cblas.h

/****************************************************************************
 * Copyright 2006 Evan Drumwright
 * This library is distributed under the terms of the GNU General Public 
 * License (found in COPYING).
 ****************************************************************************/

/* This is a copy of the CBLAS standard header. */

#ifndef __PHYSSIM_CBLAS_H__
#define __PHYSSIM_CBLAS_H__
#include <stddef.h>

#undef __BEGIN_DECLS
#undef __END_DECLS
#ifdef __cplusplus
#include <iostream>
#include <cmath>
#define __BEGIN_DECLS extern "C" {
#define __END_DECLS }
#else
#define __BEGIN_DECLS           /* empty */
#define __END_DECLS             /* empty */
#endif

__BEGIN_DECLS

/*
 * Enumerated and derived types
 */
#define CBLAS_INDEX size_t  /* this may vary between platforms */

enum CBLAS_ORDER {CblasRowMajor=101, CblasColMajor=102};
enum CBLAS_TRANSPOSE {CblasNoTrans=111, CblasTrans=112, CblasConjTrans=113};
enum CBLAS_UPLO {CblasUpper=121, CblasLower=122};
enum CBLAS_DIAG {CblasNonUnit=131, CblasUnit=132};
enum CBLAS_SIDE {CblasLeft=141, CblasRight=142};

/*
 * ===========================================================================
 * Prototypes for level 1 BLAS functions (complex are recast as routines)
 * ===========================================================================
 */
float  cblas_sdsdot(const int N, const float alpha, const float *X,
                    const int incX, const float *Y, const int incY);
double cblas_dsdot(const int N, const float *X, const int incX, const float *Y,
                   const int incY);
float  cblas_sdot(const int N, const float  *X, const int incX,
                  const float  *Y, const int incY);
double cblas_ddot(const int N, const double *X, const int incX,
                  const double *Y, const int incY);

/*
 * Functions having prefixes Z and C only
 */
void   cblas_cdotu_sub(const int N, const void *X, const int incX,
                       const void *Y, const int incY, void *dotu);
void   cblas_cdotc_sub(const int N, const void *X, const int incX,
                       const void *Y, const int incY, void *dotc);

void   cblas_zdotu_sub(const int N, const void *X, const int incX,
                       const void *Y, const int incY, void *dotu);
void   cblas_zdotc_sub(const int N, const void *X, const int incX,
                       const void *Y, const int incY, void *dotc);


/*
 * Functions having prefixes S D SC DZ
 */
float  cblas_snrm2(const int N, const float *X, const int incX);
float  cblas_sasum(const int N, const float *X, const int incX);

double cblas_dnrm2(const int N, const double *X, const int incX);
double cblas_dasum(const int N, const double *X, const int incX);

float  cblas_scnrm2(const int N, const void *X, const int incX);
float  cblas_scasum(const int N, const void *X, const int incX);

double cblas_dznrm2(const int N, const void *X, const int incX);
double cblas_dzasum(const int N, const void *X, const int incX);


/*
 * Functions having standard 4 prefixes (S D C Z)
 */
CBLAS_INDEX cblas_isamax(const int N, const float  *X, const int incX);
CBLAS_INDEX cblas_idamax(const int N, const double *X, const int incX);
CBLAS_INDEX cblas_icamax(const int N, const void   *X, const int incX);
CBLAS_INDEX cblas_izamax(const int N, const void   *X, const int incX);

/*
 * ===========================================================================
 * Prototypes for level 1 BLAS routines
 * ===========================================================================
 */

/* 
 * Routines with standard 4 prefixes (s, d, c, z)
 */
void cblas_sswap(const int N, float *X, const int incX, 
                 float *Y, const int incY);
void cblas_scopy(const int N, const float *X, const int incX, 
                 float *Y, const int incY);
void cblas_saxpy(const int N, const float alpha, const float *X,
                 const int incX, float *Y, const int incY);

void cblas_dswap(const int N, double *X, const int incX, 
                 double *Y, const int incY);
void cblas_dcopy(const int N, const double *X, const int incX, 
                 double *Y, const int incY);
void cblas_daxpy(const int N, const double alpha, const double *X,
                 const int incX, double *Y, const int incY);

void cblas_cswap(const int N, void *X, const int incX, 
                 void *Y, const int incY);
void cblas_ccopy(const int N, const void *X, const int incX, 
                 void *Y, const int incY);
void cblas_caxpy(const int N, const void *alpha, const void *X,
                 const int incX, void *Y, const int incY);

void cblas_zswap(const int N, void *X, const int incX, 
                 void *Y, const int incY);
void cblas_zcopy(const int N, const void *X, const int incX, 
                 void *Y, const int incY);
void cblas_zaxpy(const int N, const void *alpha, const void *X,
                 const int incX, void *Y, const int incY);


