xref: /aosp_15_r20/external/eigen/test/householder.cpp (revision bf2c37156dfe67e5dfebd6d394bad8b2ab5804d4)
1 // This file is part of Eigen, a lightweight C++ template library
2 // for linear algebra.
3 //
4 // Copyright (C) 2009-2010 Benoit Jacob <[email protected]>
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 "main.h"
11 #include <Eigen/QR>
12 
householder(const MatrixType & m)13 template<typename MatrixType> void householder(const MatrixType& m)
14 {
15   static bool even = true;
16   even = !even;
17   /* this test covers the following files:
18      Householder.h
19   */
20   Index rows = m.rows();
21   Index cols = m.cols();
22 
23   typedef typename MatrixType::Scalar Scalar;
24   typedef typename NumTraits<Scalar>::Real RealScalar;
25   typedef Matrix<Scalar, MatrixType::RowsAtCompileTime, 1> VectorType;
26   typedef Matrix<Scalar, internal::decrement_size<MatrixType::RowsAtCompileTime>::ret, 1> EssentialVectorType;
27   typedef Matrix<Scalar, MatrixType::RowsAtCompileTime, MatrixType::RowsAtCompileTime> SquareMatrixType;
28   typedef Matrix<Scalar, Dynamic, MatrixType::ColsAtCompileTime> HBlockMatrixType;
29   typedef Matrix<Scalar, Dynamic, 1> HCoeffsVectorType;
30 
31   typedef Matrix<Scalar, MatrixType::ColsAtCompileTime, MatrixType::RowsAtCompileTime> TMatrixType;
32 
33   Matrix<Scalar, EIGEN_SIZE_MAX(MatrixType::RowsAtCompileTime,MatrixType::ColsAtCompileTime), 1> _tmp((std::max)(rows,cols));
34   Scalar* tmp = &_tmp.coeffRef(0,0);
35 
36   Scalar beta;
37   RealScalar alpha;
38   EssentialVectorType essential;
39 
40   VectorType v1 = VectorType::Random(rows), v2;
41   v2 = v1;
42   v1.makeHouseholder(essential, beta, alpha);
43   v1.applyHouseholderOnTheLeft(essential,beta,tmp);
44   VERIFY_IS_APPROX(v1.norm(), v2.norm());
45   if(rows>=2) VERIFY_IS_MUCH_SMALLER_THAN(v1.tail(rows-1).norm(), v1.norm());
46   v1 = VectorType::Random(rows);
47   v2 = v1;
48   v1.applyHouseholderOnTheLeft(essential,beta,tmp);
49   VERIFY_IS_APPROX(v1.norm(), v2.norm());
50 
51   // reconstruct householder matrix:
52   SquareMatrixType id, H1, H2;
53   id.setIdentity(rows, rows);
54   H1 = H2 = id;
55   VectorType vv(rows);
56   vv << Scalar(1), essential;
57   H1.applyHouseholderOnTheLeft(essential, beta, tmp);
58   H2.applyHouseholderOnTheRight(essential, beta, tmp);
59   VERIFY_IS_APPROX(H1, H2);
60   VERIFY_IS_APPROX(H1, id - beta * vv*vv.adjoint());
61 
62   MatrixType m1(rows, cols),
63              m2(rows, cols);
64 
65   v1 = VectorType::Random(rows);
66   if(even) v1.tail(rows-1).setZero();
67   m1.colwise() = v1;
68   m2 = m1;
69   m1.col(0).makeHouseholder(essential, beta, alpha);
70   m1.applyHouseholderOnTheLeft(essential,beta,tmp);
71   VERIFY_IS_APPROX(m1.norm(), m2.norm());
72   if(rows>=2) VERIFY_IS_MUCH_SMALLER_THAN(m1.block(1,0,rows-1,cols).norm(), m1.norm());
73   VERIFY_IS_MUCH_SMALLER_THAN(numext::imag(m1(0,0)), numext::real(m1(0,0)));
74   VERIFY_IS_APPROX(numext::real(m1(0,0)), alpha);
75 
76   v1 = VectorType::Random(rows);
77   if(even) v1.tail(rows-1).setZero();
78   SquareMatrixType m3(rows,rows), m4(rows,rows);
79   m3.rowwise() = v1.transpose();
80   m4 = m3;
81   m3.row(0).makeHouseholder(essential, beta, alpha);
82   m3.applyHouseholderOnTheRight(essential.conjugate(),beta,tmp);
83   VERIFY_IS_APPROX(m3.norm(), m4.norm());
84   if(rows>=2) VERIFY_IS_MUCH_SMALLER_THAN(m3.block(0,1,rows,rows-1).norm(), m3.norm());
85   VERIFY_IS_MUCH_SMALLER_THAN(numext::imag(m3(0,0)), numext::real(m3(0,0)));
86   VERIFY_IS_APPROX(numext::real(m3(0,0)), alpha);
87 
88   // test householder sequence on the left with a shift
89 
90   Index shift = internal::random<Index>(0, std::max<Index>(rows-2,0));
91   Index brows = rows - shift;
92   m1.setRandom(rows, cols);
93   HBlockMatrixType hbm = m1.block(shift,0,brows,cols);
94   HouseholderQR<HBlockMatrixType> qr(hbm);
95   m2 = m1;
96   m2.block(shift,0,brows,cols) = qr.matrixQR();
97   HCoeffsVectorType hc = qr.hCoeffs().conjugate();
98   HouseholderSequence<MatrixType, HCoeffsVectorType> hseq(m2, hc);
99   hseq.setLength(hc.size()).setShift(shift);
100   VERIFY(hseq.length() == hc.size());
101   VERIFY(hseq.shift() == shift);
102 
103   MatrixType m5 = m2;
104   m5.block(shift,0,brows,cols).template triangularView<StrictlyLower>().setZero();
105   VERIFY_IS_APPROX(hseq * m5, m1); // test applying hseq directly
106   m3 = hseq;
107   VERIFY_IS_APPROX(m3 * m5, m1); // test evaluating hseq to a dense matrix, then applying
108 
109   SquareMatrixType hseq_mat = hseq;
110   SquareMatrixType hseq_mat_conj = hseq.conjugate();
111   SquareMatrixType hseq_mat_adj = hseq.adjoint();
112   SquareMatrixType hseq_mat_trans = hseq.transpose();
113   SquareMatrixType m6 = SquareMatrixType::Random(rows, rows);
114   VERIFY_IS_APPROX(hseq_mat.adjoint(),    hseq_mat_adj);
115   VERIFY_IS_APPROX(hseq_mat.conjugate(),  hseq_mat_conj);
116   VERIFY_IS_APPROX(hseq_mat.transpose(),  hseq_mat_trans);
117   VERIFY_IS_APPROX(hseq * m6,             hseq_mat * m6);
118   VERIFY_IS_APPROX(hseq.adjoint() * m6,   hseq_mat_adj * m6);
119   VERIFY_IS_APPROX(hseq.conjugate() * m6, hseq_mat_conj * m6);
120   VERIFY_IS_APPROX(hseq.transpose() * m6, hseq_mat_trans * m6);
121   VERIFY_IS_APPROX(m6 * hseq,             m6 * hseq_mat);
122   VERIFY_IS_APPROX(m6 * hseq.adjoint(),   m6 * hseq_mat_adj);
123   VERIFY_IS_APPROX(m6 * hseq.conjugate(), m6 * hseq_mat_conj);
124   VERIFY_IS_APPROX(m6 * hseq.transpose(), m6 * hseq_mat_trans);
125 
126   // test householder sequence on the right with a shift
127 
128   TMatrixType tm2 = m2.transpose();
129   HouseholderSequence<TMatrixType, HCoeffsVectorType, OnTheRight> rhseq(tm2, hc);
130   rhseq.setLength(hc.size()).setShift(shift);
131   VERIFY_IS_APPROX(rhseq * m5, m1); // test applying rhseq directly
132   m3 = rhseq;
133   VERIFY_IS_APPROX(m3 * m5, m1); // test evaluating rhseq to a dense matrix, then applying
134 }
135 
EIGEN_DECLARE_TEST(householder)136 EIGEN_DECLARE_TEST(householder)
137 {
138   for(int i = 0; i < g_repeat; i++) {
139     CALL_SUBTEST_1( householder(Matrix<double,2,2>()) );
140     CALL_SUBTEST_2( householder(Matrix<float,2,3>()) );
141     CALL_SUBTEST_3( householder(Matrix<double,3,5>()) );
142     CALL_SUBTEST_4( householder(Matrix<float,4,4>()) );
143     CALL_SUBTEST_5( householder(MatrixXd(internal::random<int>(1,EIGEN_TEST_MAX_SIZE),internal::random<int>(1,EIGEN_TEST_MAX_SIZE))) );
144     CALL_SUBTEST_6( householder(MatrixXcf(internal::random<int>(1,EIGEN_TEST_MAX_SIZE),internal::random<int>(1,EIGEN_TEST_MAX_SIZE))) );
145     CALL_SUBTEST_7( householder(MatrixXf(internal::random<int>(1,EIGEN_TEST_MAX_SIZE),internal::random<int>(1,EIGEN_TEST_MAX_SIZE))) );
146     CALL_SUBTEST_8( householder(Matrix<double,1,1>()) );
147   }
148 }
149