Spm_dev_matvec

Matrix-Vector and matrix-matrix product routines. More...

Files
file	c_spm_matrixvector.c
file	s_spm_matrixvector.c
file	z_spm_matrixvector.c
file	d_spm_matrixvector.c

Functions
int	spm_sspmv (spm_trans_t trans, float alpha, const spmatrix_t *A, const float *x, spm_int_t incx, float beta, float *y, spm_int_t incy)
	compute the matrix-vector product:
int	spm_sspmm (spm_side_t side, spm_trans_t transA, spm_trans_t transB, spm_int_t K, float alpha, const spmatrix_t *A, const float *B, spm_int_t ldb, float beta, float *C, spm_int_t ldc)
	Compute a matrix-matrix product.
int	spm_dspmv (spm_trans_t trans, double alpha, const spmatrix_t *A, const double *x, spm_int_t incx, double beta, double *y, spm_int_t incy)
	compute the matrix-vector product:
int	spm_dspmm (spm_side_t side, spm_trans_t transA, spm_trans_t transB, spm_int_t K, double alpha, const spmatrix_t *A, const double *B, spm_int_t ldb, double beta, double *C, spm_int_t ldc)
	Compute a matrix-matrix product.
int	spm_cspmv (spm_trans_t trans, spm_complex32_t alpha, const spmatrix_t *A, const spm_complex32_t *x, spm_int_t incx, spm_complex32_t beta, spm_complex32_t *y, spm_int_t incy)
	compute the matrix-vector product:
int	spm_cspmm (spm_side_t side, spm_trans_t transA, spm_trans_t transB, spm_int_t K, spm_complex32_t alpha, const spmatrix_t *A, const spm_complex32_t *B, spm_int_t ldb, spm_complex32_t beta, spm_complex32_t *C, spm_int_t ldc)
	Compute a matrix-matrix product.
int	spm_zspmv (spm_trans_t trans, spm_complex64_t alpha, const spmatrix_t *A, const spm_complex64_t *x, spm_int_t incx, spm_complex64_t beta, spm_complex64_t *y, spm_int_t incy)
	compute the matrix-vector product:
int	spm_zspmm (spm_side_t side, spm_trans_t transA, spm_trans_t transB, spm_int_t K, spm_complex64_t alpha, const spmatrix_t *A, const spm_complex64_t *B, spm_int_t ldb, spm_complex64_t beta, spm_complex64_t *C, spm_int_t ldc)
	Compute a matrix-matrix product.

Detailed Description

Matrix-Vector and matrix-matrix product routines.

Function Documentation

spm_sspmv()

int spm_sspmv	(	spm_trans_t	trans,
		float	alpha,
		const spmatrix_t *	A,
		const float *	x,
		spm_int_t	incx,
		float	beta,
		float *	y,
		spm_int_t	incy
	)

compute the matrix-vector product:

A is a SpmSymmetric spm, alpha and beta are scalars, and x and y are vectors, and A a symm.

Parameters

[in]	trans	TODO
[in]	alpha	alpha specifies the scalar alpha
[in]	A	The SpmSymmetric spm.
[in]	x	The vector x.
[in]	incx	The vector x.
[in]	beta	beta specifies the scalar beta
[in,out]	y	The vector y.
[in]	incy	The vector y.

Return values

SPM_SUCCESS	if the y vector has been computed succesfully,
SPM_ERR_BADPARAMETER	otherwise.

Definition at line 1245 of file s_spm_matrixvector.c.

References spmatrix_s::mtxtype, spmatrix_s::nexp, s_spmm_build_Btmp(), s_spmm_build_Ctmp(), s_spmReduceRHS(), spm_get_distribution(), SPM_SUCCESS, SpmDistByColumn, SpmDistByRow, SpmGeneral, SpmLeft, and SpmNoTrans.

Referenced by spmMatVec().

spm_sspmm()

int spm_sspmm	(	spm_side_t	side,
		spm_trans_t	transA,
		spm_trans_t	transB,
		spm_int_t	K,
		float	alpha,
		const spmatrix_t *	A,
		const float *	B,
		spm_int_t	ldb,
		float	beta,
		float *	C,
		spm_int_t	ldc
	)

Compute a matrix-matrix product.

