1 // This file is part of Eigen, a lightweight C++ template library 2 // for linear algebra. 3 // 4 // Copyright (C) 2008 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 13 #include "main.h" 14 15 template<typename MatrixType> void matrixRedux(const MatrixType& m) 16 { 17 typedef typename MatrixType::Index Index; 18 typedef typename MatrixType::Scalar Scalar; 19 typedef typename MatrixType::RealScalar RealScalar; 20 21 Index rows = m.rows(); 22 Index cols = m.cols(); 23 24 MatrixType m1 = MatrixType::Random(rows, cols); 25 26 // The entries of m1 are uniformly distributed in [0,1], so m1.prod() is very small. This may lead to test 27 // failures if we underflow into denormals. Thus, we scale so that entries are close to 1. 28 MatrixType m1_for_prod = MatrixType::Ones(rows, cols) + RealScalar(0.2) * m1; 29 30 VERIFY_IS_MUCH_SMALLER_THAN(MatrixType::Zero(rows, cols).sum(), Scalar(1)); 31 VERIFY_IS_APPROX(MatrixType::Ones(rows, cols).sum(), Scalar(float(rows*cols))); // the float() here to shut up excessive MSVC warning about int->complex conversion being lossy 32 Scalar s(0), p(1), minc(numext::real(m1.coeff(0))), maxc(numext::real(m1.coeff(0))); 33 for(int j = 0; j < cols; j++) 34 for(int i = 0; i < rows; i++) 35 { 36 s += m1(i,j); 37 p *= m1_for_prod(i,j); 38 minc = (std::min)(numext::real(minc), numext::real(m1(i,j))); 39 maxc = (std::max)(numext::real(maxc), numext::real(m1(i,j))); 40 } 41 const Scalar mean = s/Scalar(RealScalar(rows*cols)); 42 43 VERIFY_IS_APPROX(m1.sum(), s); 44 VERIFY_IS_APPROX(m1.mean(), mean); 45 VERIFY_IS_APPROX(m1_for_prod.prod(), p); 46 VERIFY_IS_APPROX(m1.real().minCoeff(), numext::real(minc)); 47 VERIFY_IS_APPROX(m1.real().maxCoeff(), numext::real(maxc)); 48 49 // test slice vectorization assuming assign is ok 50 Index r0 = internal::random<Index>(0,rows-1); 51 Index c0 = internal::random<Index>(0,cols-1); 52 Index r1 = internal::random<Index>(r0+1,rows)-r0; 53 Index c1 = internal::random<Index>(c0+1,cols)-c0; 54 VERIFY_IS_APPROX(m1.block(r0,c0,r1,c1).sum(), m1.block(r0,c0,r1,c1).eval().sum()); 55 VERIFY_IS_APPROX(m1.block(r0,c0,r1,c1).mean(), m1.block(r0,c0,r1,c1).eval().mean()); 56 VERIFY_IS_APPROX(m1_for_prod.block(r0,c0,r1,c1).prod(), m1_for_prod.block(r0,c0,r1,c1).eval().prod()); 57 VERIFY_IS_APPROX(m1.block(r0,c0,r1,c1).real().minCoeff(), m1.block(r0,c0,r1,c1).real().eval().minCoeff()); 58 VERIFY_IS_APPROX(m1.block(r0,c0,r1,c1).real().maxCoeff(), m1.block(r0,c0,r1,c1).real().eval().maxCoeff()); 59 60 // regression for bug 1090 61 const int R1 = MatrixType::RowsAtCompileTime>=2 ? MatrixType::RowsAtCompileTime/2 : 6; 62 const int C1 = MatrixType::ColsAtCompileTime>=2 ? MatrixType::ColsAtCompileTime/2 : 6; 63 if(R1<=rows-r0 && C1<=cols-c0) 64 { 65 VERIFY_IS_APPROX( (m1.template block<R1,C1>(r0,c0).sum()), m1.block(r0,c0,R1,C1).sum() ); 66 } 67 68 // test empty objects 69 VERIFY_IS_APPROX(m1.block(r0,c0,0,0).sum(), Scalar(0)); 70 VERIFY_IS_APPROX(m1.block(r0,c0,0,0).prod(), Scalar(1)); 71 72 // test nesting complex expression 73 VERIFY_EVALUATION_COUNT( (m1.matrix()*m1.matrix().transpose()).sum(), (MatrixType::IsVectorAtCompileTime && MatrixType::SizeAtCompileTime!=1 ? 0 : 1) ); 74 Matrix<Scalar, MatrixType::RowsAtCompileTime, MatrixType::RowsAtCompileTime> m2(rows,rows); 75 m2.setRandom(); 76 VERIFY_EVALUATION_COUNT( ((m1.matrix()*m1.matrix().transpose())+m2).sum(),(MatrixType::IsVectorAtCompileTime && MatrixType::SizeAtCompileTime!=1 ? 