/* 
 * Routines with S and D prefix only
 */
void cblas_srotg(float *a, float *b, float *c, float *s);
void cblas_srotmg(float *d1, float *d2, float *b1, const float b2, float *P);
void cblas_srot(const int N, float *X, const int incX,
                float *Y, const int incY, const float c, const float s);
void cblas_srotm(const int N, float *X, const int incX,
                float *Y, const int incY, const float *P);

void cblas_drotg(double *a, double *b, double *c, double *s);
void cblas_drotmg(double *d1, double *d2, double *b1, const double b2, double *P);
void cblas_drot(const int N, double *X, const int incX,
                double *Y, const int incY, const double c, const double  s);
void cblas_drotm(const int N, double *X, const int incX,
                double *Y, const int incY, const double *P);


/* 
 * Routines with S D C Z CS and ZD prefixes
 */
void cblas_sscal(const int N, const float alpha, float *X, const int incX);
void cblas_dscal(const int N, const double alpha, double *X, const int incX);
void cblas_cscal(const int N, const void *alpha, void *X, const int incX);
void cblas_zscal(const int N, const void *alpha, void *X, const int incX);
void cblas_csscal(const int N, const float alpha, void *X, const int incX);
void cblas_zdscal(const int N, const double alpha, void *X, const int incX);

/*
 * ===========================================================================
 * Prototypes for level 2 BLAS
 * ===========================================================================
 */

/* 
 * Routines with standard 4 prefixes (S, D, C, Z)
 */
void cblas_sgemv(const enum CBLAS_ORDER order,
                 const enum CBLAS_TRANSPOSE TransA, const int M, const int N,
                 const float alpha, const float *A, const int lda,
                 const float *X, const int incX, const float beta,
                 float *Y, const int incY);
void cblas_sgbmv(const enum CBLAS_ORDER order,
                 const enum CBLAS_TRANSPOSE TransA, const int M, const int N,
                 const int KL, const int KU, const float alpha,
                 const float *A, const int lda, const float *X,
                 const int incX, const float beta, float *Y, const int incY);
void cblas_strmv(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo,
                 const enum CBLAS_TRANSPOSE TransA, const enum CBLAS_DIAG Diag,
                 const int N, const float *A, const int lda, 
                 float *X, const int incX);
void cblas_stbmv(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo,
                 const enum CBLAS_TRANSPOSE TransA, const enum CBLAS_DIAG Diag,
                 const int N, const int K, const float *A, const int lda, 
                 float *X, const int incX);
void cblas_stpmv(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo,
                 const enum CBLAS_TRANSPOSE TransA, const enum CBLAS_DIAG Diag,
                 const int N, const float *Ap, float *X, const int incX);
void cblas_strsv(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo,
                 const enum CBLAS_TRANSPOSE TransA, const enum CBLAS_DIAG Diag,
                 const int N, const float *A, const int lda, float *X,
                 const int incX);
void cblas_stbsv(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo,
                 const enum CBLAS_TRANSPOSE TransA, const enum CBLAS_DIAG Diag,
                 const int N, const int K, const float *A, const int lda,
                 float *X, const int incX);
void cblas_stpsv(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo,
                 const enum CBLAS_TRANSPOSE TransA, const enum CBLAS_DIAG Diag,
                 const int N, const float *Ap, float *X, const int incX);

void cblas_dgemv(const enum CBLAS_ORDER order,
                 const enum CBLAS_TRANSPOSE TransA, const int M, const int N,
                 const double alpha, const double *A, const int lda,
                 const double *X, const int incX, const double beta,
                 double *Y, const int incY);
void cblas_dgbmv(const enum CBLAS_ORDER order,
                 const enum CBLAS_TRANSPOSE TransA, const int M, const int N,
                 const int KL, const int KU, const double alpha,
                 const double *A, const int lda, const double *X,
                 const int incX, const double beta, double *Y, const int incY);
void cblas_dtrmv(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo,
                 const enum CBLAS_TRANSPOSE TransA, const enum CBLAS_DIAG Diag,
                 const int N, const double *A, const int lda, 
                 double *X, const int incX);
void cblas_dtbmv(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo,
                 const enum CBLAS_TRANSPOSE TransA, const enum CBLAS_DIAG Diag,
                 const int N, const int K, const double *A, const int lda, 
                 double *X, const int incX);
void cblas_dtpmv(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo,
                 const enum CBLAS_TRANSPOSE TransA, const enum CBLAS_DIAG Diag,
                 const int N, const double *Ap, double *X, const int incX);
void cblas_dtrsv(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo,
                 const enum CBLAS_TRANSPOSE TransA, const enum CBLAS_DIAG Diag,
                 const int N, const double *A, const int lda, double *X,
                 const int incX);
void cblas_dtbsv(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo,
                 const enum CBLAS_TRANSPOSE TransA, const enum CBLAS_DIAG Diag,
                 const int N, const int K, const double *A, const int lda,
                 double *X, const int incX);
void cblas_dtpsv(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo,
                 const enum CBLAS_TRANSPOSE TransA, const enum CBLAS_DIAG Diag,
                 const int N, const double *Ap, double *X, const int incX);