C = alpha * op(A) * op(B) + beta * C or C = alpha * op(B) * op(A) + beta * C

where A is a sparse matrix, B and C two dense matrices. And op(A), op(B) are one of:

op( A ) = A or op( A ) = A' or op( A ) = conjg( A' )

alpha and beta are scalars.

Parameters

[in]	side	Specifies whether A * B is computed or B * A SpmLeft: C = alpha * op(A) * op(B) + beta * C SpmRight: C = alpha * op(B) * op(A) + beta * C
[in]	transA	Specifies whether the sparse matrix A is not transposed, transposed or conjugate transposed: SpmNoTrans SpmTrans SpmTrans
[in]	transB	Specifies whether the dense matrix B is not transposed, transposed or conjugate transposed: SpmNoTrans SpmTrans SpmTrans
[in]	K	If side == SpmLeft, specifies the number of columns of the matrices op(B) and C. If side == SpmRight, specifies the number of rows of the matrices op(B) and C.
[in]	alpha	alpha specifies the scalar alpha.
[in]	A	The sparse matrix A
[in]	B	The matrix B of size: ldb-by-Bn, with Bn = (K, A->m or A->n) based on the configuration of side, transA and transB.
[in]	ldb	The leading dimension of the matrix B. ldb >= (A->m, A->n or K) based on the configuration of side, transA, and transB
[in]	beta	beta specifies the scalar beta.
[in,out]	C	The matrix C of size ldc-by-Cn with Bn = (K, A->m or A->n) based on the configuration of side, transA and transB.
[in]	ldc	The leading dimension of the matrix C. ldc >= (A->m, A->n or K) based on the configuration of side, transA, and transB

side | transA | transB | B | C |

Left | NoTrans | NoTrans | A->n by K | A->m by K | Left | NoTrans | [Conj]Trans | K by A->n | A->m by K | Left | [Conj]Trans | NoTrans | A->m by K | A->n by K | Left | [Conj]Trans | [Conj]Trans | K by A->m | A->n by K | Right | NoTrans | NoTrans | K by A->m | K by A->n | Right | NoTrans | [Conj]Trans | A->m by K | K by A->n | Right | [Conj]Trans | NoTrans | K by A->n | K by A->m | Right | [Conj]Trans | [Conj]Trans | A->n by K | K by A->m |

Return values

SPM_SUCCESS	if the y vector has been computed successfully,
SPM_ERR_BADPARAMETER	otherwise.

Definition at line 1111 of file s_spm_matrixvector.c.

References LAPACKE_slascl_work, spmatrix_s::mtxtype, spmatrix_s::nexp, s_spmm_build_Btmp(), s_spmm_build_Ctmp(), s_spmReduceRHS(), SPM_ERR_BADPARAMETER, SPM_ERR_INTERNAL, spm_get_distribution(), SPM_SUCCESS, SpmDistByColumn, SpmDistByRow, SpmGeneral, SpmLeft, and SpmNoTrans.

Referenced by s_spmCheckAxb(), s_spmGenRHS(), and spmMatMat().

spm_dspmv()

int spm_dspmv	(	spm_trans_t	trans,
		double	alpha,
		const spmatrix_t *	A,
		const double *	x,
		spm_int_t	incx,
		double	beta,
		double *	y,
		spm_int_t	incy
	)

compute the matrix-vector product:

A is a SpmSymmetric spm, alpha and beta are scalars, and x and y are vectors, and A a symm.

Parameters

[in]	trans	TODO
[in]	alpha	alpha specifies the scalar alpha
[in]	A	The SpmSymmetric spm.
[in]	x	The vector x.
[in]	incx	The vector x.
[in]	beta	beta specifies the scalar beta
[in,out]	y	The vector y.
[in]	incy	The vector y.

Return values

SPM_SUCCESS	if the y vector has been computed succesfully,
SPM_ERR_BADPARAMETER	otherwise.

Definition at line 1245 of file d_spm_matrixvector.c.

References d_spmm_build_Btmp(), d_spmm_build_Ctmp(), d_spmReduceRHS(), spmatrix_s::mtxtype, spmatrix_s::nexp, spm_get_distribution(), SPM_SUCCESS, SpmDistByColumn, SpmDistByRow, SpmGeneral, SpmLeft, and SpmNoTrans.