0 : 1)); 77 } 78 79 template<typename VectorType> void vectorRedux(const VectorType& w) 80 { 81 using std::abs; 82 typedef typename VectorType::Index Index; 83 typedef typename VectorType::Scalar Scalar; 84 typedef typename NumTraits<Scalar>::Real RealScalar; 85 Index size = w.size(); 86 87 VectorType v = VectorType::Random(size); 88 VectorType v_for_prod = VectorType::Ones(size) + Scalar(0.2) * v; // see comment above declaration of m1_for_prod 89 90 for(int i = 1; i < size; i++) 91 { 92 Scalar s(0), p(1); 93 RealScalar minc(numext::real(v.coeff(0))), maxc(numext::real(v.coeff(0))); 94 for(int j = 0; j < i; j++) 95 { 96 s += v[j]; 97 p *= v_for_prod[j]; 98 minc = (std::min)(minc, numext::real(v[j])); 99 maxc = (std::max)(maxc, numext::real(v[j])); 100 } 101 VERIFY_IS_MUCH_SMALLER_THAN(abs(s - v.head(i).sum()), Scalar(1)); 102 VERIFY_IS_APPROX(p, v_for_prod.head(i).prod()); 103 VERIFY_IS_APPROX(minc, v.real().head(i).minCoeff()); 104 VERIFY_IS_APPROX(maxc, v.real().head(i).maxCoeff()); 105 } 106 107 for(int i = 0; i < size-1; i++) 108 { 109 Scalar s(0), p(1); 110 RealScalar minc(numext::real(v.coeff(i))), maxc(numext::real(v.coeff(i))); 111 for(int j = i; j < size; j++) 112 { 113 s += v[j]; 114 p *= v_for_prod[j]; 115 minc = (std::min)(minc, numext::real(v[j])); 116 maxc = (std::max)(maxc, numext::real(v[j])); 117 } 118 VERIFY_IS_MUCH_SMALLER_THAN(abs(s - v.tail(size-i).sum()), Scalar(1)); 119 VERIFY_IS_APPROX(p, v_for_prod.tail(size-i).prod()); 120 VERIFY_IS_APPROX(minc, v.real().tail(size-i).minCoeff()); 121 VERIFY_IS_APPROX(maxc, v.real().tail(size-i).maxCoeff()); 122 } 123 124 for(int i = 0; i < size/2; i++) 125 { 126 Scalar s(0), p(1); 127 RealScalar minc(numext::real(v.coeff(i))), maxc(numext::real(v.coeff(i))); 128 for(int j = i; j < size-i; j++) 129 { 130 s += v[j]; 131 p *= v_for_prod[j]; 132 minc = (std::min)(minc, numext::real(v[j])); 133 maxc = (std::max)(maxc, numext::real(v[j])); 134 } 135 VERIFY_IS_MUCH_SMALLER_THAN(abs(s - v.segment(i, size-2*i).sum()), Scalar(1)); 136 VERIFY_IS_APPROX(p, v_for_prod.segment(i, size-2*i).prod()); 137 VERIFY_IS_APPROX(minc, v.real().segment(i, size-2*i).minCoeff()); 138 VERIFY_IS_APPROX(maxc, v.real().segment(i, size-2*i).maxCoeff()); 139 } 140 141 // test empty objects 142 VERIFY_IS_APPROX(v.head(0).sum(), Scalar(0)); 143 VERIFY_IS_APPROX(v.tail(0).prod(), Scalar(1)); 144 VERIFY_RAISES_ASSERT(v.head(0).mean()); 145 VERIFY_RAISES_ASSERT(v.head(0).minCoeff()); 146 VERIFY_RAISES_ASSERT(v.head(0).maxCoeff()); 147 } 148 149 void test_redux() 150 { 151 // the max size cannot be too large, otherwise reduxion operations obviously generate large errors. 152 int maxsize = (std::min)(100,EIGEN_TEST_MAX_SIZE); 153 TEST_SET_BUT_UNUSED_VARIABLE(maxsize); 154 for(int i = 0; i < g_repeat; i++) { 155 CALL_SUBTEST_1( matrixRedux(Matrix<float, 1, 1>()) ); 156 CALL_SUBTEST_1( matrixRedux(Array<float, 1, 1>()) ); 157 CALL_SUBTEST_2( matrixRedux(Matrix2f()) ); 158 CALL_SUBTEST_2( matrixRedux(Array2f()) ); 159 CALL_SUBTEST_2( matrixRedux(Array22f()) ); 160 CALL_SUBTEST_3( matrixRedux(Matrix4d()) ); 161 CALL_SUBTEST_3( matrixRedux(Array4d()) ); 162 CALL_SUBTEST_3( matrixRedux(Array44d()) ); 163 CALL_SUBTEST_4( matrixRedux(MatrixXcf(internal::random<int>(1,maxsize), internal::random<int>(1,maxsize))) ); 164 CALL_SUBTEST_4( matrixRedux(ArrayXXcf(internal::random<int>(1,maxsize), internal::random<int>(1,maxsize))) ); 165 CALL_SUBTEST_5( matrixRedux(MatrixXd (internal::random<int>(1,maxsize), internal::random<int>(1,maxsize))) ); 166 CALL_SUBTEST_5( matrixRedux(ArrayXXd (internal::random<int>(1,maxsize), internal::random<int>(1,maxsize))) ); 167 CALL_SUBTEST_6( matrixRedux(MatrixXi (internal::random<int>(1,maxsize), internal::random<int>(1,maxsize))) ); 168 CALL_SUBTEST_6( matrixRedux(ArrayXXi (internal::random<int>(1,maxsize), internal::random<int>(1,maxsize))) ); 169 } 170 for(int i = 0; i < g_repeat; i++) { 171 CALL_SUBTEST_7( vectorRedux(Vector4f()) ); 172 CALL_SUBTEST_7( vectorRedux(Array4f()) ); 173 CALL_SUBTEST_5( vectorRedux(VectorXd(internal::random<int>(1,maxsize))) ); 174 CALL_SUBTEST_5( vectorRedux(ArrayXd(internal::random<int>(1,maxsize))) ); 175 CALL_SUBTEST_8( vectorRedux(VectorXf(internal::random<int>(1,maxsize))) ); 176 CALL_SUBTEST_8( vectorRedux(ArrayXf(internal::random<int>(1,maxsize))) ); 177 } 178 } 179