KmlMgSolverCreate

Create a solver handle.

Interface Definition

C interface:

KmlMgStatus KmlMgSolverCreate(KmlMgSolverH* pSolver, MPI_Comm comm, const KmlMgSolverOptions* options);

Parameters

Parameter	Type	Description	Input/Output
pSolver	KmlMgSolverH*	Solver handle. solver cannot be a null pointer.	Input/Output
comm	MPI_Comm	Solver communicator, which must be the same as that in the matrix to be solved. Otherwise, unknown problems may occur.	Input
options	const KmlMgSolverOptions*	Solver configuration.	Input

Return Value

Return Value	Type	Description
KML_MG_OK	int	Normal execution.
KML_MG_ERR_INVALID_PARAM	int	Invalid parameter.
MG_ERR_UNSUPPORTED	int	Unsupported feature.
KML_MG_ERR_NO_MEMORY	int	Memory allocation failed.
KML_MG_ERR_EXCEEDS_LIMIT	int	Value out of the valid range.
KML_MG_ERR_INCOMPATIBLE	int	Incompatible version.
KML_MG_ERR_NO_REQUIRED_FIELD	int	Required field not set.
KML_MG_ERR_UNKNOWN_TYPE	int	Unknown type.
KML_MG_ERR_THIRD_PART	int	Internal third-party call error.
KML_MG_ERR_INCONSISTENT	int	Inconsistent information transferred.
KML_MG_ERR_UNEXPECTED_CALL_ORDER	int	Unexpected call order.
KML_MG_ERR_MATRIX_VALUE_IS_NULL	int	Null matrix value.
KML_MG_ERR_NO_RESOURCE	int	Resources not ready.

Structure Description

KmlMgSolverOptions

Field	Type	Description
fieldMask	uint64_t	Mask that determines to which fields the configuration is applied. If a bit is set to 1, its corresponding field uses a specified value. If a bit is set to 0, its corresponding field uses the default value. See the individual field description for configurable mask bits.
indexType	KmlMgIndexType	Coordinate type. The options are as follows: KML_MG_INDEX_INT32 (Currently, only 32-bit int is supported.) KML_MG_INDEX_INT64 Mask bit: KML_MG_SOLVER_OPTIONS_INDEX_TYPE
kspValueType	KmlMgValueType	Value type of the input matrix. (Currently, it must match that of the pcCalcType field. The supported precision is as follows.) KML_MG_VALUE_FP16 KML_MG_VALUE_FP32 KML_MG_VALUE_FP64 Mask bit: KML_MG_SOLVER_OPTIONS_KSP_VALUE_TYPE
pcValueType	KmlMgValueType	Value type of the preconditioner (multigrid). The options are as follows: (Currently, only FP16, FP32, and FP64 are supported.) KML_MG_VALUE_FP16 KML_MG_VALUE_FP32 KML_MG_VALUE_FP64 Mask bit: KML_MG_MATRIX_INFO_STORE
pcCalcType	KmlMgValueType	Calculation value type of the preconditioner (multigrid). The options are as follows: (Currently, only FP32 and FP64 are supported.) KML_MG_VALUE_FP16 KML_MG_VALUE_FP32 KML_MG_VALUE_FP64 Mask bit: KML_MG_MATRIX_INFO_STORE
ksp	KmlMgSolverKsp	Krylov subspace method type. The options are as follows: KML_MG_SOLVER_NONE KML_MG_SOLVER_BICG KML_MG_SOLVER_CG KML_MG_SOLVER_CR KML_MG_SOLVER_GCR KML_MG_SOLVER_GMRES Mask bit: KML_MG_MATRIX_INFO_STORE

The table below describes the supported precision combinations.

Combination	kspValueType	pcValueType	pcCalcType
1	KML_MG_VALUE_FP64	KML_MG_VALUE_FP64	KML_MG_VALUE_FP64
2	KML_MG_VALUE_FP32	KML_MG_VALUE_FP32	KML_MG_VALUE_FP32
3	KML_MG_VALUE_FP32	KML_MG_VALUE_FP16	KML_MG_VALUE_FP32

Dependencies

#include "mg.h"

For details about the sample code, see Examples.

Parent topic: HMG