1// This file is part of Eigen, a lightweight C++ template library
2// for linear algebra.
3//
4// Copyright (C) 2008 Gael Guennebaud <gael.guennebaud@inria.fr>
5// Copyright (C) 2006-2008 Benoit Jacob <jacob.benoit.1@gmail.com>
6//
7// This Source Code Form is subject to the terms of the Mozilla
8// Public License v. 2.0. If a copy of the MPL was not distributed
9// with this file, You can obtain one at http://mozilla.org/MPL/2.0/.
10
11// discard stack allocation as that too bypasses malloc
12#define EIGEN_STACK_ALLOCATION_LIMIT 0
13// heap allocation will raise an assert if enabled at runtime
14#define EIGEN_RUNTIME_NO_MALLOC
15
16#include "main.h"
17#include <Eigen/Cholesky>
18#include <Eigen/Eigenvalues>
19#include <Eigen/LU>
20#include <Eigen/QR>
21#include <Eigen/SVD>
22
23template<typename MatrixType> void nomalloc(const MatrixType& m)
24{
25  /* this test check no dynamic memory allocation are issued with fixed-size matrices
26  */
27  typedef typename MatrixType::Index Index;
28  typedef typename MatrixType::Scalar Scalar;
29
30  Index rows = m.rows();
31  Index cols = m.cols();
32
33  MatrixType m1 = MatrixType::Random(rows, cols),
34             m2 = MatrixType::Random(rows, cols),
35             m3(rows, cols);
36
37  Scalar s1 = internal::random<Scalar>();
38
39  Index r = internal::random<Index>(0, rows-1),
40        c = internal::random<Index>(0, cols-1);
41
42  VERIFY_IS_APPROX((m1+m2)*s1,              s1*m1+s1*m2);
43  VERIFY_IS_APPROX((m1+m2)(r,c), (m1(r,c))+(m2(r,c)));
44  VERIFY_IS_APPROX(m1.cwiseProduct(m1.block(0,0,rows,cols)), (m1.array()*m1.array()).matrix());
45  VERIFY_IS_APPROX((m1*m1.transpose())*m2,  m1*(m1.transpose()*m2));
46
47  m2.col(0).noalias() = m1 * m1.col(0);
48  m2.col(0).noalias() -= m1.adjoint() * m1.col(0);
49  m2.col(0).noalias() -= m1 * m1.row(0).adjoint();
50  m2.col(0).noalias() -= m1.adjoint() * m1.row(0).adjoint();
51
52  m2.row(0).noalias() = m1.row(0) * m1;
53  m2.row(0).noalias() -= m1.row(0) * m1.adjoint();
54  m2.row(0).noalias() -= m1.col(0).adjoint() * m1;
55  m2.row(0).noalias() -= m1.col(0).adjoint() * m1.adjoint();
56  VERIFY_IS_APPROX(m2,m2);
57
58  m2.col(0).noalias() = m1.template triangularView<Upper>() * m1.col(0);
59  m2.col(0).noalias() -= m1.adjoint().template triangularView<Upper>() * m1.col(0);
60  m2.col(0).noalias() -= m1.template triangularView<Upper>() * m1.row(0).adjoint();
61  m2.col(0).noalias() -= m1.adjoint().template triangularView<Upper>() * m1.row(0).adjoint();
62
63  m2.row(0).noalias() = m1.row(0) * m1.template triangularView<Upper>();
64  m2.row(0).noalias() -= m1.row(0) * m1.adjoint().template triangularView<Upper>();
65  m2.row(0).noalias() -= m1.col(0).adjoint() * m1.template triangularView<Upper>();
66  m2.row(0).noalias() -= m1.col(0).adjoint() * m1.adjoint().template triangularView<Upper>();
67  VERIFY_IS_APPROX(m2,m2);
68
69  m2.col(0).noalias() = m1.template selfadjointView<Upper>() * m1.col(0);
70  m2.col(0).noalias() -= m1.adjoint().template selfadjointView<Upper>() * m1.col(0);
71  m2.col(0).noalias() -= m1.template selfadjointView<Upper>() * m1.row(0).adjoint();
