lib_ex_pour_HZpp/Partie_2/algebre_lineaire/sparselib++/sp1_5c/spblas/spmm.cc

679 lines
18 KiB
C++
Executable file

/*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++*/
/* ******** *** SparseLib++ */
/* ******* ** *** *** *** v. 1.5c */
/* ***** *** ******** ******** */
/* ***** *** ******** ******** R. Pozo */
/* ** ******* *** ** *** *** K. Remington */
/* ******** ******** A. Lumsdaine */
/*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++*/
/* */
/* */
/* SparseLib++ : Sparse Matrix Library */
/* */
/* National Institute of Standards and Technology */
/* University of Notre Dame */
/* Authors: R. Pozo, K. Remington, A. Lumsdaine */
/* */
/* NOTICE */
/* */
/* Permission to use, copy, modify, and distribute this software and */
/* its documentation for any purpose and without fee is hereby granted */
/* provided that the above notice appear in all copies and supporting */
/* documentation. */
/* */
/* Neither the Institutions (National Institute of Standards and Technology, */
/* University of Notre Dame) nor the Authors make any representations about */
/* the suitability of this software for any purpose. This software is */
/* provided ``as is'' without expressed or implied warranty. */
/* */
/*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++*/
/*
* Home Grown Sparse BLAS
*
* These are just a subset of the functions described in SPARKER
* Working Note #3.
*
* Would be great if these could be templated some day
*
*/
#include <stdlib.h>
#include <iostream>
using namespace std; //introduces namespace std
//#include <iostream.h>
#include "spblas.h"
#define _SpMatVal(_a,_lda,_row,_col) ((_a)[(_lda)*(_col)+(_row)])
static void CoordMatVec_float(int m, int n, int k, const float &alpha,
const float *val, const int *indx, const int *jndx,
const int &nnz,
const float *b, int ldb, float *c, int ldc)
{
int i, j;
// To make the compiler happy
if (k && m)
;
// Frob these so we can use one-based indexing externally
b -= 1;
c -= 1;
if (alpha == 1.0) {
if (n == 1)
for (j = 0; j < nnz; j++)
c[indx[j]] += b[jndx[j]] * val[j];
else
for (i = 0; i < n; i++)
for (j = 0; j < nnz; j++)
_SpMatVal(c, ldc, indx[j], i) += _SpMatVal(b, ldb, indx[j], i) * val[j];
} else {
if (n == 1)
for (j = 0; j < nnz; j++)
c[indx[j]] += alpha * b[jndx[j]] * val[j];
else
for (i = 0; i < n; i++)
for (j = 0; j < nnz; j++)
_SpMatVal(c, ldc, indx[j], i) +=
alpha * _SpMatVal(b, ldb, indx[j], i) * val[j];
}
}
static void CoordMatVec_double(int m, int n, int k, const double &alpha,
const double *val, const int *indx, const int *jndx,
const int &nnz,
const double *b, int ldb, double *c, int ldc)
{
int i, j;
// To make the compiler happy
if (k && m)
;
// Frob these so we can use one-based indexing externally
b -= 1;
c -= 1;
if (alpha == 1.0) {
if (n == 1)
for (j = 0; j < nnz; j++)
c[indx[j]] += b[jndx[j]] * val[j];
else
for (i = 0; i < n; i++)
for (j = 0; j < nnz; j++)
_SpMatVal(c, ldc, indx[j], i) += _SpMatVal(b, ldb, indx[j], i) * val[j];
} else {
if (n == 1)
for (j = 0; j < nnz; j++)
c[indx[j]] += alpha * b[jndx[j]] * val[j];
else
for (i = 0; i < n; i++)
for (j = 0; j < nnz; j++)
_SpMatVal(c, ldc, indx[j], i) +=
alpha * _SpMatVal(b, ldb, indx[j], i) * val[j];
}
}
static void
CompColMatVec_double(int m, int n, int k, const double &alpha,
const double *val, const int *indx, const int *pntr,
const double *b, int ldb, double *c, int ldc)
{
int i, j, l;
if (alpha == 0.0)
return;
// To make the compiler happy
if (m)
;
// Frob these so we can use one-based indexing externally
c -= 1;
val -= pntr[0];
indx -= pntr[0];
if (alpha == 1.0) {
if (n == 1)
for (i = 0; i < k; i++)
for (j = pntr[i]; j < pntr[i+1]; j++)
c[indx[j]] += b[i] * val[j];
else
for (l = 0; l < n; l++)
for (i = 0; i < k; i++)
for (j = pntr[i]; j < pntr[i+1]; j++)
_SpMatVal(c, ldc, indx[j], l) += _SpMatVal(b, ldb, i, l) * val[j];
} else {
if (n == 1)