void cblas_cgemv(const enum CBLAS_ORDER order,
                 const enum CBLAS_TRANSPOSE TransA, const int M, const int N,
                 const void *alpha, const void *A, const int lda,
                 const void *X, const int incX, const void *beta,
                 void *Y, const int incY);
void cblas_cgbmv(const enum CBLAS_ORDER order,
                 const enum CBLAS_TRANSPOSE TransA, const int M, const int N,
                 const int KL, const int KU, const void *alpha,
                 const void *A, const int lda, const void *X,
                 const int incX, const void *beta, void *Y, const int incY);
void cblas_ctrmv(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo,
                 const enum CBLAS_TRANSPOSE TransA, const enum CBLAS_DIAG Diag,
                 const int N, const void *A, const int lda, 
                 void *X, const int incX);
void cblas_ctbmv(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo,
                 const enum CBLAS_TRANSPOSE TransA, const enum CBLAS_DIAG Diag,
                 const int N, const int K, const void *A, const int lda, 
                 void *X, const int incX);
void cblas_ctpmv(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo,
                 const enum CBLAS_TRANSPOSE TransA, const enum CBLAS_DIAG Diag,
                 const int N, const void *Ap, void *X, const int incX);
void cblas_ctrsv(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo,
                 const enum CBLAS_TRANSPOSE TransA, const enum CBLAS_DIAG Diag,
                 const int N, const void *A, const int lda, void *X,
                 const int incX);
void cblas_ctbsv(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo,
                 const enum CBLAS_TRANSPOSE TransA, const enum CBLAS_DIAG Diag,
                 const int N, const int K, const void *A, const int lda,
                 void *X, const int incX);
void cblas_ctpsv(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo,
                 const enum CBLAS_TRANSPOSE TransA, const enum CBLAS_DIAG Diag,
                 const int N, const void *Ap, void *X, const int incX);

void cblas_zgemv(const enum CBLAS_ORDER order,
                 const enum CBLAS_TRANSPOSE TransA, const int M, const int N,
                 const void *alpha, const void *A, const int lda,
                 const void *X, const int incX, const void *beta,
                 void *Y, const int incY);
void cblas_zgbmv(const enum CBLAS_ORDER order,
                 const enum CBLAS_TRANSPOSE TransA, const int M, const int N,
                 const int KL, const int KU, const void *alpha,
                 const void *A, const int lda, const void *X,
                 const int incX, const void *beta, void *Y, const int incY);
void cblas_ztrmv(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo,
                 const enum CBLAS_TRANSPOSE TransA, const enum CBLAS_DIAG Diag,
                 const int N, const void *A, const int lda, 
                 void *X, const int incX);
void cblas_ztbmv(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo,
                 const enum CBLAS_TRANSPOSE TransA, const enum CBLAS_DIAG Diag,
                 const int N, const int K, const void *A, const int lda, 
                 void *X, const int incX);
void cblas_ztpmv(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo,
                 const enum CBLAS_TRANSPOSE TransA, const enum CBLAS_DIAG Diag,
                 const int N, const void *Ap, void *X, const int incX);
void cblas_ztrsv(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo,
                 const enum CBLAS_TRANSPOSE TransA, const enum CBLAS_DIAG Diag,
                 const int N, const void *A, const int lda, void *X,
                 const int incX);
void cblas_ztbsv(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo,
                 const enum CBLAS_TRANSPOSE TransA, const enum CBLAS_DIAG Diag,
                 const int N, const int K, const void *A, const int lda,
                 void *X, const int incX);
void cblas_ztpsv(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo,
                 const enum CBLAS_TRANSPOSE TransA, const enum CBLAS_DIAG Diag,
                 const int N, const void *Ap, void *X, const int incX);


