#include <multigrid.h>

Inheritance diagram for quda::MGParam:

Public Member Functions
	MGParam (QudaMultigridParam &param, std::vector< ColorSpinorField * > &B, DiracMatrix matResidual, DiracMatrix matSmooth, DiracMatrix *matSmoothSloppy, int level=0)

	MGParam (const MGParam &param, std::vector< ColorSpinorField * > &B, DiracMatrix matResidual, DiracMatrix matSmooth, DiracMatrix *matSmoothSloppy, int level=0)

Public Member Functions inherited from quda::SolverParam
	SolverParam ()

	SolverParam (const QudaInvertParam &param)

	SolverParam (const SolverParam &param)

	~SolverParam ()

void	updateInvertParam (QudaInvertParam &param, int offset=-1)

void	updateRhsIndex (QudaInvertParam &param)

Public Attributes
QudaMultigridParam &	mg_global

int	level

int	Nlevel

int	geoBlockSize [QUDA_MAX_DIM]

int	spinBlockSize

int	Nvec

int	NblockOrtho

MG *	coarse

MG *	fine

std::vector< ColorSpinorField * > &	B

int	nu_pre

int	nu_post

double	smoother_tol

QudaMultigridCycleType	cycle_type

QudaBoolean	global_reduction

DiracMatrix *	matResidual

DiracMatrix *	matSmooth

DiracMatrix *	matSmoothSloppy

QudaInverterType	smoother

QudaSolutionType	coarse_grid_solution_type

QudaSolveType	smoother_solve_type

QudaFieldLocation	location

QudaFieldLocation	setup_location

char	filename [100]

QudaTransferType	transfer_type

bool	use_mma

Public Attributes inherited from quda::SolverParam
QudaInverterType	inv_type

QudaInverterType	inv_type_precondition

void *	preconditioner

void *	deflation_op

QudaResidualType	residual_type

bool	deflate

QudaEigParam	eig_param

QudaUseInitGuess	use_init_guess

QudaComputeNullVector	compute_null_vector

double	delta

bool	use_alternative_reliable

bool	use_sloppy_partial_accumulator

int	solution_accumulator_pipeline

int	max_res_increase

int	max_res_increase_total

int	max_hq_res_increase

int	max_hq_res_restart_total

int	heavy_quark_check

int	pipeline

double	tol

double	tol_restart

double	tol_hq

bool	compute_true_res

bool	sloppy_converge

double	true_res

double	true_res_hq

int	maxiter

int	iter

QudaPrecision	precision

QudaPrecision	precision_sloppy

QudaPrecision	precision_refinement_sloppy

QudaPrecision	precision_precondition

QudaPrecision	precision_eigensolver

QudaPreserveSource	preserve_source

bool	return_residual

int	overlap_precondition

int	num_src

int	num_offset

double	offset [QUDA_MAX_MULTI_SHIFT]

double	tol_offset [QUDA_MAX_MULTI_SHIFT]

double	tol_hq_offset [QUDA_MAX_MULTI_SHIFT]

double	true_res_offset [QUDA_MAX_MULTI_SHIFT]

double	iter_res_offset [QUDA_MAX_MULTI_SHIFT]

double	true_res_hq_offset [QUDA_MAX_MULTI_SHIFT]

int	Nsteps

int	Nkrylov

int	precondition_cycle

double	tol_precondition

int	maxiter_precondition

double	omega

QudaCABasis	ca_basis

double	ca_lambda_min

double	ca_lambda_max

QudaSchwarzType	schwarz_type

double	secs

double	gflops

QudaPrecision	precision_ritz

int	n_ev

int	m

int	deflation_grid

int	rhs_idx

int	eigcg_max_restarts

int	max_restart_num

double	inc_tol

double	eigenval_tol

QudaVerbosity	verbosity_precondition

bool	is_preconditioner
	verbosity to use for preconditioner More...

bool	global_reduction
	whether the solver acting as a preconditioner for another solver More...

bool	mg_instance
	whether to use a global or local (node) reduction for this solver More...

QudaExtLibType	extlib_type

Detailed Description

This struct contains all the metadata required to define the multigrid solver. For each level of multigrid we will have an instance of MGParam describing all the meta data appropriate for given level.

Definition at line 25 of file multigrid.h.

Constructor & Destructor Documentation

◆ MGParam() [1/2]

quda::MGParam::MGParam	(	QudaMultigridParam &	param,
		std::vector< ColorSpinorField * > &	B,
		DiracMatrix *	matResidual,
		DiracMatrix *	matSmooth,
		DiracMatrix *	matSmoothSloppy,
		int	level = `0`
	)

inline

This is top level instantiation done when we start creating the multigrid operator.