for (i = 0; i < k; i++)
for (j = pntr[i]; j < pntr[i+1]; j++)
c[indx[j]] += alpha * b[i] * val[j];
else
for (l = 0; l < n; l++)
for (i = 0; i < k; i++)
for (j = pntr[i]; j < pntr[i+1]; j++)
_SpMatVal(c, ldc, indx[j], l) +=
alpha * _SpMatVal(b, ldb, i, l) * val[j];
}
}
static void CompColMatVec_float(int m, int n, int k, const float &alpha,
const float *val, const int *indx, const int *pntr,
const float *b, int ldb, float *c, int ldc)
{
int i, j, l;
if (alpha == 0.0)
return;
// To make the compiler happy
if (m)
;
// Frob these so we can use one-based indexing externally
c -= 1;
val -= pntr[0];
indx -= pntr[0];
if (alpha == 1.0) {
if (n == 1)
for (i = 0; i < k; i++)
for (j = pntr[i]; j < pntr[i+1]; j++)
c[indx[j]] += b[i] * val[j];
else
for (l = 0; l < n; l++)
for (i = 0; i < k; i++)
for (j = pntr[i]; j < pntr[i+1]; j++)
_SpMatVal(c, ldc, indx[j], l) += _SpMatVal(b, ldb, i, l) * val[j];
} else {
if (n == 1)
for (i = 0; i < k; i++)
for (j = pntr[i]; j < pntr[i+1]; j++)
c[indx[j]] += alpha * b[i] * val[j];
else
for (l = 0; l < n; l++)
for (i = 0; i < k; i++)
for (j = pntr[i]; j < pntr[i+1]; j++)
_SpMatVal(c, ldc, indx[j], l) +=
alpha * _SpMatVal(b, ldb, i, l) * val[j];
}
}
static void
CompRowMatVec_double(int m, int n, int k, const double &alpha,
const double *val, const int *indx, const int *pntr,
const double *b, int ldb, double *c, int ldc)
{
int i, j, l;
if (alpha == 0.0)
return;
// To make the compiler happy
if (m || k)
;
// Frob these so we can use one-based indexing externally
b -= 1;
val -= pntr[0];
indx -= pntr[0];
if (alpha == 1.0) {
if (n == 1)
for (i = 0; i < m; i++)
for (j = pntr[i]; j < pntr[i+1]; j++)
c[i] += b[indx[j]] * val[j];
else
for (l = 0; l < n; l++)
for (i = 0; i < m; i++)
for (j = pntr[i]; j < pntr[i+1]; j++)
_SpMatVal(c, ldc, i, l) += _SpMatVal(b, ldb, indx[j], l) * val[j];
} else {
if (n == 1)
for (i = 0; i < m; i++)
for (j = pntr[i]; j < pntr[i+1]; j++)
c[i] += alpha * b[indx[j]] * val[j];
else
for (l = 0; l < n; l++)
for (i = 0; i < m; i++)
for (j = pntr[i]; j < pntr[i+1]; j++)
_SpMatVal(c, ldc, i, l) +=
alpha * _SpMatVal(b, ldb, indx[j], l) * val[j];
}
}
static void
CompRowMatVec_float(int m, int n, int k, const float &alpha,
const float *val, const int *indx, const int *pntr,
const float *b, int ldb, float *c, int ldc)
{
int i, j, l;
if (alpha == 0.0)
return;
// To make the compiler happy
if (m || k)
;
// Frob these so we can use one-based indexing externally
b -= 1;
val -= pntr[0];
indx -= pntr[0];
if (alpha == 1.0) {
if (n == 1)
for (i = 0; i < m; i++)
for (j = pntr[i]; j < pntr[i+1]; j++)
c[i] += b[indx[j]] * val[j];
else
for (l = 0; l < n; l++)
for (i = 0; i < m; i++)
for (j = pntr[i]; j < pntr[i+1]; j++)
_SpMatVal(c, ldc, i, l) += _SpMatVal(b, ldb, indx[j], l) * val[j];
} else {
if (n == 1)
for (i = 0; i < m; i++)
for (j = pntr[i]; j < pntr[i+1]; j++)
c[i] += alpha * b[indx[j]] * val[j];
else
for (l = 0; l < n; l++)
for (i = 0; i < m; i++)
for (j = pntr[i]; j < pntr[i+1]; j++)
_SpMatVal(c, ldc, i, l) +=
alpha * _SpMatVal(b, ldb, indx[j], l) * val[j];
}
}
static void
ScaleRectangularArray_double(int m, int n, double *c, int ldc,
const double &beta)
{
int i, j;
if (beta == 1.0)
return;
if (beta == 0.0) {
if (n == 1)
for (j = 0; j < m; j++)
c[j] = 0.0;
else
for (i = 0; i < n; i++)
for (j = 0; j < m; j++)
_SpMatVal(c, ldc, j, i) = 0.0;
} else {
if (n == 1)
for (j = 0; j < m; j++)
c[j] *= beta;
else
for (i = 0; i < n; i++)
for (j = 0; j < m; j++)
_SpMatVal(c, ldc, j, i) *= beta;
}
}
static void
ScaleRectangularArray_float(int m, int n, float *c, int ldc,
const double &beta)
{
int i, j;
if (beta == 1.0)
return;
if (beta == 0.0) {
if (n == 1)
for (j = 0; j < m; j++)
c[j] = 0.0;
else
for (i = 0; i < n; i++)
for (j = 0; j < m; j++)
_SpMatVal(c, ldc, j, i) = 0.0;
} else {
if (n == 1)
for (j = 0; j < m; j++)
c[j] *= beta;
else
for (i = 0; i < n; i++)
