• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 // This file is part of Eigen, a lightweight C++ template library
2 // for linear algebra.
3 //
4 // Copyright (C) 2014 Gael Guennebaud <gael.guennebaud@inria.fr>
5 //
6 // This Source Code Form is subject to the terms of the Mozilla
7 // Public License v. 2.0. If a copy of the MPL was not distributed
8 // with this file, You can obtain one at http://mozilla.org/MPL/2.0/.
9 
10 #include "lapack_common.h"
11 #include <Eigen/SVD>
12 
13 // computes the singular values/vectors a general M-by-N matrix A using divide-and-conquer
14 EIGEN_LAPACK_FUNC(gesdd,(char *jobz, int *m, int* n, Scalar* a, int *lda, RealScalar *s, Scalar *u, int *ldu, Scalar *vt, int *ldvt, Scalar* /*work*/, int* lwork,
15                          EIGEN_LAPACK_ARG_IF_COMPLEX(RealScalar */*rwork*/) int * /*iwork*/, int *info))
16 {
17   // TODO exploit the work buffer
18   bool query_size = *lwork==-1;
19   int diag_size = (std::min)(*m,*n);
20 
21   *info = 0;
22         if(*jobz!='A' && *jobz!='S' && *jobz!='O' && *jobz!='N')  *info = -1;
23   else  if(*m<0)                                                  *info = -2;
24   else  if(*n<0)                                                  *info = -3;
25   else  if(*lda<std::max(1,*m))                                   *info = -5;
26   else  if(*lda<std::max(1,*m))                                   *info = -8;
27   else  if(*ldu <1 || (*jobz=='A' && *ldu <*m)
28                    || (*jobz=='O' && *m<*n && *ldu<*m))           *info = -8;
29   else  if(*ldvt<1 || (*jobz=='A' && *ldvt<*n)
30                    || (*jobz=='S' && *ldvt<diag_size)
31                    || (*jobz=='O' && *m>=*n && *ldvt<*n))         *info = -10;
32 
33   if(*info!=0)
34   {
35     int e = -*info;
36     return xerbla_(SCALAR_SUFFIX_UP"GESDD ", &e, 6);
37   }
38 
39   if(query_size)
40   {
41     *lwork = 0;
42     return 0;
43   }
44 
45   if(*n==0 || *m==0)
46     return 0;
47 
48   PlainMatrixType mat(*m,*n);
49   mat = matrix(a,*m,*n,*lda);
50 
51   int option = *jobz=='A' ? ComputeFullU|ComputeFullV
52              : *jobz=='S' ? ComputeThinU|ComputeThinV
53              : *jobz=='O' ? ComputeThinU|ComputeThinV
54              : 0;
55 
56   BDCSVD<PlainMatrixType> svd(mat,option);
57 
58   make_vector(s,diag_size) = svd.singularValues().head(diag_size);
59 
60   if(*jobz=='A')
61   {
62     matrix(u,*m,*m,*ldu)   = svd.matrixU();
63     matrix(vt,*n,*n,*ldvt) = svd.matrixV().adjoint();
64   }
65   else if(*jobz=='S')
66   {
67     matrix(u,*m,diag_size,*ldu)   = svd.matrixU();
68     matrix(vt,diag_size,*n,*ldvt) = svd.matrixV().adjoint();
69   }
70   else if(*jobz=='O' && *m>=*n)
71   {
72     matrix(a,*m,*n,*lda)   = svd.matrixU();
73     matrix(vt,*n,*n,*ldvt) = svd.matrixV().adjoint();
74   }
75   else if(*jobz=='O')
76   {
77     matrix(u,*m,*m,*ldu)        = svd.matrixU();
78     matrix(a,diag_size,*n,*lda) = svd.matrixV().adjoint();
79   }
80 
81   return 0;
82 }
83 
84 // computes the singular values/vectors a general M-by-N matrix A using two sided jacobi algorithm
85 EIGEN_LAPACK_FUNC(gesvd,(char *jobu, char *jobv, int *m, int* n, Scalar* a, int *lda, RealScalar *s, Scalar *u, int *ldu, Scalar *vt, int *ldvt, Scalar* /*work*/, int* lwork,
86                          EIGEN_LAPACK_ARG_IF_COMPLEX(RealScalar */*rwork*/) int *info))
87 {
88   // TODO exploit the work buffer
89   bool query_size = *lwork==-1;
90   int diag_size = (std::min)(*m,*n);
91 
92   *info = 0;
93         if( *jobu!='A' && *jobu!='S' && *jobu!='O' && *jobu!='N') *info = -1;
94   else  if((*jobv!='A' && *jobv!='S' && *jobv!='O' && *jobv!='N')
95            || (*jobu=='O' && *jobv=='O'))                         *info = -2;
96   else  if(*m<0)                                                  *info = -3;
97   else  if(*n<0)                                                  *info = -4;
98   else  if(*lda<std::max(1,*m))                                   *info = -6;
99   else  if(*ldu <1 || ((*jobu=='A' || *jobu=='S') && *ldu<*m))    *info = -9;
100   else  if(*ldvt<1 || (*jobv=='A' && *ldvt<*n)
101                    || (*jobv=='S' && *ldvt<diag_size))            *info = -11;
102 
103   if(*info!=0)
104   {
105     int e = -*info;
106     return xerbla_(SCALAR_SUFFIX_UP"GESVD ", &e, 6);
107   }
108 
109   if(query_size)
110   {
111     *lwork = 0;
112     return 0;
113   }
114 
115   if(*n==0 || *m==0)
116     return 0;
117 
118   PlainMatrixType mat(*m,*n);
119   mat = matrix(a,*m,*n,*lda);
120 
121   int option = (*jobu=='A' ? ComputeFullU : *jobu=='S' || *jobu=='O' ? ComputeThinU : 0)
122              | (*jobv=='A' ? ComputeFullV : *jobv=='S' || *jobv=='O' ? ComputeThinV : 0);
123 
124   JacobiSVD<PlainMatrixType> svd(mat,option);
125 
126   make_vector(s,diag_size) = svd.singularValues().head(diag_size);
127   {
128         if(*jobu=='A') matrix(u,*m,*m,*ldu)           = svd.matrixU();
129   else  if(*jobu=='S') matrix(u,*m,diag_size,*ldu)    = svd.matrixU();
130   else  if(*jobu=='O') matrix(a,*m,diag_size,*lda)    = svd.matrixU();
131   }
132   {
133         if(*jobv=='A') matrix(vt,*n,*n,*ldvt)         = svd.matrixV().adjoint();
134   else  if(*jobv=='S') matrix(vt,diag_size,*n,*ldvt)  = svd.matrixV().adjoint();
135   else  if(*jobv=='O') matrix(a,diag_size,*n,*lda)    = svd.matrixV().adjoint();
136   }
137   return 0;
138 }
139