cs_sles_default.cpp File Reference

#include "cs_defs.h"
#include <stdarg.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <assert.h>
#include <math.h>
#include "bft_mem.h"
#include "bft_error.h"
#include "bft_printf.h"
#include "cs_base.h"
#include "cs_field.h"
#include "cs_field_default.h"
#include "cs_grid.h"
#include "cs_halo.h"
#include "cs_internal_coupling.h"
#include "cs_log.h"
#include "cs_mesh.h"
#include "cs_mesh_adjacencies.h"
#include "cs_mesh_quantities.h"
#include "cs_matrix.h"
#include "cs_matrix_default.h"
#include "cs_matrix_util.h"
#include "cs_multigrid.h"
#include "cs_parameters.h"
#include "cs_sles.h"
#include "cs_sles_it.h"
#include "cs_sles_pc.h"
#include "cs_timer.h"
#include "cs_sles_default.h"

Include dependency graph for cs_sles_default.cpp:

Functions
void	cs_sles_default (int f_id, const char *name, const cs_matrix_t *a)
	Default definition of a sparse linear equation solver. More...
void	cs_sles_default_setup (void)
	Default setup for sparse linear equation solver API. More...
void	cs_sles_default_finalize (void)
	Default finalization for sparse linear equation solver API. More...
int	cs_sles_default_get_verbosity (int f_id, const char *name)
	Return default verbosity associated to a field id, name couple. More...
cs_matrix_t *	cs_sles_default_get_matrix (int f_id, const char *name, cs_lnum_t db_size, cs_lnum_t eb_size, bool symmetric)
	Return pointer to matrix structure matching equation solve. More...
void	cs_sles_default_release_matrix (cs_matrix_t **m)
void	cs_sles_setup_native_conv_diff (int f_id, const char *name, const cs_lnum_t diag_block_size, const cs_lnum_t extra_diag_block_size, const cs_real_t *da, const cs_real_t *xa, bool conv_diff)
	Call sparse linear equation solver setup for convection-diffusion systems. More...
cs_sles_convergence_state_t	cs_sles_solve_ccc_fv (cs_sles_t *sc, cs_matrix_t *a, double precision, double r_norm, int *n_iter, double *residual, const cs_real_t *rhs, cs_real_t *vx)
	Call sparse linear equation solver for general colocated cell-centered finite volume scheme. More...
cs_sles_convergence_state_t	cs_sles_solve_native (int f_id, const char *name, bool symmetric, cs_lnum_t diag_block_size, cs_lnum_t extra_diag_block_size, const cs_real_t *da, const cs_real_t *xa, double precision, double r_norm, int *n_iter, double *residual, const cs_real_t *rhs, cs_real_t *vx)
	Call sparse linear equation solver using native matrix arrays. More...
void	cs_sles_free_native (int f_id, const char *name)
	Free sparse linear equation solver setup using native matrix arrays. More...
bool	cs_sles_default_error (cs_sles_t *sles, cs_sles_convergence_state_t state, const cs_matrix_t *a, const cs_real_t rhs[], cs_real_t vx[])
	Error handler attempting fallback to alternative solution procedure for sparse linear equation solver. More...

Function Documentation

cs_sles_default()

void cs_sles_default	(	int	f_id,
		const char *	name,
		const cs_matrix_t *	a
	)

Default definition of a sparse linear equation solver.

Parameters

[in]	f_id	associated field id, or < 0
[in]	name	associated name if f_id < 0, or nullptr
[in]	a	matrix

cs_sles_default_error()

bool cs_sles_default_error	(	cs_sles_t *	sles,
		cs_sles_convergence_state_t	state,
		const cs_matrix_t *	a,
		const cs_real_t	rhs[],
		cs_real_t	vx[]
	)

Error handler attempting fallback to alternative solution procedure for sparse linear equation solver.

In case of divergence with an iterative solver, this error handler switches to a default preconditioner, then resets the solution vector.

The default error for the solver type handler is then set, in case the solution fails again.

Note that this error handler may rebuild solver contexts, so should not be used in conjunction with shared contexts (such as multigrid ascent/descent contexts), but only for "outer" solvers.

Parameters

[in,out]	sles	pointer to solver object
[in]	state	convergence status
[in]	a	matrix
[in]	rhs	right hand side
[in,out]	vx	system solution

Returns

true if fallback solution is possible, false otherwise

cs_sles_default_finalize()

void cs_sles_default_finalize

(

void

)

Default finalization for sparse linear equation solver API.