Referenced by spmMatVec().

spm_dspmm()

int spm_dspmm	(	spm_side_t	side,
		spm_trans_t	transA,
		spm_trans_t	transB,
		spm_int_t	K,
		double	alpha,
		const spmatrix_t *	A,
		const double *	B,
		spm_int_t	ldb,
		double	beta,
		double *	C,
		spm_int_t	ldc
	)

Compute a matrix-matrix product.

C = alpha * op(A) * op(B) + beta * C or C = alpha * op(B) * op(A) + beta * C

where A is a sparse matrix, B and C two dense matrices. And op(A), op(B) are one of:

op( A ) = A or op( A ) = A' or op( A ) = conjg( A' )

alpha and beta are scalars.

Parameters

[in]	side	Specifies whether A * B is computed or B * A SpmLeft: C = alpha * op(A) * op(B) + beta * C SpmRight: C = alpha * op(B) * op(A) + beta * C
[in]	transA	Specifies whether the sparse matrix A is not transposed, transposed or conjugate transposed: SpmNoTrans SpmTrans SpmTrans
[in]	transB	Specifies whether the dense matrix B is not transposed, transposed or conjugate transposed: SpmNoTrans SpmTrans SpmTrans
[in]	K	If side == SpmLeft, specifies the number of columns of the matrices op(B) and C. If side == SpmRight, specifies the number of rows of the matrices op(B) and C.
[in]	alpha	alpha specifies the scalar alpha.
[in]	A	The sparse matrix A
[in]	B	The matrix B of size: ldb-by-Bn, with Bn = (K, A->m or A->n) based on the configuration of side, transA and transB.
[in]	ldb	The leading dimension of the matrix B. ldb >= (A->m, A->n or K) based on the configuration of side, transA, and transB
[in]	beta	beta specifies the scalar beta.
[in,out]	C	The matrix C of size ldc-by-Cn with Bn = (K, A->m or A->n) based on the configuration of side, transA and transB.
[in]	ldc	The leading dimension of the matrix C. ldc >= (A->m, A->n or K) based on the configuration of side, transA, and transB

side | transA | transB | B | C |

Left | NoTrans | NoTrans | A->n by K | A->m by K | Left | NoTrans | [Conj]Trans | K by A->n | A->m by K | Left | [Conj]Trans | NoTrans | A->m by K | A->n by K | Left | [Conj]Trans | [Conj]Trans | K by A->m | A->n by K | Right | NoTrans | NoTrans | K by A->m | K by A->n | Right | NoTrans | [Conj]Trans | A->m by K | K by A->n | Right | [Conj]Trans | NoTrans | K by A->n | K by A->m | Right | [Conj]Trans | [Conj]Trans | A->n by K | K by A->m |

Return values

SPM_SUCCESS	if the y vector has been computed successfully,
SPM_ERR_BADPARAMETER	otherwise.

Definition at line 1111 of file d_spm_matrixvector.c.

References d_spmm_build_Btmp(), d_spmm_build_Ctmp(), d_spmReduceRHS(), LAPACKE_dlascl_work, spmatrix_s::mtxtype, spmatrix_s::nexp, SPM_ERR_BADPARAMETER, SPM_ERR_INTERNAL, spm_get_distribution(), SPM_SUCCESS, SpmDistByColumn, SpmDistByRow, SpmGeneral, SpmLeft, and SpmNoTrans.

Referenced by d_spmCheckAxb(), d_spmGenRHS(), and spmMatMat().

spm_cspmv()

int spm_cspmv	(	spm_trans_t	trans,
		spm_complex32_t	alpha,
		const spmatrix_t *	A,
		const spm_complex32_t *	x,
		spm_int_t	incx,
		spm_complex32_t	beta,
		spm_complex32_t *	y,
		spm_int_t	incy
	)

compute the matrix-vector product:

A is a SpmHermitian spm, alpha and beta are scalars, and x and y are vectors, and A a symm.

Parameters

[in]	trans	TODO
[in]	alpha	alpha specifies the scalar alpha
[in]	A	The SpmHermitian spm.
[in]	x	The vector x.
[in]	incx	The vector x.
[in]	beta	beta specifies the scalar beta
[in,out]	y	The vector y.
[in]	incy	The vector y.

Return values

SPM_SUCCESS	if the y vector has been computed succesfully,
SPM_ERR_BADPARAMETER	otherwise.

Definition at line 1245 of file c_spm_matrixvector.c.

