From 771262669202239258b54394840cc75114f9cb2b Mon Sep 17 00:00:00 2001 From: ukashanoor Date: Tue, 4 Jul 2017 11:55:14 +0530 Subject: after debugging 1 --- 2.3-1/src/c/linearAlgebra/svd/dsvda.c | 184 ++++++++++++++++++++++++++++++++++ 1 file changed, 184 insertions(+) create mode 100644 2.3-1/src/c/linearAlgebra/svd/dsvda.c (limited to '2.3-1/src/c/linearAlgebra/svd') diff --git a/2.3-1/src/c/linearAlgebra/svd/dsvda.c b/2.3-1/src/c/linearAlgebra/svd/dsvda.c new file mode 100644 index 00000000..c3bcfc29 --- /dev/null +++ b/2.3-1/src/c/linearAlgebra/svd/dsvda.c @@ -0,0 +1,184 @@ +/* Copyright (C) 2017 - IIT Bombay - FOSSEE + + This file must be used under the terms of the CeCILL. + This source file is licensed as described in the file COPYING, which + you should have received as part of this distribution. The terms + are also available at + http://www.cecill.info/licences/Licence_CeCILL_V2-en.txt + Author: Sandeep Gupta + Organization: FOSSEE, IIT Bombay + Email: toolbox@scilab.in + + */ + +/*Funtion to find singular value decomposition of given matrix */ + +#include "lapack.h" +#include +#include +#include "string.h" +#include +#include "svd.h" +#include "matrixTranspose.h" + +int min(int a,int b); +int max(int a,int b); + +extern double dgesvd_(char*,char*,int*,int*,double*,int*,double*,double*,int*,\ + double*,int*,double *,int*,int*); + +#define eps 2.22044604925e-16 /* pow(2,-52) */ + +/* DGESVD computes the singular value decomposition (SVD) of a real + M-by-N matrix A, optionally computing the left and/or right singular + vectors. The SVD is written + + A = U * SIGMA * transpose(V) */ + +/*Function support - + +s=svd(X) +[U,S,V]=svd(X) +[U,S,V]=svd(X,0) (obsolete) +[U,S,V]=svd(X,"e") +[U,S,V,rk]=svd(X [,tol]) + +*/ + +double dsvda(double tol,double *in1,int row,int col,double in2,double nout,double *out1, \ + double *out2,double *out3){ + + char JOBU,JOBVT; + int i,j,k; + int LDU=1; /*Leading Dimension of U */ + int LDVT=1; /*Leading Dimension of VT */ + int M = row; + int N = col; + double *buf; + double *S,*U,*VT; + double *WORK; + + int rk; /*Fourth output if needed */ + + /*if((nout > 1 && in2 == 1) && (M != N)){ // [U,S,VT] = svd(x,'e') + if(M > N){ + JOBU = 'S'; + JOBVT = 'A'; + LDVT = N; + } + else{ + JOBU = 'A'; + JOBVT = 'S'; + LDVT = min(M,N); + } + LDU = M; + U = (double*) malloc((double) (LDU)*min(M,N)*sizeof(double)); + VT = (double*) malloc((double) (LDVT)*N*sizeof(double)); + } + else */if(nout > 1){ /* [U,S,VT = svd(x)] */ + JOBU = 'A'; /*If JOBU = 'A', U contains the M-by-M orthogonal matrix U */ + JOBVT = 'A'; /*JOBVT = 'A': all N rows of V**T are returned in the array VT;*/ + LDU = M; + LDVT = N; + U = (double*) malloc((double) M*M*sizeof(double)); + VT = (double*) malloc((double) N*N*sizeof(double)); + } + else{ /* ans = svd(x) */ + JOBU = 'N'; + JOBVT = 'N'; + } + int LDA = max(1,M); + + /* Making a copy of input matrix */ + buf = (double*) malloc((double)M*N*sizeof(double)); + memcpy(buf,in1,M*N*sizeof(double)); + + S = (double*)malloc((double)min(col,row)*sizeof(double)); + + int LWORK = 5*min(M,N); + WORK = (double*)malloc((double)LWORK*sizeof(double)); + int INFO = 0; /*For successful exit */ + + dgesvd_(&JOBU,&JOBVT,&M,&N,buf,&LDA,S,U,&LDU,VT,&LDVT,WORK,&LWORK,&INFO); + /*Subroutine DGESVD from Lapack lib. */ + + if (nout == 1){ /* ans = svd(x)*/ + memcpy(out1,S,min(row,col)*sizeof(double)); + //printf("%lf %lf %lf",*(S),*(S+1),*(S+2)); + } /* [U,S,VT] = svd(x) */ + else if(in2 == 0 && nout > 1){ + memcpy(out1,U,LDU*M*sizeof(double)); + //memcpy(out3,VT,LDVT*min(row,col)*sizeof(double)); + for(j=0;j tol){ + rk = i+1; + } + } + return rk; + } + return 0; +} + +int min(int a,int b){ + if(a > b) return b; + return a; +} + +int max(int a,int b){ + if(a > b) return a; + return b; +} -- cgit