Functions to initialize sparse matrices. More...

Functions
ghost_error	ghost_sparsemat_init_rowfunc (ghost_sparsemat mat, ghost_sparsemat_src_rowfunc src, ghost_mpi_comm mpicomm, double weight)
	Initializes a sparsemat from a row-based callback function. More...

ghost_error	ghost_sparsemat_init_bin (ghost_sparsemat mat, char path, ghost_mpi_comm mpicomm, double weight)
	Initializes a sparsemat from a binary CRS file. More...

ghost_error	ghost_sparsemat_init_mm (ghost_sparsemat mat, char path, ghost_mpi_comm mpicomm, double weight)
	Initializes a sparsemat from a Matrix Market file. More...

ghost_error	ghost_sparsemat_init_crs (ghost_sparsemat mat, ghost_gidx offs, ghost_lidx n, ghost_gidx col, void val, ghost_lidx rpt, ghost_mpi_comm mpicomm, double weight)
	Initializes a sparsemat from local CRS data. More...

Detailed Description

Functions to initialize sparse matrices.

Function Documentation

ghost_error ghost_sparsemat_init_bin	(	ghost_sparsemat *	mat,
		char *	path,
		ghost_mpi_comm	mpicomm,
		double	weight
	)

Initializes a sparsemat from a binary CRS file.

Parameters

mat	The matrix.
path	The source file.
mpicomm	The MPI communicator in which to create this sparsemat.
weight	The weight of each rank in the given MPI communicator.

Returns: GHOST_SUCCESS on success or an error indicator.

Allows the matrix' datatype to be GHOST_DT_NONE. In this case the datatype for the matrix is read from file. Otherwise the matrix datatype has to be valid and compatible.

Here is the call graph for this function:

ghost_error ghost_sparsemat_init_crs	(	ghost_sparsemat *	mat,
		ghost_gidx	offs,
		ghost_lidx	n,
		ghost_gidx *	col,
		void *	val,
		ghost_lidx *	rpt,
		ghost_mpi_comm	mpicomm,
		double	weight
	)

Initializes a sparsemat from local CRS data.

Parameters

mat	The matrix.
offs	The global index of this rank's first row.
n	The local number of rows.
col	The (global) column indices.
val	The values.
rpt	The row pointers.
mpicomm	The MPI communicator in which to create this sparsemat.
weight	The weight of each rank in the given MPI communicator.

Returns: GHOST_SUCCESS on success or an error indicator.

Requires the matrix to have a valid and compatible datatype.

Here is the call graph for this function:

ghost_error ghost_sparsemat_init_mm	(	ghost_sparsemat *	mat,
		char *	path,
		ghost_mpi_comm	mpicomm,
		double	weight
	)

Initializes a sparsemat from a Matrix Market file.

Parameters

mat	The matrix.
path	The source file.
mpicomm	The MPI communicator in which to create this sparsemat.
weight	The weight of each rank in the given MPI communicator.

Returns: GHOST_SUCCESS on success or an error indicator.

Allows the matrix' datatype to be GHOST_DT_NONE or one of the incomplete datatypes GHOST_DT_FLOAT and GHOST_DT_DOUBLE. If the matrix' datatype on entry is GHOST_DT_FLOAT or GHOST_DT_DOUBLE, the file will be interpreted either in single or double precision, respectively. In this case, the datatype will be completed with GHOST_DT_REAL or GHOST_DT_COMPLEX as specified in the input file. If the matrix' datatype on entry is GHOST_DT_NONE, GHOST_DT_DOUBLE is assumed. Otherwise the matrix datatype has to be valid and compatible.

Here is the call graph for this function:

ghost_error ghost_sparsemat_init_rowfunc	(	ghost_sparsemat *	mat,
		ghost_sparsemat_src_rowfunc *	src,
		ghost_mpi_comm	mpicomm,
		double	weight
	)

Initializes a sparsemat from a row-based callback function.

Parameters

mat	The matrix.
src	The source.
mpicomm	The MPI communicator in which to create this sparsemat.
weight	The weight of each rank in the given MPI communicator.

Returns: GHOST_SUCCESS on success or an error indicator.

Requires the matrix to have a valid and compatible datatype.

Here is the call graph for this function: