1 // This file is part of Eigen, a lightweight C++ template library 2 // for linear algebra. 3 // 4 // Copyright (C) 2006-2008 Benoit Jacob <jacob.benoit.1 (at) gmail.com> 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 #define EIGEN_NO_STATIC_ASSERT 11 12 #include "main.h" 13 14 template<bool IsInteger> struct adjoint_specific; 15 16 template<> struct adjoint_specific<true> { 17 template<typename Vec, typename Mat, typename Scalar> 18 static void run(const Vec& v1, const Vec& v2, Vec& v3, const Mat& square, Scalar s1, Scalar s2) { 19 VERIFY(test_isApproxWithRef((s1 * v1 + s2 * v2).dot(v3), numext::conj(s1) * v1.dot(v3) + numext::conj(s2) * v2.dot(v3), 0)); 20 VERIFY(test_isApproxWithRef(v3.dot(s1 * v1 + s2 * v2), s1*v3.dot(v1)+s2*v3.dot(v2), 0)); 21 22 // check compatibility of dot and adjoint 23 VERIFY(test_isApproxWithRef(v1.dot(square * v2), (square.adjoint() * v1).dot(v2), 0)); 24 } 25 }; 26 27 template<> struct adjoint_specific<false> { 28 template<typename Vec, typename Mat, typename Scalar> 29 static void run(const Vec& v1, const Vec& v2, Vec& v3, const Mat& square, Scalar s1, Scalar s2) { 30 typedef typename NumTraits<Scalar>::Real RealScalar; 31 using std::abs; 32 33 RealScalar ref = NumTraits<Scalar>::IsInteger ? RealScalar(0) : (std::max)((s1 * v1 + s2 * v2).norm(),v3.norm()); 34 VERIFY(test_isApproxWithRef((s1 * v1 + s2 * v2).dot(v3), numext::conj(s1) * v1.dot(v3) + numext::conj(s2) * v2.dot(v3), ref)); 35 VERIFY(test_isApproxWithRef(v3.dot(s1 * v1 + s2 * v2), s1*v3.dot(v1)+s2*v3.dot(v2), ref)); 36 37 VERIFY_IS_APPROX(v1.squaredNorm(), v1.norm() * v1.norm()); 38 // check normalized() and normalize() 39 VERIFY_IS_APPROX(v1, v1.norm() * v1.normalized()); 40 v3 = v1; 41 v3.normalize(); 42 VERIFY_IS_APPROX(v1, v1.norm() * v3); 43 VERIFY_IS_APPROX(v3, v1.normalized()); 44 VERIFY_IS_APPROX(v3.norm(), RealScalar(1)); 45 46 // check null inputs 47 VERIFY_IS_APPROX((v1*0).normalized(), (v1*0)); 48 #if (!EIGEN_ARCH_i386) || defined(EIGEN_VECTORIZE) 49 RealScalar very_small = (std::numeric_limits<RealScalar>::min)(); 50 VERIFY( (v1*very_small).norm() == 0 ); 51 VERIFY_IS_APPROX((v1*very_small).normalized(), (v1*very_small)); 52 v3 = v1*very_small; 53 v3.normalize(); 54 VERIFY_IS_APPROX(v3, (v1*very_small)); 55 #endif 56 57 // check compatibility of dot and adjoint 58 ref = NumTraits<Scalar>::IsInteger ? 0 : (std::max)((std::max)(v1.norm(),v2.norm()),(std::max)((square * v2).norm(),(square.adjoint() * v1).norm())); 59 VERIFY(internal::isMuchSmallerThan(abs(v1.dot(square * v2) - (square.adjoint() * v1).dot(v2)), ref, test_precision<Scalar>())); 60 61 // check that Random().normalized() works: tricky as the random xpr must be evaluated by 62 // normalized() in order to produce a consistent result. 