/* 
 * Routines with S and D prefixes only
 */
void cblas_ssymv(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo,
                 const int N, const float alpha, const float *A,
                 const int lda, const float *X, const int incX,
                 const float beta, float *Y, const int incY);
void cblas_ssbmv(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo,
                 const int N, const int K, const float alpha, const float *A,
                 const int lda, const float *X, const int incX,
                 const float beta, float *Y, const int incY);
void cblas_sspmv(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo,
                 const int N, const float alpha, const float *Ap,
                 const float *X, const int incX,
                 const float beta, float *Y, const int incY);
void cblas_sger(const enum CBLAS_ORDER order, const int M, const int N,
                const float alpha, const float *X, const int incX,
                const float *Y, const int incY, float *A, const int lda);
void cblas_ssyr(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo,
                const int N, const float alpha, const float *X,
                const int incX, float *A, const int lda);
void cblas_sspr(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo,
                const int N, const float alpha, const float *X,
                const int incX, float *Ap);
void cblas_ssyr2(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo,
                const int N, const float alpha, const float *X,
                const int incX, const float *Y, const int incY, float *A,
                const int lda);
void cblas_sspr2(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo,
                const int N, const float alpha, const float *X,
                const int incX, const float *Y, const int incY, float *A);

void cblas_dsymv(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo,
                 const int N, const double alpha, const double *A,
                 const int lda, const double *X, const int incX,
                 const double beta, double *Y, const int incY);
void cblas_dsbmv(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo,
                 const int N, const int K, const double alpha, const double *A,
                 const int lda, const double *X, const int incX,
                 const double beta, double *Y, const int incY);
void cblas_dspmv(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo,
                 const int N, const double alpha, const double *Ap,
                 const double *X, const int incX,
                 const double beta, double *Y, const int incY);
void cblas_dger(const enum CBLAS_ORDER order, const int M, const int N,
                const double alpha, const double *X, const int incX,
                const double *Y, const int incY, double *A, const int lda);
void cblas_dsyr(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo,
                const int N, const double alpha, const double *X,
                const int incX, double *A, const int lda);
void cblas_dspr(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo,
                const int N, const double alpha, const double *X,
                const int incX, double *Ap);
void cblas_dsyr2(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo,
                const int N, const double alpha, const double *X,
                const int incX, const double *Y, const int incY, double *A,
                const int lda);
void cblas_dspr2(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo,
                const int N, const double alpha, const double *X,
                const int incX, const double *Y, const int incY, double *A);


/* 
 * Routines with C and Z prefixes only
 */
void cblas_chemv(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo,
                 const int N, const void *alpha, const void *A,
                 const int lda, const void *X, const int incX,
                 const void *beta, void *Y, const int incY);
void cblas_chbmv(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo,
                 const int N, const int K, const void *alpha, const void *A,
                 const int lda, const void *X, const int incX,
                 const void *beta, void *Y, const int incY);
void cblas_chpmv(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo,
                 const int N, const void *alpha, const void *Ap,
                 const void *X, const int incX,
                 const void *beta, void *Y, const int incY);
void cblas_cgeru(const enum CBLAS_ORDER order, const int M, const int N,
                 const void *alpha, const void *X, const int incX,
                 const void *Y, const int incY, void *A, const int lda);
void cblas_cgerc(const enum CBLAS_ORDER order, const int M, const int N,
                 const void *alpha, const void *X, const int incX,
                 const void *Y, const int incY, void *A, const int lda);
void cblas_cher(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo,
                const int N, const float alpha, const void *X, const int incX,
                void *A, const int lda);
void cblas_chpr(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo,
                const int N, const float alpha, const void *X,
                const int incX, void *A);
void cblas_cher2(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo, const int N,
                const void *alpha, const void *X, const int incX,
                const void *Y, const int incY, void *A, const int lda);
void cblas_chpr2(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo, const int N,
                const void *alpha, const void *X, const int incX,
                const void *Y, const int incY, void *Ap);

void cblas_zhemv(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo,
                 const int N, const void *alpha, const void *A,
                 const int lda, const void *X, const int incX,
                 const void *beta, void *Y, const int incY);
void cblas_zhbmv(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo,
                 const int N, const int K, const void *alpha, const void *A,
                 const int lda, const void *X, const int incX,
                 const void *beta, void *Y, const int incY);
void cblas_zhpmv(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo,
                 const int N, const void *alpha, const void *Ap,
                 const void *X, const int incX,
                 const void *beta, void *Y, const int incY);
void cblas_zgeru(const enum CBLAS_ORDER order, const int M, const int N,
                 const void *alpha, const void *X, const int incX,
                 const void *Y, const int incY, void *A, const int lda);
void cblas_zgerc(const enum CBLAS_ORDER order, const int M, const int N,
                 const void *alpha, const void *X, const int incX,
                 const void *Y, const int incY, void *A, const int lda);
void cblas_zher(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo,
                const int N, const double alpha, const void *X, const int incX,
                void *A, const int lda);
void cblas_zhpr(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo,
                const int N, const double alpha, const void *X,
                const int incX, void *A);
void cblas_zher2(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo, const int N,
                const void *alpha, const void *X, const int incX,
                const void *Y, const int incY, void *A, const int lda);
void cblas_zhpr2(const enum CBLAS_ORDER order, const enum CBLAS_UPLO Uplo, const int N,
                const void *alpha, const void *X, const int incX,
                const void *Y, const int incY, void *Ap);

