• 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) 2014-2015 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 template<typename T>
11 Array<T,4,1> four_denorms();
12 
13 template<>
four_denorms()14 Array4f four_denorms() { return Array4f(5.60844e-39f, -5.60844e-39f, 4.94e-44f, -4.94e-44f); }
15 template<>
four_denorms()16 Array4d four_denorms() { return Array4d(5.60844e-313, -5.60844e-313, 4.94e-324, -4.94e-324); }
17 template<typename T>
four_denorms()18 Array<T,4,1> four_denorms() { return four_denorms<double>().cast<T>(); }
19 
20 template<typename MatrixType>
21 void svd_fill_random(MatrixType &m, int Option = 0)
22 {
23   using std::pow;
24   typedef typename MatrixType::Scalar Scalar;
25   typedef typename MatrixType::RealScalar RealScalar;
26   typedef typename MatrixType::Index Index;
27   Index diagSize = (std::min)(m.rows(), m.cols());
28   RealScalar s = std::numeric_limits<RealScalar>::max_exponent10/4;
29   s = internal::random<RealScalar>(1,s);
30   Matrix<RealScalar,Dynamic,1> d =  Matrix<RealScalar,Dynamic,1>::Random(diagSize);
31   for(Index k=0; k<diagSize; ++k)
32     d(k) = d(k)*pow(RealScalar(10),internal::random<RealScalar>(-s,s));
33 
34   bool dup     = internal::random<int>(0,10) < 3;
35   bool unit_uv = internal::random<int>(0,10) < (dup?7:3); // if we duplicate some diagonal entries, then increase the chance to preserve them using unitary U and V factors
36 
37   // duplicate some singular values
38   if(dup)
39   {
40     Index n = internal::random<Index>(0,d.size()-1);
41     for(Index i=0; i<n; ++i)
42       d(internal::random<Index>(0,d.size()-1)) = d(internal::random<Index>(0,d.size()-1));
43   }
44 
45   Matrix<Scalar,Dynamic,Dynamic> U(m.rows(),diagSize);
46   Matrix<Scalar,Dynamic,Dynamic> VT(diagSize,m.cols());
47   if(unit_uv)
48   {
49     // in very rare cases let's try with a pure diagonal matrix
50     if(internal::random<int>(0,10) < 1)
51     {
52       U.setIdentity();
53       VT.setIdentity();
54     }
55     else
56     {
57       createRandomPIMatrixOfRank(diagSize,U.rows(), U.cols(), U);
58       createRandomPIMatrixOfRank(diagSize,VT.rows(), VT.cols(), VT);
59     }
60   }
61   else
62   {
63     U.setRandom();
64     VT.setRandom();
65   }
66 
67   Matrix<Scalar,Dynamic,1> samples(9);
68   samples << 0, four_denorms<RealScalar>(),
69             -RealScalar(1)/NumTraits<RealScalar>::highest(), RealScalar(1)/NumTraits<RealScalar>::highest(), (std::numeric_limits<RealScalar>::min)(), pow((std::numeric_limits<RealScalar>::min)(),0.8);
70 
71   if(Option==Symmetric)
72   {
73     m = U * d.asDiagonal() * U.transpose();
74 
75     // randomly nullify some rows/columns
76     {
77       Index count = internal::random<Index>(-diagSize,diagSize);
78       for(Index k=0; k<count; ++k)
79       {
80         Index i = internal::random<Index>(0,diagSize-1);
81         m.row(i).setZero();
82         m.col(i).setZero();
83       }
84       if(count<0)
85       // (partly) cancel some coeffs
86       if(!(dup && unit_uv))
87       {
88 
89         Index n = internal::random<Index>(0,m.size()-1);
90         for(Index k=0; k<n; ++k)
91         {
92           Index i = internal::random<Index>(0,m.rows()-1);
93           Index j = internal::random<Index>(0,m.cols()-1);
94           m(j,i) = m(i,j) = samples(internal::random<Index>(0,samples.size()-1));
95           if(NumTraits<Scalar>::IsComplex)
96             *(&numext::real_ref(m(j,i))+1) = *(&numext::real_ref(m(i,j))+1) = samples.real()(internal::random<Index>(0,samples.size()-1));
97         }
98       }
99     }
100   }
101   else
102   {
103     m = U * d.asDiagonal() * VT;
104     // (partly) cancel some coeffs
105     if(!(dup && unit_uv))
106     {
107       Index n = internal::random<Index>(0,m.size()-1);
108       for(Index k=0; k<n; ++k)
109       {
110         Index i = internal::random<Index>(0,m.rows()-1);
111         Index j = internal::random<Index>(0,m.cols()-1);
112         m(i,j) = samples(internal::random<Index>(0,samples.size()-1));
113         if(NumTraits<Scalar>::IsComplex)
114           *(&numext::real_ref(m(i,j))+1) = samples.real()(internal::random<Index>(0,samples.size()-1));
115       }
116     }
117   }
118 }
119 
120