References c_spmm_build_Btmp(), c_spmm_build_Ctmp(), c_spmReduceRHS(), CBLAS_SADDR, spmatrix_s::mtxtype, spmatrix_s::nexp, spm_get_distribution(), SPM_SUCCESS, SpmDistByColumn, SpmDistByRow, SpmGeneral, SpmLeft, and SpmNoTrans.

Referenced by spmMatVec().

spm_cspmm()

int spm_cspmm	(	spm_side_t	side,
		spm_trans_t	transA,
		spm_trans_t	transB,
		spm_int_t	K,
		spm_complex32_t	alpha,
		const spmatrix_t *	A,
		const spm_complex32_t *	B,
		spm_int_t	ldb,
		spm_complex32_t	beta,
		spm_complex32_t *	C,
		spm_int_t	ldc
	)

Compute a matrix-matrix product.

C = alpha * op(A) * op(B) + beta * C or C = alpha * op(B) * op(A) + beta * C

where A is a sparse matrix, B and C two dense matrices. And op(A), op(B) are one of:

op( A ) = A or op( A ) = A' or op( A ) = conjg( A' )

alpha and beta are scalars.

Parameters

[in]	side	Specifies whether A * B is computed or B * A SpmLeft: C = alpha * op(A) * op(B) + beta * C SpmRight: C = alpha * op(B) * op(A) + beta * C
[in]	transA	Specifies whether the sparse matrix A is not transposed, transposed or conjugate transposed: SpmNoTrans SpmTrans SpmConjTrans
[in]	transB	Specifies whether the dense matrix B is not transposed, transposed or conjugate transposed: SpmNoTrans SpmTrans SpmConjTrans
[in]	K	If side == SpmLeft, specifies the number of columns of the matrices op(B) and C. If side == SpmRight, specifies the number of rows of the matrices op(B) and C.
[in]	alpha	alpha specifies the scalar alpha.
[in]	A	The sparse matrix A
[in]	B	The matrix B of size: ldb-by-Bn, with Bn = (K, A->m or A->n) based on the configuration of side, transA and transB.
[in]	ldb	The leading dimension of the matrix B. ldb >= (A->m, A->n or K) based on the configuration of side, transA, and transB
[in]	beta	beta specifies the scalar beta.
[in,out]	C	The matrix C of size ldc-by-Cn with Bn = (K, A->m or A->n) based on the configuration of side, transA and transB.
[in]	ldc	The leading dimension of the matrix C. ldc >= (A->m, A->n or K) based on the configuration of side, transA, and transB

side | transA | transB | B | C |

Left | NoTrans | NoTrans | A->n by K | A->m by K | Left | NoTrans | [Conj]Trans | K by A->n | A->m by K | Left | [Conj]Trans | NoTrans | A->m by K | A->n by K | Left | [Conj]Trans | [Conj]Trans | K by A->m | A->n by K | Right | NoTrans | NoTrans | K by A->m | K by A->n | Right | NoTrans | [Conj]Trans | A->m by K | K by A->n | Right | [Conj]Trans | NoTrans | K by A->n | K by A->m | Right | [Conj]Trans | [Conj]Trans | A->n by K | K by A->m |

Return values

SPM_SUCCESS	if the y vector has been computed successfully,
SPM_ERR_BADPARAMETER	otherwise.

Definition at line 1111 of file c_spm_matrixvector.c.

References c_spmm_build_Btmp(), c_spmm_build_Ctmp(), c_spmReduceRHS(), LAPACKE_clascl_work, spmatrix_s::mtxtype, spmatrix_s::nexp, SPM_ERR_BADPARAMETER, SPM_ERR_INTERNAL, spm_get_distribution(), SPM_SUCCESS, SpmDistByColumn, SpmDistByRow, SpmGeneral, SpmLeft, and SpmNoTrans.

Referenced by c_spmCheckAxb(), c_spmGenRHS(), and spmMatMat().

spm_zspmv()

int spm_zspmv	(	spm_trans_t	trans,
		spm_complex64_t	alpha,
		const spmatrix_t *	A,
		const spm_complex64_t *	x,
		spm_int_t	incx,
		spm_complex64_t	beta,
		spm_complex64_t *	y,
		spm_int_t	incy
	)

compute the matrix-vector product:

A is a SpmHermitian spm, alpha and beta are scalars, and x and y are vectors, and A a symm.