for (j = 0; j < m; j++)
_SpMatVal(c, ldc, j, i) *= beta;
}
}
/*
* dcoom -- coordinate format matrix-matrix multiply
*
* C <- alpha A B + beta C
*
* Arguments:
*
* int &transa Indicates how to operate with the sparse matrix
* 0 : operate with matrix
* 1 : operate with transpose matrix
* 2 : operate with conjugate transpose matrix
*
* int &m Number of rows in matrix c
*
* int &n Number of columns in matrix c
*
* int &k Number of rows in matrix b
*
* double &alpha Scalar parameter
*
* double &beta Scalar parameter
*
* int descra[] Descriptor argument. Nine element integer array
* descra[0] matrix structure
* 0 : general
* 1 : symmetric
* 2 : Hermition
* 3 : Triangular
* 4 : Anti-Symmetric
* 5 : Diagonal
* descra[1] upper/lower triangular indicator
* 1 : lower
* 2 : upper
* descra[2] main diagonal type
* 0 : non-unit
* 1 : unit
* descra[4] repeated indices?
* 0 : unknown
* 1 : no repeated indices
*
*
* double *val scalar array of length nnz containing matrix entries
*
* int *indx integer array of length nnz containing row indices
*
* int *jndx integer array of length nnz containing column indices
*
* double *b rectangular array with first dimension ldb
*
* double *c rectangular array with first dimension ldc
*
* double *work scratch array of length lwork. lwork should be at least
* max(m,n)
*
*/
void F77NAME(scoomm)
(const int &transa, const int &m, const int &n, const int &k,
const float &alpha,
const int descra[], const float *val,
const int *indx, const int *jndx, const int &nnz,
const float *b, const int &ldb,
const float &beta, float *c, const int &ldc,
float *work, const int &lwork)
{
if (descra[0] != 0) {
cerr << "Must have general matrix" << endl;
exit(1);
}
// To make the compiler happy
if (work && lwork)
;
ScaleRectangularArray_float(m, n, c, ldc, beta);
if (alpha == 0.0)
return;
// Use this hack if transpose is desired
if (transa == 1 || transa == 2) {
const int *itmp = indx;
indx = jndx;
jndx = itmp;
}
CoordMatVec_float(m, n, k, alpha, val, indx, jndx, nnz, b, ldb, c, ldc);
}
void F77NAME(dcoomm)
(const int &transa, const int &m, const int &n, const int &k,
const double &alpha,
const int descra[], const double *val,
const int *indx, const int *jndx, const int &nnz,
const double *b, const int &ldb,
const double &beta, double *c, const int &ldc,
double *work, const int &lwork)
{
if (descra[0] != 0) {
cerr << "Must have general matrix" << endl;
exit(1);
}
// To make the compiler happy
if (work && lwork)
;
ScaleRectangularArray_double(m, n, c, ldc, beta);
if (alpha == 0.0)
return;
// Use this hack if transpose is desired
if (transa == 1 || transa == 2) {
const int *itmp = indx;
indx = jndx;
jndx = itmp;
}
CoordMatVec_double(m, n, k, alpha, val, indx, jndx, nnz, b, ldb, c, ldc);
}
/*
* dcscm -- comp sparse column matrix-matrix multiply
*
* Arguments:
*
* int &transa Indicates how to operate with the sparse matrix
* 0 : operate with matrix
* 1 : operate with transpose matrix
* 2 : operate with conjugate transpose matrix
*
* int &m Number of rows in matrix c
*
* int &n Number of columns in matrix c
*
* int &k Number of rows in matrix b
*
* double &alpha Scalar parameter
*
* double &beta Scalar parameter
*
* int descra[] Descriptor argument. Nine element integer array
* descra[0] matrix structure
* 0 : general
* 1 : symmetric
* 2 : Hermition
* 3 : Triangular
* 4 : Anti-Symmetric
* 5 : Diagonal
* descra[1] upper/lower triangular indicator
* 1 : lower
* 2 : upper
* descra[2] main diagonal type
* 0 : non-unit
* 1 : unit
*
* double *val scalar array of length nnz containing matrix entries
*
* int *indx integer array of length nnz containing row indices
*
* int *pntr integer array of length k+1 such that pntr(j)-pntr(1)
* points to location in val of the first element in column j
*
* double *b rectangular array with first dimension ldb
*
* double *c rectangular array with first dimension ldc
*