/*
 * ===========================================================================
 * Prototypes for level 3 BLAS
 * ===========================================================================
 */

/* 
 * Routines with standard 4 prefixes (S, D, C, Z)
 */
void cblas_sgemm(const enum CBLAS_ORDER Order, const enum CBLAS_TRANSPOSE TransA,
                 const enum CBLAS_TRANSPOSE TransB, const int M, const int N,
                 const int K, const float alpha, const float *A,
                 const int lda, const float *B, const int ldb,
                 const float beta, float *C, const int ldc);
void cblas_ssymm(const enum CBLAS_ORDER Order, const enum CBLAS_SIDE Side,
                 const enum CBLAS_UPLO Uplo, const int M, const int N,
                 const float alpha, const float *A, const int lda,
                 const float *B, const int ldb, const float beta,
                 float *C, const int ldc);
void cblas_ssyrk(const enum CBLAS_ORDER Order, const enum CBLAS_UPLO Uplo,
                 const enum CBLAS_TRANSPOSE Trans, const int N, const int K,
                 const float alpha, const float *A, const int lda,
                 const float beta, float *C, const int ldc);
void cblas_ssyr2k(const enum CBLAS_ORDER Order, const enum CBLAS_UPLO Uplo,
                  const enum CBLAS_TRANSPOSE Trans, const int N, const int K,
                  const float alpha, const float *A, const int lda,
                  const float *B, const int ldb, const float beta,
                  float *C, const int ldc);
void cblas_strmm(const enum CBLAS_ORDER Order, const enum CBLAS_SIDE Side,
                 const enum CBLAS_UPLO Uplo, const enum CBLAS_TRANSPOSE TransA,
                 const enum CBLAS_DIAG Diag, const int M, const int N,
                 const float alpha, const float *A, const int lda,
                 float *B, const int ldb);
void cblas_strsm(const enum CBLAS_ORDER Order, const enum CBLAS_SIDE Side,
                 const enum CBLAS_UPLO Uplo, const enum CBLAS_TRANSPOSE TransA,
                 const enum CBLAS_DIAG Diag, const int M, const int N,
                 const float alpha, const float *A, const int lda,
                 float *B, const int ldb);

void cblas_dgemm(const enum CBLAS_ORDER Order, const enum CBLAS_TRANSPOSE TransA,
                 const enum CBLAS_TRANSPOSE TransB, const int M, const int N,
                 const int K, const double alpha, const double *A,
                 const int lda, const double *B, const int ldb,
                 const double beta, double *C, const int ldc);
void cblas_dsymm(const enum CBLAS_ORDER Order, const enum CBLAS_SIDE Side,
                 const enum CBLAS_UPLO Uplo, const int M, const int N,
                 const double alpha, const double *A, const int lda,
                 const double *B, const int ldb, const double beta,
                 double *C, const int ldc);
void cblas_dsyrk(const enum CBLAS_ORDER Order, const enum CBLAS_UPLO Uplo,
                 const enum CBLAS_TRANSPOSE Trans, const int N, const int K,
                 const double alpha, const double *A, const int lda,
                 const double beta, double *C, const int ldc);
void cblas_dsyr2k(const enum CBLAS_ORDER Order, const enum CBLAS_UPLO Uplo,
                  const enum CBLAS_TRANSPOSE Trans, const int N, const int K,
                  const double alpha, const double *A, const int lda,
                  const double *B, const int ldb, const double beta,
                  double *C, const int ldc);
void cblas_dtrmm(const enum CBLAS_ORDER Order, const enum CBLAS_SIDE Side,
                 const enum CBLAS_UPLO Uplo, const enum CBLAS_TRANSPOSE TransA,
                 const enum CBLAS_DIAG Diag, const int M, const int N,
                 const double alpha, const double *A, const int lda,
                 double *B, const int ldb);
void cblas_dtrsm(const enum CBLAS_ORDER Order, const enum CBLAS_SIDE Side,
                 const enum CBLAS_UPLO Uplo, const enum CBLAS_TRANSPOSE TransA,
                 const enum CBLAS_DIAG Diag, const int M, const int N,
                 const double alpha, const double *A, const int lda,
                 double *B, const int ldb);

