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 (at) inria.fr> 5 // Copyright (C) 2006-2008 Benoit Jacob <jacob.benoit.1 (at) 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 // this hack is needed to make this file compiles with -pedantic (gcc) 12 #ifdef __GNUC__ 13 #define throw(X) 14 #endif 15 16 #ifdef __INTEL_COMPILER 17 // disable "warning #76: argument to macro is empty" produced by the above hack 18 #pragma warning disable 76 19 #endif 20 21 // discard stack allocation as that too bypasses malloc 22 #define EIGEN_STACK_ALLOCATION_LIMIT 0 23 // any heap allocation will raise an assert 24 #define EIGEN_NO_MALLOC 25 26 #include "main.h" 27 #include <Eigen/Cholesky> 28 #include <Eigen/Eigenvalues> 29 #include <Eigen/LU> 30 #include <Eigen/QR> 31 #include <Eigen/SVD> 32 33 template<typename MatrixType> void nomalloc(const MatrixType& m) 34 { 35 /* this test check no dynamic memory allocation are issued with fixed-size matrices 36 */ 37 typedef typename MatrixType::Index Index; 38 typedef typename MatrixType::Scalar Scalar; 39 40 Index rows = m.rows(); 41 Index cols = m.cols(); 42 43 MatrixType m1 = MatrixType::Random(rows, cols), 44 m2 = MatrixType::Random(rows, cols), 45 m3(rows, cols); 46 47 Scalar s1 = internal::random<Scalar>(); 48 49 Index r = internal::random<Index>(0, rows-1), 50 c = internal::random<Index>(0, cols-1); 51 52 VERIFY_IS_APPROX((m1+m2)*s1, s1*m1+s1*m2); 53 VERIFY_IS_APPROX((m1+m2)(r,c), (m1(r,c))+(m2(r,c))); 54 VERIFY_IS_APPROX(m1.cwiseProduct(m1.block(0,0,rows,cols)), (m1.array()*m1.array()).matrix()); 55 VERIFY_IS_APPROX((m1*m1.transpose())*m2, m1*(m1.transpose()*m2)); 56 57 m2.col(0).noalias() = m1 * m1.col(0); 58 m2.col(0).noalias() -= m1.adjoint() * m1.col(0); 59 m2.col(0).noalias() -= m1 * m1.row(0).adjoint(); 60 m2.col(0).noalias() -= m1.adjoint() * m1.row(0).adjoint(); 61 62 m2.row(0).noalias() = m1.row(0) * m1; 63 m2.row(0).noalias() -= m1.row(0) * m1.adjoint(); 64 m2.row(0).noalias() -= m1.col(0).adjoint() * m1; 65 m2.row(0).noalias() -= m1.col(0).adjoint() * m1.adjoint(); 66 VERIFY_IS_APPROX(m2,m2); 67 68 m2.col(0).noalias() = m1.template triangularView<Upper>() * m1.col(0); 69 m2.col(0).noalias() -= m1.adjoint().template triangularView<Upper>() * m1.col(0); 70 m2.col(0).noalias() -= m1.template triangularView<Upper>() * m1.row(0).adjoint(); 71 m2.col(0).noalias() -= m1.adjoint().template triangularView<Upper>() * m1.row(0).adjoint(); 72 73 m2.row(0).noalias() = m1.row(0) * m1.template triangularView<Upper>(); 74 m2.row(0).noalias() -= m1.row(0) * m1.adjoint().template triangularView<Upper>(); 75 m2.row(0).noalias() -= m1.col(0).adjoint() * m1.template triangularView<Upper>(); 76 m2.row(0).noalias() -= m1.col(0).adjoint() * m1.adjoint().template triangularView<Upper>(); 77 VERIFY_IS_APPROX(m2,m2); 78 79 m2.col(0).noalias() = m1.template selfadjointView<Upper>() * m1.col(0); 80 m2.col(0).noalias() -= m1.adjoint().template selfadjointView<Upper>() * m1.col(0); 81 m2.col(0).noalias() -= m1.template selfadjointView<Upper>() * m1.row(0).adjoint(); 82 m2.col(0).noalias() -= m1.adjoint().template selfadjointView<Upper>() * m1.row(0).adjoint(); 83 84 m2.row(0).noalias() = m1.row(0) * m1.template selfadjointView<Upper>(); 85 m2.row(0).noalias() -= m1.row(0) * m1.adjoint().template selfadjointView<Upper>(); 86 m2.row(0).noalias() -= m1.col(0).adjoint() * m1.template selfadjointView<Upper>(); 87 m2.row(0).noalias() -= m1.col(0).adjoint() * m1.adjoint().template selfadjointView<Upper>(); 88 VERIFY_IS_APPROX(m2,m2); 89 90 m2.template selfadjointView<Lower>().rankUpdate(m1.col(0),-1); 91 m2.template selfadjointView<Lower>().rankUpdate(m1.row(0),-1); 92 93 // The following fancy matrix-matrix products are not safe yet regarding static allocation 94 // m1 += m1.template triangularView<Upper>() * m2.col(; 95 // m1.template selfadjointView<Lower>().rankUpdate(m2); 96 // m1 += m1.template triangularView<Upper>() * m2; 97 // m1 += m1.template