Parameters

[in]	trans	TODO
[in]	alpha	alpha specifies the scalar alpha
[in]	A	The SpmHermitian spm.
[in]	x	The vector x.
[in]	incx	The vector x.
[in]	beta	beta specifies the scalar beta
[in,out]	y	The vector y.
[in]	incy	The vector y.

Return values

SPM_SUCCESS	if the y vector has been computed succesfully,
SPM_ERR_BADPARAMETER	otherwise.

Definition at line 1245 of file z_spm_matrixvector.c.

References CBLAS_SADDR, spmatrix_s::mtxtype, spmatrix_s::nexp, spm_get_distribution(), SPM_SUCCESS, SpmDistByColumn, SpmDistByRow, SpmGeneral, SpmLeft, SpmNoTrans, z_spmm_build_Btmp(), z_spmm_build_Ctmp(), and z_spmReduceRHS().

Referenced by spmMatVec().

spm_zspmm()

int spm_zspmm	(	spm_side_t	side,
		spm_trans_t	transA,
		spm_trans_t	transB,
		spm_int_t	K,
		spm_complex64_t	alpha,
		const spmatrix_t *	A,
		const spm_complex64_t *	B,
		spm_int_t	ldb,
		spm_complex64_t	beta,
		spm_complex64_t *	C,
		spm_int_t	ldc
	)

Compute a matrix-matrix product.

C = alpha * op(A) * op(B) + beta * C or C = alpha * op(B) * op(A) + beta * C

where A is a sparse matrix, B and C two dense matrices. And op(A), op(B) are one of:

op( A ) = A or op( A ) = A' or op( A ) = conjg( A' )

alpha and beta are scalars.

Parameters

[in]	side	Specifies whether A * B is computed or B * A SpmLeft: C = alpha * op(A) * op(B) + beta * C SpmRight: C = alpha * op(B) * op(A) + beta * C
[in]	transA	Specifies whether the sparse matrix A is not transposed, transposed or conjugate transposed: SpmNoTrans SpmTrans SpmConjTrans
[in]	transB	Specifies whether the dense matrix B is not transposed, transposed or conjugate transposed: SpmNoTrans SpmTrans SpmConjTrans
[in]	K	If side == SpmLeft, specifies the number of columns of the matrices op(B) and C. If side == SpmRight, specifies the number of rows of the matrices op(B) and C.
[in]	alpha	alpha specifies the scalar alpha.
[in]	A	The sparse matrix A
[in]	B	The matrix B of size: ldb-by-Bn, with Bn = (K, A->m or A->n) based on the configuration of side, transA and transB.
[in]	ldb	The leading dimension of the matrix B. ldb >= (A->m, A->n or K) based on the configuration of side, transA, and transB
[in]	beta	beta specifies the scalar beta.
[in,out]	C	The matrix C of size ldc-by-Cn with Bn = (K, A->m or A->n) based on the configuration of side, transA and transB.
[in]	ldc	The leading dimension of the matrix C. ldc >= (A->m, A->n or K) based on the configuration of side, transA, and transB

side | transA | transB | B | C |

Left | NoTrans | NoTrans | A->n by K | A->m by K | Left | NoTrans | [Conj]Trans | K by A->n | A->m by K | Left | [Conj]Trans | NoTrans | A->m by K | A->n by K | Left | [Conj]Trans | [Conj]Trans | K by A->m | A->n by K | Right | NoTrans | NoTrans | K by A->m | K by A->n | Right | NoTrans | [Conj]Trans | A->m by K | K by A->n | Right | [Conj]Trans | NoTrans | K by A->n | K by A->m | Right | [Conj]Trans | [Conj]Trans | A->n by K | K by A->m |

Return values

SPM_SUCCESS	if the y vector has been computed successfully,
SPM_ERR_BADPARAMETER	otherwise.

Definition at line 1111 of file z_spm_matrixvector.c.

References LAPACKE_zlascl_work, spmatrix_s::mtxtype, spmatrix_s::nexp, SPM_ERR_BADPARAMETER, SPM_ERR_INTERNAL, spm_get_distribution(), SPM_SUCCESS, SpmDistByColumn, SpmDistByRow, SpmGeneral, SpmLeft, SpmNoTrans, z_spmm_build_Btmp(), z_spmm_build_Ctmp(), and z_spmReduceRHS().

Referenced by spmMatMat(), z_spmCheckAxb(), and z_spmGenRHS().