1 // This file is part of Eigen, a lightweight C++ template library 2 // for linear algebra. 3 // 4 // Copyright (C) 2011 Benoit Jacob <jacob.benoit.1 (at) gmail.com> 5 // Copyright (C) 2015 Gael Guennebaud <gael.guennebaud (at) inria.fr> 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 #define TEST_ENABLE_TEMPORARY_TRACKING 12 #define EIGEN_NO_STATIC_ASSERT 13 14 #include "main.h" 15 16 template<typename ArrayType> void vectorwiseop_array(const ArrayType& m) 17 { 18 typedef typename ArrayType::Index Index; 19 typedef typename ArrayType::Scalar Scalar; 20 typedef Array<Scalar, ArrayType::RowsAtCompileTime, 1> ColVectorType; 21 typedef Array<Scalar, 1, ArrayType::ColsAtCompileTime> RowVectorType; 22 23 Index rows = m.rows(); 24 Index cols = m.cols(); 25 Index r = internal::random<Index>(0, rows-1), 26 c = internal::random<Index>(0, cols-1); 27 28 ArrayType m1 = ArrayType::Random(rows, cols), 29 m2(rows, cols), 30 m3(rows, cols); 31 32 ColVectorType colvec = ColVectorType::Random(rows); 33 RowVectorType rowvec = RowVectorType::Random(cols); 34 35 // test addition 36 37 m2 = m1; 38 m2.colwise() += colvec; 39 VERIFY_IS_APPROX(m2, m1.colwise() + colvec); 40 VERIFY_IS_APPROX(m2.col(c), m1.col(c) + colvec); 41 42 VERIFY_RAISES_ASSERT(m2.colwise() += colvec.transpose()); 43 VERIFY_RAISES_ASSERT(m1.colwise() + colvec.transpose()); 44 45 m2 = m1; 46 m2.rowwise() += rowvec; 47 VERIFY_IS_APPROX(m2, m1.rowwise() + rowvec); 48 VERIFY_IS_APPROX(m2.row(r), m1.row(r) + rowvec); 49 50 VERIFY_RAISES_ASSERT(m2.rowwise() += rowvec.transpose()); 51 VERIFY_RAISES_ASSERT(m1.rowwise() + rowvec.transpose()); 52 53 // test substraction 54 55 m2 = m1; 56 m2.colwise() -= colvec; 57 VERIFY_IS_APPROX(m2, m1.colwise() - colvec); 58 VERIFY_IS_APPROX(m2.col(c), m1.col(c) - colvec); 59 60 VERIFY_RAISES_ASSERT(m2.colwise() -= colvec.transpose()); 61 VERIFY_RAISES_ASSERT(m1.colwise() - colvec.transpose()); 62 63 m2 = m1; 64 m2.rowwise() -= rowvec; 65 VERIFY_IS_APPROX(m2, m1.rowwise() - rowvec); 66 VERIFY_IS_APPROX(m2.row(r), m1.row(r) - rowvec); 67 68 VERIFY_RAISES_ASSERT(m2.rowwise() -= rowvec.transpose()); 69 VERIFY_RAISES_ASSERT(m1.rowwise() - rowvec.transpose()); 70 71 // test multiplication 72 73 m2 = m1; 74 m2.colwise() *= colvec; 75 VERIFY_IS_APPROX(m2, m1.colwise() * colvec); 76 VERIFY_IS_APPROX(m2.col(c), m1.col(c) * colvec); 77 78 VERIFY_RAISES_ASSERT(m2.colwise() *= colvec.transpose()); 79 VERIFY_RAISES_ASSERT(m1.colwise() * colvec.transpose()); 80 81 m2 = m1; 82 m2.rowwise() *= rowvec; 83 VERIFY_IS_APPROX(m2, m1.rowwise() * rowvec); 84 VERIFY_IS_APPROX(m2.row(r), m1.row(r) * rowvec); 85 86 VERIFY_RAISES_ASSERT(m2.rowwise() *= rowvec.transpose()); 87 VERIFY_RAISES_ASSERT(m1.rowwise() * rowvec.transpose()); 88 89 // test quotient 90 91 m2 = m1; 92 m2.colwise() /= colvec; 93 VERIFY_IS_APPROX(m2, m1.colwise() / colvec); 94 VERIFY_IS_APPROX(m2.col(c), m1.col(c) / colvec); 95 96 VERIFY_RAISES_ASSERT(m2.colwise() /= colvec.transpose()); 97 