72  m2.col(0).noalias() -= m1.adjoint().template selfadjointView<Upper>() * m1.row(0).adjoint();
73
74  m2.row(0).noalias() = m1.row(0) * m1.template selfadjointView<Upper>();
75  m2.row(0).noalias() -= m1.row(0) * m1.adjoint().template selfadjointView<Upper>();
76  m2.row(0).noalias() -= m1.col(0).adjoint() * m1.template selfadjointView<Upper>();
77  m2.row(0).noalias() -= m1.col(0).adjoint() * m1.adjoint().template selfadjointView<Upper>();
78  VERIFY_IS_APPROX(m2,m2);
79
80  m2.template selfadjointView<Lower>().rankUpdate(m1.col(0),-1);
81  m2.template selfadjointView<Upper>().rankUpdate(m1.row(0),-1);
82  m2.template selfadjointView<Lower>().rankUpdate(m1.col(0), m1.col(0)); // rank-2
83
84  // The following fancy matrix-matrix products are not safe yet regarding static allocation
85  m2.template selfadjointView<Lower>().rankUpdate(m1);
86  m2 += m2.template triangularView<Upper>() * m1;
87  m2.template triangularView<Upper>() = m2 * m2;
88  m1 += m1.template selfadjointView<Lower>() * m2;
89  VERIFY_IS_APPROX(m2,m2);
90}
91
92template<typename Scalar>
93void ctms_decompositions()
94{
95  const int maxSize = 16;
96  const int size    = 12;
97
98  typedef Eigen::Matrix<Scalar,
99                        Eigen::Dynamic, Eigen::Dynamic,
100                        0,
101                        maxSize, maxSize> Matrix;
102
103  typedef Eigen::Matrix<Scalar,
104                        Eigen::Dynamic, 1,
105                        0,
106                        maxSize, 1> Vector;
107
108  typedef Eigen::Matrix<std::complex<Scalar>,
109                        Eigen::Dynamic, Eigen::Dynamic,
110                        0,
111                        maxSize, maxSize> ComplexMatrix;
112
113  const Matrix A(Matrix::Random(size, size)), B(Matrix::Random(size, size));
114  Matrix X(size,size);
115  const ComplexMatrix complexA(ComplexMatrix::Random(size, size));
116  const Matrix saA = A.adjoint() * A;
117  const Vector b(Vector::Random(size));
118  Vector x(size);
119
120  // Cholesky module
121  Eigen::LLT<Matrix>  LLT;  LLT.compute(A);
122  X = LLT.solve(B);
123  x = LLT.solve(b);
124  Eigen::LDLT<Matrix> LDLT; LDLT.compute(A);
125  X = LDLT.solve(B);
126  x = LDLT.solve(b);
127
128  // Eigenvalues module
129  Eigen::HessenbergDecomposition<ComplexMatrix> hessDecomp;        hessDecomp.compute(complexA);
130  Eigen::ComplexSchur<ComplexMatrix>            cSchur(size);      cSchur.compute(complexA);
131  Eigen::ComplexEigenSolver<ComplexMatrix>      cEigSolver;        cEigSolver.compute(complexA);
132  Eigen::EigenSolver<Matrix>                    eigSolver;         eigSolver.compute(A);
133  Eigen::SelfAdjointEigenSolver<Matrix>         saEigSolver(size); saEigSolver.compute(saA);
134  Eigen::Tridiagonalization<Matrix>             tridiag;           tridiag.compute(saA);
135
136  // LU module
137  Eigen::PartialPivLU<Matrix> ppLU; ppLU.compute(A);
138  X = ppLU.solve(B);
139  x = ppLU.solve(b);
140  Eigen::FullPivLU<Matrix>    fpLU; fpLU.compute(A);
141  X = fpLU.solve(B);
142  x = fpLU.solve(b);
143
144  // QR module
145  Eigen::HouseholderQR<Matrix>        hQR;  hQR.compute(A);