63 VERIFY_IS_APPROX(Vec::Random(v1.size()).normalized().norm(), RealScalar(1)); 64 } 65 }; 66 67 template<typename MatrixType> void adjoint(const MatrixType& m) 68 { 69 /* this test covers the following files: 70 Transpose.h Conjugate.h Dot.h 71 */ 72 using std::abs; 73 typedef typename MatrixType::Index Index; 74 typedef typename MatrixType::Scalar Scalar; 75 typedef typename NumTraits<Scalar>::Real RealScalar; 76 typedef Matrix<Scalar, MatrixType::RowsAtCompileTime, 1> VectorType; 77 typedef Matrix<Scalar, MatrixType::RowsAtCompileTime, MatrixType::RowsAtCompileTime> SquareMatrixType; 78 const Index PacketSize = internal::packet_traits<Scalar>::size; 79 80 Index rows = m.rows(); 81 Index cols = m.cols(); 82 83 MatrixType m1 = MatrixType::Random(rows, cols), 84 m2 = MatrixType::Random(rows, cols), 85 m3(rows, cols), 86 square = SquareMatrixType::Random(rows, rows); 87 VectorType v1 = VectorType::Random(rows), 88 v2 = VectorType::Random(rows), 89 v3 = VectorType::Random(rows), 90 vzero = VectorType::Zero(rows); 91 92 Scalar s1 = internal::random<Scalar>(), 93 s2 = internal::random<Scalar>(); 94 95 // check basic compatibility of adjoint, transpose, conjugate 96 VERIFY_IS_APPROX(m1.transpose().conjugate().adjoint(), m1); 97 VERIFY_IS_APPROX(m1.adjoint().conjugate().transpose(), m1); 98 99 // check multiplicative behavior 100 VERIFY_IS_APPROX((m1.adjoint() * m2).adjoint(), m2.adjoint() * m1); 101 VERIFY_IS_APPROX((s1 * m1).adjoint(), numext::conj(s1) * m1.adjoint()); 102 103 // check basic properties of dot, squaredNorm 104 VERIFY_IS_APPROX(numext::conj(v1.dot(v2)), v2.dot(v1)); 105 VERIFY_IS_APPROX(numext::real(v1.dot(v1)), v1.squaredNorm()); 106 107 adjoint_specific<NumTraits<Scalar>::IsInteger>::run(v1, v2, v3, square, s1, s2); 108 109 VERIFY_IS_MUCH_SMALLER_THAN(abs(vzero.dot(v1)), static_cast<RealScalar>(1)); 110 111 // like in testBasicStuff, test operator() to check const-qualification 112 Index r = internal::random<Index>(0, rows-1), 113 c = internal::random<Index>(0, cols-1); 114 VERIFY_IS_APPROX(m1.conjugate()(r,c), numext::conj(m1(r,c))); 115 VERIFY_IS_APPROX(m1.adjoint()(c,r), numext::conj(m1(r,c))); 116 117 // check inplace transpose 118 m3 = m1; 119 m3.transposeInPlace(); 120 VERIFY_IS_APPROX(m3,m1.transpose()); 121 m3.transposeInPlace(); 122 VERIFY_IS_APPROX(m3,m1); 123 124 if(PacketSize<m3.rows() && PacketSize<m3.cols()) 125 { 126 m3 = m1; 127 Index i = internal::random<Index>(0,m3.rows()-PacketSize); 128 Index j = internal::random<Index>(0,m3.cols()-PacketSize); 129 m3.template block<PacketSize,PacketSize>(i,j).transposeInPlace(); 130 VERIFY_IS_APPROX( (m3.template block<PacketSize,PacketSize>(i,j)), (m1.template