void cblas_cgemm(const enum CBLAS_ORDER Order, const enum CBLAS_TRANSPOSE TransA,
                 const enum CBLAS_TRANSPOSE TransB, const int M, const int N,
                 const int K, const void *alpha, const void *A,
                 const int lda, const void *B, const int ldb,
                 const void *beta, void *C, const int ldc);
void cblas_csymm(const enum CBLAS_ORDER Order, const enum CBLAS_SIDE Side,
                 const enum CBLAS_UPLO Uplo, const int M, const int N,
                 const void *alpha, const void *A, const int lda,
                 const void *B, const int ldb, const void *beta,
                 void *C, const int ldc);
void cblas_csyrk(const enum CBLAS_ORDER Order, const enum CBLAS_UPLO Uplo,
                 const enum CBLAS_TRANSPOSE Trans, const int N, const int K,
                 const void *alpha, const void *A, const int lda,
                 const void *beta, void *C, const int ldc);
void cblas_csyr2k(const enum CBLAS_ORDER Order, const enum CBLAS_UPLO Uplo,
                  const enum CBLAS_TRANSPOSE Trans, const int N, const int K,
                  const void *alpha, const void *A, const int lda,
                  const void *B, const int ldb, const void *beta,
                  void *C, const int ldc);
void cblas_ctrmm(const enum CBLAS_ORDER Order, const enum CBLAS_SIDE Side,
                 const enum CBLAS_UPLO Uplo, const enum CBLAS_TRANSPOSE TransA,
                 const enum CBLAS_DIAG Diag, const int M, const int N,
                 const void *alpha, const void *A, const int lda,
                 void *B, const int ldb);
void cblas_ctrsm(const enum CBLAS_ORDER Order, const enum CBLAS_SIDE Side,
                 const enum CBLAS_UPLO Uplo, const enum CBLAS_TRANSPOSE TransA,
                 const enum CBLAS_DIAG Diag, const int M, const int N,
                 const void *alpha, const void *A, const int lda,
                 void *B, const int ldb);

void cblas_zgemm(const enum CBLAS_ORDER Order, const enum CBLAS_TRANSPOSE TransA,
                 const enum CBLAS_TRANSPOSE TransB, const int M, const int N,
                 const int K, const void *alpha, const void *A,
                 const int lda, const void *B, const int ldb,
                 const void *beta, void *C, const int ldc);
void cblas_zsymm(const enum CBLAS_ORDER Order, const enum CBLAS_SIDE Side,
                 const enum CBLAS_UPLO Uplo, const int M, const int N,
                 const void *alpha, const void *A, const int lda,
                 const void *B, const int ldb, const void *beta,
                 void *C, const int ldc);
void cblas_zsyrk(const enum CBLAS_ORDER Order, const enum CBLAS_UPLO Uplo,
                 const enum CBLAS_TRANSPOSE Trans, const int N, const int K,
                 const void *alpha, const void *A, const int lda,
                 const void *beta, void *C, const int ldc);
void cblas_zsyr2k(const enum CBLAS_ORDER Order, const enum CBLAS_UPLO Uplo,
                  const enum CBLAS_TRANSPOSE Trans, const int N, const int K,
                  const void *alpha, const void *A, const int lda,
                  const void *B, const int ldb, const void *beta,
                  void *C, const int ldc);
void cblas_ztrmm(const enum CBLAS_ORDER Order, const enum CBLAS_SIDE Side,
                 const enum CBLAS_UPLO Uplo, const enum CBLAS_TRANSPOSE TransA,
                 const enum CBLAS_DIAG Diag, const int M, const int N,
                 const void *alpha, const void *A, const int lda,
                 void *B, const int ldb);
void cblas_ztrsm(const enum CBLAS_ORDER Order, const enum CBLAS_SIDE Side,
                 const enum CBLAS_UPLO Uplo, const enum CBLAS_TRANSPOSE TransA,
                 const enum CBLAS_DIAG Diag, const int M, const int N,
                 const void *alpha, const void *A, const int lda,
                 void *B, const int ldb);