146  X = hQR.solve(B);
147  x = hQR.solve(b);
148  Eigen::ColPivHouseholderQR<Matrix>  cpQR; cpQR.compute(A);
149  X = cpQR.solve(B);
150  x = cpQR.solve(b);
151  Eigen::FullPivHouseholderQR<Matrix> fpQR; fpQR.compute(A);
152  // FIXME X = fpQR.solve(B);
153  x = fpQR.solve(b);
154
155  // SVD module
156  Eigen::JacobiSVD<Matrix> jSVD; jSVD.compute(A, ComputeFullU | ComputeFullV);
157}
158
159void test_zerosized() {
160  // default constructors:
161  Eigen::MatrixXd A;
162  Eigen::VectorXd v;
163  // explicit zero-sized:
164  Eigen::ArrayXXd A0(0,0);
165  Eigen::ArrayXd v0(0);
166
167  // assigning empty objects to each other:
168  A=A0;
169  v=v0;
170}
171
172template<typename MatrixType> void test_reference(const MatrixType& m) {
173  typedef typename MatrixType::Scalar Scalar;
174  enum { Flag          =  MatrixType::IsRowMajor ? Eigen::RowMajor : Eigen::ColMajor};
175  enum { TransposeFlag = !MatrixType::IsRowMajor ? Eigen::RowMajor : Eigen::ColMajor};
176  typename MatrixType::Index rows = m.rows(), cols=m.cols();
177  typedef Eigen::Matrix<Scalar, Eigen::Dynamic, Eigen::Dynamic, Flag         > MatrixX;
178  typedef Eigen::Matrix<Scalar, Eigen::Dynamic, Eigen::Dynamic, TransposeFlag> MatrixXT;
179  // Dynamic reference:
180  typedef Eigen::Ref<const MatrixX  > Ref;
181  typedef Eigen::Ref<const MatrixXT > RefT;
182
183  Ref r1(m);
184  Ref r2(m.block(rows/3, cols/4, rows/2, cols/2));
185  RefT r3(m.transpose());
186  RefT r4(m.topLeftCorner(rows/2, cols/2).transpose());
187
188  VERIFY_RAISES_ASSERT(RefT r5(m));
189  VERIFY_RAISES_ASSERT(Ref r6(m.transpose()));
190  VERIFY_RAISES_ASSERT(Ref r7(Scalar(2) * m));
191
192  // Copy constructors shall also never malloc
193  Ref r8 = r1;
194  RefT r9 = r3;
195
196  // Initializing from a compatible Ref shall also never malloc
197  Eigen::Ref<const MatrixX, Unaligned, Stride<Dynamic, Dynamic> > r10=r8, r11=m;
198
199  // Initializing from an incompatible Ref will malloc:
200  typedef Eigen::Ref<const MatrixX, Aligned> RefAligned;
201  VERIFY_RAISES_ASSERT(RefAligned r12=r10);
202  VERIFY_RAISES_ASSERT(Ref r13=r10); // r10 has more dynamic strides
203
204}
205
206void test_nomalloc()
207{
208  // create some dynamic objects
209  Eigen::MatrixXd M1 = MatrixXd::Random(3,3);
210  Ref<const MatrixXd> R1 = 2.0*M1; // Ref requires temporary
211
212  // from here on prohibit malloc:
213  Eigen::internal::set_is_malloc_allowed(false);
214
215  // check that our operator new is indeed called:
216  VERIFY_RAISES_ASSERT(MatrixXd dummy(MatrixXd::Random(3,3)));
217  CALL_SUBTEST_1(nomalloc(Matrix<float, 1, 1>()) );
218  CALL_SUBTEST_2(nomalloc(Matrix4d()) );
219  CALL_SUBTEST_3(nomalloc(Matrix<float,32,32>()) );
220
221  // Check decomposition modules with dynamic matrices that have a known compile-time max size (ctms)
222  CALL_SUBTEST_4(ctms_decompositions<float>());
223
224  CALL_SUBTEST_5(test_zerosized());
225
226  CALL_SUBTEST_6(test_reference(Matrix<float,32,32>()));
227  CALL_SUBTEST_7(test_reference(R1));
228  CALL_SUBTEST_8(Ref<MatrixXd> R2 = M1.topRows<2>(); test_reference(R2));
229}
230