* double *work scratch array of length lwork. lwork should be at least
* max(m,n)
*
*/
void F77NAME(scscmm)
(const int &transa, const int &m, const int &n, const int &k,
const float &alpha,
const int descra[], const float *val,
const int *indx, const int *pntr, const float *b, int &ldb,
const float &beta, float *c, const int &ldc,
float *work, const int &lwork)
{
if (descra[0] != 0) {
cerr << "Must have general matrix" << endl;
exit(1);
}
// To make the compiler happy
if (work && lwork)
;
ScaleRectangularArray_float(m, n, c, ldc, beta);
if (transa == 1 || transa == 2)
CompRowMatVec_float(m, n, k, alpha, val, indx, pntr, b, ldb, c, ldc);
else
CompColMatVec_float(m, n, k, alpha, val, indx, pntr, b, ldb, c, ldc);
}
void F77NAME(dcscmm)
(const int &transa, const int &m, const int &n, const int &k,
const double &alpha,
const int descra[], const double *val,
const int *indx, const int *pntr, const double *b, int &ldb,
const double &beta, double *c, const int &ldc,
double *work, const int &lwork)
{
if (descra[0] != 0) {
cerr << "Must have general matrix" << endl;
exit(1);
}
// To make the compiler happy
if (work && lwork)
;
ScaleRectangularArray_double(m, n, c, ldc, beta);
if (transa == 1 || transa == 2)
CompRowMatVec_double(m, n, k, alpha, val, indx, pntr, b, ldb, c, ldc);
else
CompColMatVec_double(m, n, k, alpha, val, indx, pntr, b, ldb, c, ldc);
}
/*
* dcsrm -- comp sparse row matrix-matrix multiply
*
* Arguments:
*
* int &transa Indicates how to operate with the sparse matrix
* 0 : operate with matrix
* 1 : operate with transpose matrix
* 2 : operate with conjugate transpose matrix
*
* int &m Number of rows in matrix c
*
* int &n Number of columns in matrix c
*
* int &k Number of rows in matrix b
*
* double &alpha Scalar parameter
*
* double &beta Scalar parameter
*
* int descra[] Descriptor argument. Nine element integer array
* descra[0] matrix structure
* 0 : general
* 1 : symmetric
* 2 : Hermition
* 3 : Triangular
* 4 : Anti-Symmetric
* 5 : Diagonal
* descra[1] upper/lower triangular indicator
* 1 : lower
* 2 : upper
* descra[2] main diagonal type
* 0 : non-unit
* 1 : unit
*
* double *val scalar array of length nnz containing matrix entries
*
* int *indx integer array of length nnz containing column indices
*
* int *pntr integer array of length k+1 such that pntr(j)-pntr(1)
* points to location in val of the first element in row j
*
* double *b rectangular array with first dimension ldb
*
* double *c rectangular array with first dimension ldc
*
* double *work scratch array of length lwork. lwork should be at least
* max(m,n)
*
*/
void F77NAME(scsrmm)
(const int &transa, const int &m, const int &n, const int &k,
const float &alpha,
const int descra[], const float *val,
const int *indx, const int *pntr, const float *b, int &ldb,
const float &beta, float *c, const int &ldc,
float *work, const int &lwork)
{
if (descra[0] != 0) {
cerr << "Must have general matrix" << endl;
exit(1);
}
// To make the compiler happy
if (work && lwork)
;
ScaleRectangularArray_float(m, n, c, ldc, beta);
if (transa == 1 || transa == 2)
CompColMatVec_float(m, n, k, alpha, val, indx, pntr, b, ldb, c, ldc);
else
CompRowMatVec_float(m, n, k, alpha, val, indx, pntr, b, ldb, c, ldc);
}
void F77NAME(dcsrmm)
(const int &transa, const int &m, const int &n, const int &k,
const double &alpha,
const int descra[], const double *val,
const int *indx, const int *pntr, const double *b, int &ldb,
const double &beta, double *c, const int &ldc,
double *work, const int &lwork)
{
if (descra[0] != 0) {
cerr << "Must have general matrix" << endl;
exit(1);
}
// To make the compiler happy
if (work && lwork)
;
ScaleRectangularArray_double(m, n, c, ldc, beta);
if (transa == 1 || transa == 2)
CompColMatVec_double(m, n, k, alpha, val, indx, pntr, b, ldb, c, ldc);
else
CompRowMatVec_double(m, n, k, alpha, val, indx, pntr, b, ldb, c, ldc);
}