This includes performance data logging output.

cs_sles_default_get_matrix()

cs_matrix_t * cs_sles_default_get_matrix	(	int	f_id,
		const char *	name,
		cs_lnum_t	db_size,
		cs_lnum_t	eb_size,
		bool	symmetric
	)

Return pointer to matrix structure matching equation solve.

Some matrix properties (such as assembly options and geometric association) are set immediately, but coefficients are not assigned at this stage.

Parameters

[in]	f_id	associated field id, or < 0
[in]	name	associated name if f_id < 0, or nullptr
[in]	db_size	block sizes for diagonal
[in]	eb_size	block sizes for extra diagonal
[in]	symmetric	indicate if matrix is symmetric

Returns

pointer to matrix structure

cs_sles_default_get_verbosity()

int cs_sles_default_get_verbosity	(	int	f_id,
		const char *	name
	)

Return default verbosity associated to a field id, name couple.

Parameters

[in]	f_id	associated field id, or < 0
[in]	name	associated name if f_id < 0, or nullptr

Returns

verbosity associated with field or name

cs_sles_default_release_matrix()

void cs_sles_default_release_matrix

(

cs_matrix_t **

m

)

cs_sles_default_setup()

void cs_sles_default_setup

(

void

)

Default setup for sparse linear equation solver API.

This includes setup logging.

cs_sles_free_native()

void cs_sles_free_native	(	int	f_id,
		const char *	name
	)

Free sparse linear equation solver setup using native matrix arrays.

Parameters

[in]	f_id	associated field id, or < 0
[in]	name	associated name if f_id < 0, or nullptr

cs_sles_setup_native_conv_diff()

void cs_sles_setup_native_conv_diff	(	int	f_id,
		const char *	name,
		const cs_lnum_t	diag_block_size,
		const cs_lnum_t	extra_diag_block_size,
		const cs_real_t *	da,
		const cs_real_t *	xa,
		bool	conv_diff
	)

Call sparse linear equation solver setup for convection-diffusion systems.

Parameters

[in]	f_id	associated field id, or < 0
[in]	name	associated name if f_id < 0, or nullptr
[in]	diag_block_size	block sizes for diagonal, or nullptr
[in]	extra_diag_block_size	block sizes for extra diagonal, or nullptr
[in]	da	diagonal values (nullptr if zero)
[in]	xa	extradiagonal values (nullptr if zero)
[in]	conv_diff	convection-diffusion mode

cs_sles_solve_ccc_fv()

cs_sles_convergence_state_t cs_sles_solve_ccc_fv	(	cs_sles_t *	sc,
		cs_matrix_t *	a,
		double	precision,
		double	r_norm,
		int *	n_iter,
		double *	residual,
		const cs_real_t *	rhs,
		cs_real_t *	vx
	)

Call sparse linear equation solver for general colocated cell-centered finite volume scheme.

The initial solution is assumed to be 0 (and does not need to be initialized before calling this function).

Parameters

[in]	sc	solver context
[in]	a	matrix
[in]	precision	solver precision
[in]	r_norm	residual normalization
[out]	n_iter	number of "equivalent" iterations
[out]	residual	residual
[in]	rhs	right hand side
[out]	vx	system solution

Returns

convergence state

cs_sles_solve_native()

cs_sles_convergence_state_t cs_sles_solve_native	(	int	f_id,
		const char *	name,
		bool	symmetric,
		cs_lnum_t	diag_block_size,
		cs_lnum_t	extra_diag_block_size,
		const cs_real_t *	da,
		const cs_real_t *	xa,
		double	precision,
		double	r_norm,
		int *	n_iter,
		double *	residual,
		const cs_real_t *	rhs,
		cs_real_t *	vx
	)

Call sparse linear equation solver using native matrix arrays.

Parameters

[in]	f_id	associated field id, or < 0
[in]	name	associated name if f_id < 0, or nullptr
[in]	symmetric	indicates if matrix coefficients are symmetric
[in]	diag_block_size	block sizes for diagonal
[in]	extra_diag_block_size	block sizes for extra diagonal
[in]	da	diagonal values (nullptr if zero)
[in]	xa	extradiagonal values (nullptr if zero)
[in]	precision	solver precision
[in]	r_norm	residual normalization
[out]	n_iter	number of "equivalent" iterations
[out]	residual	residual
[in]	rhs	right hand side
[in,out]	vx	system solution

Returns

convergence state