Definition at line 110 of file multigrid.h.

◆ MGParam() [2/2]

quda::MGParam::MGParam	(	const MGParam &	param,
		std::vector< ColorSpinorField * > &	B,
		DiracMatrix *	matResidual,
		DiracMatrix *	matSmooth,
		DiracMatrix *	matSmoothSloppy,
		int	level = `0`
	)

inline

Definition at line 143 of file multigrid.h.

Member Data Documentation

◆ B

std::vector<ColorSpinorField*>& quda::MGParam::B

The null space vectors

Definition at line 56 of file multigrid.h.

◆ coarse

MG* quda::MGParam::coarse

This is the next lower level

Definition at line 50 of file multigrid.h.

◆ coarse_grid_solution_type

QudaSolutionType quda::MGParam::coarse_grid_solution_type

The type of residual to send to the next coarse grid, and thus the type of solution to receive back from this coarse grid

Definition at line 87 of file multigrid.h.

◆ cycle_type

QudaMultigridCycleType quda::MGParam::cycle_type

Multigrid cycle type

Definition at line 68 of file multigrid.h.

◆ filename

char quda::MGParam::filename[100]

Filename for where to load/store the null space

Definition at line 99 of file multigrid.h.

◆ fine

MG* quda::MGParam::fine

This is the immediate finer level

Definition at line 53 of file multigrid.h.

◆ geoBlockSize

int quda::MGParam::geoBlockSize[QUDA_MAX_DIM]

Geometric block size

Definition at line 38 of file multigrid.h.

◆ global_reduction

QudaBoolean quda::MGParam::global_reduction

Whether to use global or local (node) reductions

Definition at line 71 of file multigrid.h.

◆ level

int quda::MGParam::level

What is the level of this instance

Definition at line 32 of file multigrid.h.

◆ location

QudaFieldLocation quda::MGParam::location

Where to compute this level of multigrid

Definition at line 93 of file multigrid.h.

◆ matResidual

DiracMatrix* quda::MGParam::matResidual

The Dirac operator to use for residual computation

Definition at line 74 of file multigrid.h.

◆ matSmooth

DiracMatrix* quda::MGParam::matSmooth

The Dirac operator to use for smoothing

Definition at line 77 of file multigrid.h.

◆ matSmoothSloppy

DiracMatrix* quda::MGParam::matSmoothSloppy

The sloppy Dirac operator to use for smoothing

Definition at line 80 of file multigrid.h.

◆ mg_global

QudaMultigridParam& quda::MGParam::mg_global

This points to the parameter struct that is passed into QUDA. We use this to set per-level parameters

Definition at line 29 of file multigrid.h.

◆ NblockOrtho

int quda::MGParam::NblockOrtho

Number of times to apply Gram-Schmidt within a block

Definition at line 47 of file multigrid.h.

◆ Nlevel

int quda::MGParam::Nlevel

Number of levels in the solver

Definition at line 35 of file multigrid.h.

◆ nu_post

int quda::MGParam::nu_post

Number of pre-smoothing applications to perform

Definition at line 62 of file multigrid.h.

◆ nu_pre

int quda::MGParam::nu_pre

Number of pre-smoothing applications to perform

Definition at line 59 of file multigrid.h.

◆ Nvec

int quda::MGParam::Nvec

Number of vectors used to define coarse space

Definition at line 44 of file multigrid.h.

◆ setup_location

QudaFieldLocation quda::MGParam::setup_location

Where to compute this level of the multigrid setup

Definition at line 96 of file multigrid.h.

◆ smoother

QudaInverterType quda::MGParam::smoother

What type of smoother to use

Definition at line 83 of file multigrid.h.

◆ smoother_solve_type

QudaSolveType quda::MGParam::smoother_solve_type

The type of smoother solve to do on each grid (e/o preconditioning or not)

Definition at line 90 of file multigrid.h.

◆ smoother_tol

double quda::MGParam::smoother_tol

Tolerance to use for the solver / smoother (if applicable)

Definition at line 65 of file multigrid.h.

◆ spinBlockSize

int quda::MGParam::spinBlockSize

Spin block size

Definition at line 41 of file multigrid.h.

◆ transfer_type

QudaTransferType quda::MGParam::transfer_type

Whether or not this is a staggered solve or not

Definition at line 102 of file multigrid.h.

◆ use_mma

bool quda::MGParam::use_mma

Whether to use tensor cores (if available)

Definition at line 105 of file multigrid.h.

The documentation for this struct was generated from the following file:

quda/include/multigrid.h

Public Member Functions

Public Attributes