block<PacketSize,PacketSize>(i,j).transpose()) ); 131 m3.template block<PacketSize,PacketSize>(i,j).transposeInPlace(); 132 VERIFY_IS_APPROX(m3,m1); 133 } 134 135 // check inplace adjoint 136 m3 = m1; 137 m3.adjointInPlace(); 138 VERIFY_IS_APPROX(m3,m1.adjoint()); 139 m3.transposeInPlace(); 140 VERIFY_IS_APPROX(m3,m1.conjugate()); 141 142 // check mixed dot product 143 typedef Matrix<RealScalar, MatrixType::RowsAtCompileTime, 1> RealVectorType; 144 RealVectorType rv1 = RealVectorType::Random(rows); 145 VERIFY_IS_APPROX(v1.dot(rv1.template cast<Scalar>()), v1.dot(rv1)); 146 VERIFY_IS_APPROX(rv1.template cast<Scalar>().dot(v1), rv1.dot(v1)); 147 } 148 149 void test_adjoint() 150 { 151 for(int i = 0; i < g_repeat; i++) { 152 CALL_SUBTEST_1( adjoint(Matrix<float, 1, 1>()) ); 153 CALL_SUBTEST_2( adjoint(Matrix3d()) ); 154 CALL_SUBTEST_3( adjoint(Matrix4f()) ); 155 156 CALL_SUBTEST_4( adjoint(MatrixXcf(internal::random<int>(1,EIGEN_TEST_MAX_SIZE/2), internal::random<int>(1,EIGEN_TEST_MAX_SIZE/2))) ); 157 CALL_SUBTEST_5( adjoint(MatrixXi(internal::random<int>(1,EIGEN_TEST_MAX_SIZE), internal::random<int>(1,EIGEN_TEST_MAX_SIZE))) ); 158 CALL_SUBTEST_6( adjoint(MatrixXf(internal::random<int>(1,EIGEN_TEST_MAX_SIZE), internal::random<int>(1,EIGEN_TEST_MAX_SIZE))) ); 159 160 // Complement for 128 bits vectorization: 161 CALL_SUBTEST_8( adjoint(Matrix2d()) ); 162 CALL_SUBTEST_9( adjoint(Matrix<int,4,4>()) ); 163 164 // 256 bits vectorization: 165 CALL_SUBTEST_10( adjoint(Matrix<float,8,8>()) ); 166 CALL_SUBTEST_11( adjoint(Matrix<double,4,4>()) ); 167 CALL_SUBTEST_12( adjoint(Matrix<int,8,8>()) ); 168 } 169 // test a large static matrix only once 170 CALL_SUBTEST_7( adjoint(Matrix<float, 100, 100>()) ); 171 172 #ifdef EIGEN_TEST_PART_13 173 { 174 MatrixXcf a(10,10), b(10,10); 175 VERIFY_RAISES_ASSERT(a = a.transpose()); 176 VERIFY_RAISES_ASSERT(a = a.transpose() + b); 177 VERIFY_RAISES_ASSERT(a = b + a.transpose()); 178 VERIFY_RAISES_ASSERT(a = a.conjugate().transpose()); 179 VERIFY_RAISES_ASSERT(a = a.adjoint()); 180 VERIFY_RAISES_ASSERT(a = a.adjoint() + b); 181 VERIFY_RAISES_ASSERT(a = b + a.adjoint()); 182 183 // no assertion should be triggered for these cases: 184 a.transpose() = a.transpose(); 185 a.transpose() += a.transpose(); 186 a.transpose() += a.transpose() + b; 187 a.transpose() = a.adjoint(); 188 a.transpose() += a.adjoint(); 189 a.transpose() += a.adjoint() + b; 190 191 // regression tests for check_for_aliasing 192 MatrixXd c(10,10); 193 c = 1.0 * MatrixXd::Ones(10,10) + c; 194 c = MatrixXd::Ones(10,10) * 1.0 + c; 195 c = c + MatrixXd::Ones(10,10) .cwiseProduct( MatrixXd::Zero(10,10) ); 196 c = MatrixXd::Ones(10,10) * MatrixXd::Zero(10,10); 197 } 198 #endif 199 } 200 201