/* 
 * Routines with prefixes C and Z only
 */
void cblas_chemm(const enum CBLAS_ORDER Order, const enum CBLAS_SIDE Side,
                 const enum CBLAS_UPLO Uplo, const int M, const int N,
                 const void *alpha, const void *A, const int lda,
                 const void *B, const int ldb, const void *beta,
                 void *C, const int ldc);
void cblas_cherk(const enum CBLAS_ORDER Order, const enum CBLAS_UPLO Uplo,
                 const enum CBLAS_TRANSPOSE Trans, const int N, const int K,
                 const float alpha, const void *A, const int lda,
                 const float beta, void *C, const int ldc);
void cblas_cher2k(const enum CBLAS_ORDER Order, const enum CBLAS_UPLO Uplo,
                  const enum CBLAS_TRANSPOSE Trans, const int N, const int K,
                  const void *alpha, const void *A, const int lda,
                  const void *B, const int ldb, const float beta,
                  void *C, const int ldc);

void cblas_zhemm(const enum CBLAS_ORDER Order, const enum CBLAS_SIDE Side,
                 const enum CBLAS_UPLO Uplo, const int M, const int N,
                 const void *alpha, const void *A, const int lda,
                 const void *B, const int ldb, const void *beta,
                 void *C, const int ldc);
void cblas_zherk(const enum CBLAS_ORDER Order, const enum CBLAS_UPLO Uplo,
                 const enum CBLAS_TRANSPOSE Trans, const int N, const int K,
                 const double alpha, const void *A, const int lda,
                 const double beta, void *C, const int ldc);
void cblas_zher2k(const enum CBLAS_ORDER Order, const enum CBLAS_UPLO Uplo,
                  const enum CBLAS_TRANSPOSE Trans, const int N, const int K,
                  const void *alpha, const void *A, const int lda,
                  const void *B, const int ldb, const double beta,
                  void *C, const int ldc);

void cblas_xerbla(int p, const char *rout, const char *form, ...);

__END_DECLS

class CBLAS
{
        public:
                template <class T>
                static void scal(int N, T alpha, T* X, int incX);

                template <class T>
                static void copy(int N, const T* X, int incX, T* Y, int incY);

                template <class T>
                static void axpy(int N, T alpha, const T* X, int incX, T* Y, int incY);

                template <class T>
                static T dot(int N, const T* X, int incX, const T* Y, int incY);

                template <class T>
                static T nrm2(int N, const T* X, int incX);             

                template <class T>
                static void ger(enum CBLAS_ORDER order, int M, int N, T alpha, const T* X, int incX, const T* Y, int incY, T* A, int lda);

                template <class T>
                static void gemv(enum CBLAS_ORDER order, CBLAS_TRANSPOSE transA, int M, int N, T alpha, const T* A, int lda, const T* X, int incX, T beta, T* Y, int incY);

                template <class T>
                static void gemm(enum CBLAS_ORDER order, CBLAS_TRANSPOSE transA, CBLAS_TRANSPOSE transB, int M, int N, int K, T alpha, const T* A, int lda, const T* B, int ldb, T beta, T* C, int ldc);
};

/*
template <class T>
void CBLAS::scal(int N, T alpha, T* X, int incX)
{
        #ifndef _DISABLE_SLOW_CBLAS_WARNING_
        std::cout << "CBLAS::scal() warning- using slow version!" << std::endl;
        #endif

        // if increment is 1, use simple version
        if (incX == 1)
                for (int i=0; i< N; i++)
                        X[i] *= alpha;
        else
        {
                int ix = 1;
                if (incX < 0)
                        ix = (-N+1)*incX + 1;
                for (int i=0; i< N; i++, ix += incX)
                        X[ix] *= alpha;
        }
}

template <class T>
void CBLAS::copy(int N, const T* X, int incX, T* Y, int incY)
{
        // if both increments are 1, use memcpy
        if (incX == 1 && incY == 1)
                memcpy(Y, X, sizeof(T) * N);
        else
        {
                int ix = 1, iy = 1;
                if (incX < 0)
                        ix = (-N+1)*incX + 1;
                if (incY < 0)
                        iy = (-N+1)*incY + 1;
                for (int i=0; i< N; i++, ix+= incX, iy += incY)
                        Y[iy] = X[ix];
        }
}

template <class T>
void CBLAS::axpy(int N, T alpha, const T* X, int incX, T* Y, int incY)
{
        #ifndef _DISABLE_SLOW_CBLAS_WARNING_
        std::cout << "CBLAS::axpy() warning- using slow version!" << std::endl;
        #endif