selfadjointView<Lower>() * m2; 98 // VERIFY_IS_APPROX(m1,m1); 99 } 100 101 template<typename Scalar> 102 void ctms_decompositions() 103 { 104 const int maxSize = 16; 105 const int size = 12; 106 107 typedef Eigen::Matrix<Scalar, 108 Eigen::Dynamic, Eigen::Dynamic, 109 0, 110 maxSize, maxSize> Matrix; 111 112 typedef Eigen::Matrix<Scalar, 113 Eigen::Dynamic, 1, 114 0, 115 maxSize, 1> Vector; 116 117 typedef Eigen::Matrix<std::complex<Scalar>, 118 Eigen::Dynamic, Eigen::Dynamic, 119 0, 120 maxSize, maxSize> ComplexMatrix; 121 122 const Matrix A(Matrix::Random(size, size)), B(Matrix::Random(size, size)); 123 Matrix X(size,size); 124 const ComplexMatrix complexA(ComplexMatrix::Random(size, size)); 125 const Matrix saA = A.adjoint() * A; 126 const Vector b(Vector::Random(size)); 127 Vector x(size); 128 129 // Cholesky module 130 Eigen::LLT<Matrix> LLT; LLT.compute(A); 131 X = LLT.solve(B); 132 x = LLT.solve(b); 133 Eigen::LDLT<Matrix> LDLT; LDLT.compute(A); 134 X = LDLT.solve(B); 135 x = LDLT.solve(b); 136 137 // Eigenvalues module 138 Eigen::HessenbergDecomposition<ComplexMatrix> hessDecomp; hessDecomp.compute(complexA); 139 Eigen::ComplexSchur<ComplexMatrix> cSchur(size); cSchur.compute(complexA); 140 Eigen::ComplexEigenSolver<ComplexMatrix> cEigSolver; cEigSolver.compute(complexA); 141 Eigen::EigenSolver<Matrix> eigSolver; eigSolver.compute(A); 142 Eigen::SelfAdjointEigenSolver<Matrix> saEigSolver(size); saEigSolver.compute(saA); 143 Eigen::Tridiagonalization<Matrix> tridiag; tridiag.compute(saA); 144 145 // LU module 146 Eigen::PartialPivLU<Matrix> ppLU; ppLU.compute(A); 147 X = ppLU.solve(B); 148 x = ppLU.solve(b); 149 Eigen::FullPivLU<Matrix> fpLU; fpLU.compute(A); 150 X = fpLU.solve(B); 151 x = fpLU.solve(b); 152 153 // QR module 154 Eigen::HouseholderQR<Matrix> hQR; hQR.compute(A); 155 X = hQR.solve(B); 156 x = hQR.solve(b); 157 Eigen::ColPivHouseholderQR<Matrix> cpQR; cpQR.compute(A); 158 X = cpQR.solve(B); 159 x = cpQR.solve(b); 160 Eigen::FullPivHouseholderQR<Matrix> fpQR; fpQR.compute(A); 161 // FIXME X = fpQR.solve(B); 162 x = fpQR.solve(b); 163 164 // SVD module 165 Eigen::JacobiSVD<Matrix> jSVD; jSVD.compute(A, ComputeFullU | ComputeFullV); 166 } 167 168 void test_zerosized() { 169 // default constructors: 170 Eigen::MatrixXd A; 171 Eigen::VectorXd v; 172 // explicit zero-sized: 173 Eigen::ArrayXXd A0(0,0); 174 Eigen::ArrayXd v0(std::ptrdiff_t(0)); // FIXME ArrayXd(0) is ambiguous 175 176 // assigning empty objects to each other: 177 A=A0; 178 v=v0; 179 } 180 181 template<typename MatrixType> void test_reference(const MatrixType& m) { 182 typedef typename MatrixType::Scalar Scalar; 183 enum { Flag = MatrixType::IsRowMajor ? Eigen::RowMajor : Eigen::ColMajor}; 184 enum { TransposeFlag = !MatrixType::IsRowMajor ? Eigen::RowMajor : Eigen::ColMajor}; 185 typename MatrixType::Index rows = m.rows(), cols=m.cols(); 186 // Dynamic reference: 187 typedef Eigen::Ref<const Eigen::Matrix<Scalar, Eigen::Dynamic, Eigen::Dynamic, Flag > > Ref; 188 typedef Eigen::Ref<const Eigen::Matrix<Scalar, Eigen::Dynamic, Eigen::Dynamic, TransposeFlag> > RefT; 189 190 Ref r1(m); 191 Ref r2(m.block(rows/3, cols/4, rows/2, cols/2)); 192 RefT r3(m.transpose()); 193 RefT r4(m.topLeftCorner(rows/2, cols/2).transpose()); 194 195 VERIFY_RAISES_ASSERT(RefT r5(m)); 196 VERIFY_RAISES_ASSERT(Ref r6(m.transpose())); 197 VERIFY_RAISES_ASSERT(Ref r7(Scalar(2) * m)); 198 } 199 200 void test_nomalloc() 201 { 202 // check that our operator new is indeed called: 203 VERIFY_RAISES_ASSERT(MatrixXd dummy(MatrixXd::Random(3,3))); 204 CALL_SUBTEST_1(nomalloc(Matrix<float, 1, 1>()) ); 205 CALL_SUBTEST_2(nomalloc(Matrix4d()) ); 206 CALL_SUBTEST_3(nomalloc(Matrix<float,32,32>()) ); 207 208 // Check decomposition modules with dynamic matrices that have a known compile-time max size (ctms) 209 CALL_SUBTEST_4(ctms_decompositions<float>()); 210 CALL_SUBTEST_5(test_zerosized()); 211 CALL_SUBTEST_6(test_reference(Matrix<float,32,32>())); 212 } 213