VERIFY_RAISES_ASSERT(m1.colwise() / colvec.transpose()); 98 99 m2 = m1; 100 m2.rowwise() /= rowvec; 101 VERIFY_IS_APPROX(m2, m1.rowwise() / rowvec); 102 VERIFY_IS_APPROX(m2.row(r), m1.row(r) / rowvec); 103 104 VERIFY_RAISES_ASSERT(m2.rowwise() /= rowvec.transpose()); 105 VERIFY_RAISES_ASSERT(m1.rowwise() / rowvec.transpose()); 106 107 m2 = m1; 108 // yes, there might be an aliasing issue there but ".rowwise() /=" 109 // is supposed to evaluate " m2.colwise().sum()" into a temporary to avoid 110 // evaluating the reduction multiple times 111 if(ArrayType::RowsAtCompileTime>2 || ArrayType::RowsAtCompileTime==Dynamic) 112 { 113 m2.rowwise() /= m2.colwise().sum(); 114 VERIFY_IS_APPROX(m2, m1.rowwise() / m1.colwise().sum()); 115 } 116 117 // all/any 118 Array<bool,Dynamic,Dynamic> mb(rows,cols); 119 mb = (m1.real()<=0.7).colwise().all(); 120 VERIFY( (mb.col(c) == (m1.real().col(c)<=0.7).all()).all() ); 121 mb = (m1.real()<=0.7).rowwise().all(); 122 VERIFY( (mb.row(r) == (m1.real().row(r)<=0.7).all()).all() ); 123 124 mb = (m1.real()>=0.7).colwise().any(); 125 VERIFY( (mb.col(c) == (m1.real().col(c)>=0.7).any()).all() ); 126 mb = (m1.real()>=0.7).rowwise().any(); 127 VERIFY( (mb.row(r) == (m1.real().row(r)>=0.7).any()).all() ); 128 } 129 130 template<typename MatrixType> void vectorwiseop_matrix(const MatrixType& m) 131 { 132 typedef typename MatrixType::Index Index; 133 typedef typename MatrixType::Scalar Scalar; 134 typedef typename NumTraits<Scalar>::Real RealScalar; 135 typedef Matrix<Scalar, MatrixType::RowsAtCompileTime, 1> ColVectorType; 136 typedef Matrix<Scalar, 1, MatrixType::ColsAtCompileTime> RowVectorType; 137 typedef Matrix<RealScalar, MatrixType::RowsAtCompileTime, 1> RealColVectorType; 138 typedef Matrix<RealScalar, 1, MatrixType::ColsAtCompileTime> RealRowVectorType; 139 140 Index rows = m.rows(); 141 Index cols = m.cols(); 142 Index r = internal::random<Index>(0, rows-1), 143 c = internal::random<Index>(0, cols-1); 144 145 MatrixType m1 = MatrixType::Random(rows, cols), 146 m2(rows, cols), 147 m3(rows, cols); 148 149 ColVectorType colvec = ColVectorType::Random(rows); 150 RowVectorType rowvec = RowVectorType::Random(cols); 151 RealColVectorType rcres; 152 RealRowVectorType rrres; 153 154 // test addition 155 156 m2 = m1; 157 m2.colwise() += colvec; 158 VERIFY_IS_APPROX(m2, m1.colwise() + colvec); 159 VERIFY_IS_APPROX(m2.col(c), m1.col(c) + colvec); 160 161 if(rows>1) 162 { 163 VERIFY_RAISES_ASSERT(m2.colwise() += colvec.transpose()); 164 VERIFY_RAISES_ASSERT(m1.colwise() + colvec.transpose()); 165 } 166 167 m2 = m1; 168 m2.rowwise() += rowvec; 169 VERIFY_IS_APPROX(m2, m1.rowwise() + rowvec); 170 VERIFY_IS_APPROX(m2.row(r), m1.row(r) + rowvec); 171 172 if(cols>1) 173 { 174 VERIFY_RAISES_ASSERT(m2.rowwise() += rowvec.transpose()); 175 VERIFY_RAISES_ASSERT(m1.rowwise() + rowvec.transpose()); 176 } 177 178 // test substraction 179 180 m2 = m1; 181 m2.colwise() -= colvec; 182 VERIFY_IS_APPROX(m2, m1.colwise() - colvec); 183 VERIFY_IS_APPROX(m2.col(c), m1.col(c) - colvec); 