Communication-avoiding GCR solver. This solver does un-preconditioned GCR, first building up a polynomial in the linear operator of length n_krylov, and then performs a minimum residual extrapolation on the resulting basis vectors. For use as a multigrid smoother with minimum global synchronization. More...

#include <invert_quda.h>

Inheritance diagram for quda::CAGCR:

Public Member Functions
	CAGCR (const DiracMatrix &mat, const DiracMatrix &matSloppy, const DiracMatrix &matPrecon, const DiracMatrix &matEig, SolverParam &param, TimeProfile &profile)

virtual	~CAGCR ()

void	operator() (ColorSpinorField &out, ColorSpinorField &in)

virtual bool	hermitian ()

Public Member Functions inherited from quda::Solver
	Solver (const DiracMatrix &mat, const DiracMatrix &matSloppy, const DiracMatrix &matPrecon, const DiracMatrix &matEig, SolverParam &param, TimeProfile &profile)

virtual	~Solver ()

virtual void	blocksolve (ColorSpinorField &out, ColorSpinorField &in)

const DiracMatrix &	M ()

const DiracMatrix &	Msloppy ()

const DiracMatrix &	Mprecon ()

const DiracMatrix &	Meig ()

bool	convergence (double r2, double hq2, double r2_tol, double hq_tol)

bool	convergenceHQ (double r2, double hq2, double r2_tol, double hq_tol)
	Test for HQ solver convergence – ignore L2 residual. More...

bool	convergenceL2 (double r2, double hq2, double r2_tol, double hq_tol)
	Test for L2 solver convergence – ignore HQ residual. More...

void	PrintStats (const char *name, int k, double r2, double b2, double hq2)
	Prints out the running statistics of the solver (requires a verbosity of QUDA_VERBOSE) More...

void	PrintSummary (const char *name, int k, double r2, double b2, double r2_tol, double hq_tol)
	Prints out the summary of the solver convergence (requires a verbosity of QUDA_SUMMARIZE). Assumes SolverParam.true_res and SolverParam.true_res_hq has been set. More...

double	precisionEpsilon (QudaPrecision prec=QUDA_INVALID_PRECISION) const
	Returns the epsilon tolerance for a given precision, by default returns the solver precision. More...

void	constructDeflationSpace (const ColorSpinorField &meta, const DiracMatrix &mat)
	Constructs the deflation space and eigensolver. More...

void	destroyDeflationSpace ()
	Destroy the allocated deflation space. More...

void	extendSVDDeflationSpace ()
	Extends the deflation space to twice its size for SVD deflation. More...

void	injectDeflationSpace (std::vector< ColorSpinorField * > &defl_space)
	Injects a deflation space into the solver from the vector argument. Note the input space is reduced to zero size as a result of calling this function, with responsibility for the space transferred to the solver. More...

void	extractDeflationSpace (std::vector< ColorSpinorField * > &defl_space)
	Extracts the deflation space from the solver to the vector argument. Note the solver deflation space is reduced to zero size as a result of calling this function, with responsibility for the space transferred to the argument. More...

int	deflationSpaceSize () const
	Returns the size of deflation space. More...

void	setDeflateCompute (bool flag)
	Sets the deflation compute boolean. More...

void	setRecomputeEvals (bool flag)
	Sets the recompute evals boolean. More...

virtual double	flops () const
	Return flops. More...

Additional Inherited Members
Static Public Member Functions inherited from quda::Solver
static Solver *	create (SolverParam &param, const DiracMatrix &mat, const DiracMatrix &matSloppy, const DiracMatrix &matPrecon, const DiracMatrix &matEig, TimeProfile &profile)
	Solver factory. More...

static double	stopping (double tol, double b2, QudaResidualType residual_type)
	Set the solver L2 stopping condition. More...

Protected Attributes inherited from quda::Solver
const DiracMatrix &	mat

const DiracMatrix &	matSloppy

const DiracMatrix &	matPrecon

const DiracMatrix &	matEig

SolverParam &	param

TimeProfile &	profile

int	node_parity

EigenSolver *	eig_solve

bool	deflate_init

bool	deflate_compute

bool	recompute_evals

std::vector< ColorSpinorField * >	evecs

std::vector< Complex >	evals

Detailed Description

Communication-avoiding GCR solver. This solver does un-preconditioned GCR, first building up a polynomial in the linear operator of length n_krylov, and then performs a minimum residual extrapolation on the resulting basis vectors. For use as a multigrid smoother with minimum global synchronization.

Definition at line 1099 of file invert_quda.h.

Constructor & Destructor Documentation

◆ CAGCR()

quda::CAGCR::CAGCR	(	const DiracMatrix &	mat,
		const DiracMatrix &	matSloppy,
		const DiracMatrix &	matPrecon,
		const DiracMatrix &	matEig,
		SolverParam &	param,
		TimeProfile &	profile
	)

Definition at line 7 of file inv_ca_gcr.cpp.

◆ ~CAGCR()

quda::CAGCR::~CAGCR ( )

virtual

Definition at line 22 of file inv_ca_gcr.cpp.

Member Function Documentation

◆ hermitian()

virtual bool quda::CAGCR::hermitian ( )

inlinevirtual

Returns: Whether the solver is only for Hermitian systems

Implements quda::Solver.

Definition at line 1143 of file invert_quda.h.

◆ operator()()

void quda::CAGCR::operator()	(	ColorSpinorField &	out,
		ColorSpinorField &	in
	)

virtual

Implements quda::Solver.

Definition at line 168 of file inv_ca_gcr.cpp.

The documentation for this class was generated from the following files:

quda/include/invert_quda.h
quda/lib/inv_ca_gcr.cpp

Public Member Functions