        // if increments are both 1, use simple version
        if (incX == 1 && incY == 1)
                for (int i=0; i< N; i++)
                        Y[i] += X[i] * alpha;
        else
        {
                int ix = 1, iy = 1;
                if (incX < 0)
                        ix = (-N+1)*incX + 1;
                if (incY < 0)
                        iy = (-N+1)*incY + 1;
                for (int i=0; i< N; i++, ix+= incX, iy += incY)
                        Y[iy] += X[ix] * alpha;
        }
}

template <class T>
T CBLAS::dot(int N, const T* X, int incX, const T* Y, int incY)
{
        #ifndef _DISABLE_SLOW_CBLAS_WARNING_
        std::cout << "CBLAS::dot() warning- using slow version!" << std::endl;
        #endif

        T inner_prod = 0;

        // if increments are both 1, use simple version
        if (incX == 1 && incY == 1)
                for (int i=0; i< N; i++)
                        inner_prod += X[i] * Y[i];
        else
        {
                int ix = 1, iy = 1;
                if (incX < 0)
                        ix = (-N+1)*incX + 1;
                if (incY < 0)
                        iy = (-N+1)*incY + 1;
                for (int i=0; i< N; i++, ix+= incX, iy += incY)
                        inner_prod += X[ix] * Y[iy];
        }

        return inner_prod;
}

template <class T>
T CBLAS::nrm2(int N, const T* X, int incX)
{
        #ifndef _DISABLE_SLOW_CBLAS_WARNING_
        std::cout << "CBLAS::nrm2() warning- using slow version!" << std::endl;
        #endif

        T norm = 0;

        // if increment is 1, use simple version
        if (incX == 1)
                for (int i=0; i< N; i++)
                        norm += X[i] * X[i];
        else
        {
                int ix = 1;
                if (incX < 0)
                        ix = (-N+1)*incX + 1;
                for (int i=0; i< N; i++, ix += incX)
                        norm += X[ix] * X[ix];
        }

        return std::sqrt(norm);
}

template <class T>
void CBLAS::ger(enum CBLAS_ORDER order, int M, int N, T alpha, const T* X, int incX, const T* Y, int incY, T* A, int lda)
{
        #ifndef _DISABLE_SLOW_CBLAS_WARNING_
        std::cout << "CBLAS::ger() warning- using slow version!" << std::endl;
        #endif

        // if incX and incY are both 1, use simpler version
        if (incX == 1 && incY == 1)
        {
                for (int i=0; i< M; i++)
                        for (int j=0; j< N; j++)
                        {
                                if (order == CblasRowMajor)
                                        A[i*N+j] = X[i] * Y[j];
                                else
                                        A[j*M+i] = X[i] * Y[j];
                        }
        }
        else
        {
                int ix = 1;
                if (incX < 0)
                        ix = (-M+1)*incX + 1;
                for (int i=0; i< M; i++, ix+= incX)
                {
                        int iy = 1;
                        if (incY < 0)
                                iy = (-N+1)*incY + 1;
                        for (int j=0; j< N; j++, iy += incY)
                        {
                                if (order == CblasRowMajor)
                                        A[i*N+j] = X[ix] * Y[iy];
                                else
                                        A[j*M+i] = X[ix] * Y[iy];
                        }
                }
        }
}

template <class T>
void CBLAS::gemv(enum CBLAS_ORDER order, CBLAS_TRANSPOSE transA, int M, int N, T alpha, const T* A, int lda, const T* X, int incX, T beta, T* Y, int incY)
{
        #ifndef _DISABLE_SLOW_CBLAS_WARNING_
        std::cout << "CBLAS::gemv() warning- using slow version!" << std::endl;
        #endif

        bool trans = ((order == CblasColMajor && transA == CblasTrans) ||
                                                                (order == CblasRowMajor && transA == CblasNoTrans));

        int iy = 1;
        if (incY < 0)
                iy = (-N+1)*incY + 1;
        for (int i=0; i< lda; i++, iy+= incY)
        {
                if (!trans)
                                Y[iy] = beta*Y[iy] + alpha * dot(N, A+i, lda, X, incX);
                else
                                Y[iy] = beta*Y[iy] + alpha * dot(N, A+lda*i, 1, X, incX);
        }
}

template <class T>
void CBLAS::gemm(enum CBLAS_ORDER order, CBLAS_TRANSPOSE transA, CBLAS_TRANSPOSE transB, int M, int N, int K, T alpha, const T* A, int lda, const T* B, int ldb, T beta, T* C, int ldc)
{
        #ifndef _DISABLE_SLOW_CBLAS_WARNING_
        std::cout << "CBLAS::gemm() warning- using slow version!" << std::endl;
        #endif

        std::cerr << "CBLAS::gemm() error- not yet implemented!" << std::endl;
}
*/

#endif /* __GSL_CBLAS_H__ */

---