184 185 if(rows>1) 186 { 187 VERIFY_RAISES_ASSERT(m2.colwise() -= colvec.transpose()); 188 VERIFY_RAISES_ASSERT(m1.colwise() - colvec.transpose()); 189 } 190 191 m2 = m1; 192 m2.rowwise() -= rowvec; 193 VERIFY_IS_APPROX(m2, m1.rowwise() - rowvec); 194 VERIFY_IS_APPROX(m2.row(r), m1.row(r) - rowvec); 195 196 if(cols>1) 197 { 198 VERIFY_RAISES_ASSERT(m2.rowwise() -= rowvec.transpose()); 199 VERIFY_RAISES_ASSERT(m1.rowwise() - rowvec.transpose()); 200 } 201 202 // test norm 203 rrres = m1.colwise().norm(); 204 VERIFY_IS_APPROX(rrres(c), m1.col(c).norm()); 205 rcres = m1.rowwise().norm(); 206 VERIFY_IS_APPROX(rcres(r), m1.row(r).norm()); 207 208 VERIFY_IS_APPROX(m1.cwiseAbs().colwise().sum(), m1.colwise().template lpNorm<1>()); 209 VERIFY_IS_APPROX(m1.cwiseAbs().rowwise().sum(), m1.rowwise().template lpNorm<1>()); 210 VERIFY_IS_APPROX(m1.cwiseAbs().colwise().maxCoeff(), m1.colwise().template lpNorm<Infinity>()); 211 VERIFY_IS_APPROX(m1.cwiseAbs().rowwise().maxCoeff(), m1.rowwise().template lpNorm<Infinity>()); 212 213 // regression for bug 1158 214 VERIFY_IS_APPROX(m1.cwiseAbs().colwise().sum().x(), m1.col(0).cwiseAbs().sum()); 215 216 // test normalized 217 m2 = m1.colwise().normalized(); 218 VERIFY_IS_APPROX(m2.col(c), m1.col(c).normalized()); 219 m2 = m1.rowwise().normalized(); 220 VERIFY_IS_APPROX(m2.row(r), m1.row(r).normalized()); 221 222 // test normalize 223 m2 = m1; 224 m2.colwise().normalize(); 225 VERIFY_IS_APPROX(m2.col(c), m1.col(c).normalized()); 226 m2 = m1; 227 m2.rowwise().normalize(); 228 VERIFY_IS_APPROX(m2.row(r), m1.row(r).normalized()); 229 230 // test with partial reduction of products 231 Matrix<Scalar,MatrixType::RowsAtCompileTime,MatrixType::RowsAtCompileTime> m1m1 = m1 * m1.transpose(); 232 VERIFY_IS_APPROX( (m1 * m1.transpose()).colwise().sum(), m1m1.colwise().sum()); 233 Matrix<Scalar,1,MatrixType::RowsAtCompileTime> tmp(rows); 234 VERIFY_EVALUATION_COUNT( tmp = (m1 * m1.transpose()).colwise().sum(), 1); 235 236 m2 = m1.rowwise() - (m1.colwise().sum()/RealScalar(m1.rows())).eval(); 237 m1 = m1.rowwise() - (m1.colwise().sum()/RealScalar(m1.rows())); 238 VERIFY_IS_APPROX( m1, m2 ); 239 VERIFY_EVALUATION_COUNT( m2 = (m1.rowwise() - m1.colwise().sum()/RealScalar(m1.rows())), (MatrixType::RowsAtCompileTime!=1 ? 1 : 0) ); 240 } 241 242 void test_vectorwiseop() 243 { 244 CALL_SUBTEST_1( vectorwiseop_array(Array22cd()) ); 245 CALL_SUBTEST_2( vectorwiseop_array(Array<double, 3, 2>()) ); 246 CALL_SUBTEST_3( vectorwiseop_array(ArrayXXf(3, 4)) ); 247 CALL_SUBTEST_4( vectorwiseop_matrix(Matrix4cf()) ); 248 CALL_SUBTEST_5( vectorwiseop_matrix(Matrix<float,4,5>()) ); 249 CALL_SUBTEST_6( vectorwiseop_matrix(MatrixXd(internal::random<int>(1,EIGEN_TEST_MAX_SIZE), internal::random<int>(1,EIGEN_TEST_MAX_SIZE))) ); 250 CALL_SUBTEST_7( vectorwiseop_matrix(VectorXd(internal::random<int>(1,EIGEN_TEST_MAX_SIZE))) ); 251 CALL_SUBTEST_7( vectorwiseop_matrix(RowVectorXd(internal::random<int>(1,EIGEN_TEST_MAX_SIZE))) ); 252 } 253
