &in, FermionField &out,int dir,int disp);
Jackson smoothed chebyshev and (untested) completion of force terms for Cayley, Partial and Cont fraction dwf and overlap. have even odd and unprec forces.
2015-07-31 21:58:35 +01:00
Rework/global edit to enforce type templating of fermion operators. Allows multi-precision work and paves the way for alternate BC's and such like allowing for example G-parity which is important for K pipi programme. In particular, can drive an extra flavour index into the fermion fields using template types.
2015-08-10 20:47:44 +01:00
void Meooe5D (const FermionField &in, FermionField &out);
void MeooeDag5D (const FermionField &in, FermionField &out);
Conjugate residual algorithm; some more unary functions
2015-06-08 12:04:59 +01:00
Domain wall fermions now invert ; have the basis set up for Tanh/Zolo * (Cayley/PartFrac/ContFrac) * (Mobius/Shamir/Wilson) Approx Representation Kernel. All are done with space-time taking part in checkerboarding, Ls uncheckerboarded Have only so far tested the Domain Wall limit of mobius, and at that only checked that it i) Inverts ii) 5dim DW == Ls copies of 4dim D2 iii) MeeInv Mee == 1 iv) Meo+Mee+Moe+Moo == M unprec. v) MpcDagMpc is hermitan vi) Mdag is the adjoint of M between stochastic vectors. That said, the RB schur solve, RB MpcDagMpc solve, Unprec solve all converge and the true residual becomes small; so pretty good tests.
2015-06-02 16:57:12 +01:00
// protected:
RealD mass;
Mobius Caley form, Mobius Zolotarev operators. Pass Even Odd vs unprec test and hermiticity checks in tests/Grid_any_evenodd.cc; will work on inversion tests shortly.
2015-06-03 09:36:26 +01:00
Domain wall fermions now invert ; have the basis set up for Tanh/Zolo * (Cayley/PartFrac/ContFrac) * (Mobius/Shamir/Wilson) Approx Representation Kernel. All are done with space-time taking part in checkerboarding, Ls uncheckerboarded Have only so far tested the Domain Wall limit of mobius, and at that only checked that it i) Inverts ii) 5dim DW == Ls copies of 4dim D2 iii) MeeInv Mee == 1 iv) Meo+Mee+Moe+Moo == M unprec. v) MpcDagMpc is hermitan vi) Mdag is the adjoint of M between stochastic vectors. That said, the RB schur solve, RB MpcDagMpc solve, Unprec solve all converge and the true residual becomes small; so pretty good tests.
2015-06-02 16:57:12 +01:00
// Cayley form Moebius (tanh and zolotarev)
std::vector<RealD> omega;
std::vector<RealD> bs; // S dependent coeffs
std::vector<RealD> cs;
std::vector<RealD> as;
// For preconditioning Cayley form
std::vector<RealD> bee;
std::vector<RealD> cee;
std::vector<RealD> aee;
std::vector<RealD> beo;
std::vector<RealD> ceo;
std::vector<RealD> aeo;
// LDU factorisation of the eeoo matrix
std::vector<RealD> lee;
std::vector<RealD> leem;
std::vector<RealD> uee;
std::vector<RealD> ueem;
std::vector<RealD> dee;
// Constructors
Rework/global edit to enforce type templating of fermion operators. Allows multi-precision work and paves the way for alternate BC's and such like allowing for example G-parity which is important for K pipi programme. In particular, can drive an extra flavour index into the fermion fields using template types.
2015-08-10 20:47:44 +01:00
CayleyFermion5D(GaugeField &_Umu,
Domain wall fermions now invert ; have the basis set up for Tanh/Zolo * (Cayley/PartFrac/ContFrac) * (Mobius/Shamir/Wilson) Approx Representation Kernel. All are done with space-time taking part in checkerboarding, Ls uncheckerboarded Have only so far tested the Domain Wall limit of mobius, and at that only checked that it i) Inverts ii) 5dim DW == Ls copies of 4dim D2 iii) MeeInv Mee == 1 iv) Meo+Mee+Moe+Moo == M unprec. v) MpcDagMpc is hermitan vi) Mdag is the adjoint of M between stochastic vectors. That said, the RB schur solve, RB MpcDagMpc solve, Unprec solve all converge and the true residual becomes small; so pretty good tests.
2015-06-02 16:57:12 +01:00
GridCartesian &FiveDimGrid,
GridRedBlackCartesian &FiveDimRedBlackGrid,
GridCartesian &FourDimGrid,
GridRedBlackCartesian &FourDimRedBlackGrid,
* Finished the template/policy style introduction of gparity, except the gparity force terms. So valence sector looks ok. FermionOperatorImpl.h provides the policy classes. Expect HMC will introduce a smearing policy and a fermion representation change policy template param. Will also probably need multi-precision work. * HMC is running even-odd and non-checkerboarded (checked 4^4 wilson fermion/wilson gauge). There appears to be a bug in the multi-level integrator -- <e-dH> passes with single level but not with multi-level. In any case there looks to be quite a bit to clean up. This is the "const det" style implementation that is not appropriate yet for clover since it assumes that Mee is indept of the gauge fields. Easily fixed in future.
2015-08-15 23:25:49 +01:00
RealD _mass,RealD _M5,const ImplParams &p= ImplParams());
Domain wall fermions now invert ; have the basis set up for Tanh/Zolo * (Cayley/PartFrac/ContFrac) * (Mobius/Shamir/Wilson) Approx Representation Kernel. All are done with space-time taking part in checkerboarding, Ls uncheckerboarded Have only so far tested the Domain Wall limit of mobius, and at that only checked that it i) Inverts ii) 5dim DW == Ls copies of 4dim D2 iii) MeeInv Mee == 1 iv) Meo+Mee+Moe+Moo == M unprec. v) MpcDagMpc is hermitan vi) Mdag is the adjoint of M between stochastic vectors. That said, the RB schur solve, RB MpcDagMpc solve, Unprec solve all converge and the true residual becomes small; so pretty good tests.
2015-06-02 16:57:12 +01:00
Mobius Caley form, Mobius Zolotarev operators. Pass Even Odd vs unprec test and hermiticity checks in tests/Grid_any_evenodd.cc; will work on inversion tests shortly.
2015-06-03 09:36:26 +01:00
protected:
First pass at continued fraction; solver and even odd decomposition tests pass. Have to make ContFrac class virtual and derive end non-abstract actions for the particular cases.
2015-06-04 00:00:45 +01:00
void SetCoefficientsZolotarev(RealD zolohi,Approx::zolotarev_data *zdata,RealD b,RealD c);
void SetCoefficientsTanh(Approx::zolotarev_data *zdata,RealD b,RealD c);
Domain wall fermions now invert ; have the basis set up for Tanh/Zolo * (Cayley/PartFrac/ContFrac) * (Mobius/Shamir/Wilson) Approx Representation Kernel. All are done with space-time taking part in checkerboarding, Ls uncheckerboarded Have only so far tested the Domain Wall limit of mobius, and at that only checked that it i) Inverts ii) 5dim DW == Ls copies of 4dim D2 iii) MeeInv Mee == 1 iv) Meo+Mee+Moe+Moo == M unprec. v) MpcDagMpc is hermitan vi) Mdag is the adjoint of M between stochastic vectors. That said, the RB schur solve, RB MpcDagMpc solve, Unprec solve all converge and the true residual becomes small; so pretty good tests.
2015-06-02 16:57:12 +01:00
};
}
}
#endif
Copy Permalink