Brian Silverman | 72890c2 | 2015-09-19 14:37:37 -0400 | [diff] [blame^] | 1 | // This file is part of Eigen, a lightweight C++ template library |
| 2 | // for linear algebra. |
| 3 | // |
| 4 | // Copyright (C) 2008-2009 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 | #ifndef EIGEN_ORTHOMETHODS_H |
| 12 | #define EIGEN_ORTHOMETHODS_H |
| 13 | |
| 14 | namespace Eigen { |
| 15 | |
| 16 | /** \geometry_module |
| 17 | * |
| 18 | * \returns the cross product of \c *this and \a other |
| 19 | * |
| 20 | * Here is a very good explanation of cross-product: http://xkcd.com/199/ |
| 21 | * \sa MatrixBase::cross3() |
| 22 | */ |
| 23 | template<typename Derived> |
| 24 | template<typename OtherDerived> |
| 25 | inline typename MatrixBase<Derived>::template cross_product_return_type<OtherDerived>::type |
| 26 | MatrixBase<Derived>::cross(const MatrixBase<OtherDerived>& other) const |
| 27 | { |
| 28 | EIGEN_STATIC_ASSERT_VECTOR_SPECIFIC_SIZE(Derived,3) |
| 29 | EIGEN_STATIC_ASSERT_VECTOR_SPECIFIC_SIZE(OtherDerived,3) |
| 30 | |
| 31 | // Note that there is no need for an expression here since the compiler |
| 32 | // optimize such a small temporary very well (even within a complex expression) |
| 33 | typename internal::nested<Derived,2>::type lhs(derived()); |
| 34 | typename internal::nested<OtherDerived,2>::type rhs(other.derived()); |
| 35 | return typename cross_product_return_type<OtherDerived>::type( |
| 36 | numext::conj(lhs.coeff(1) * rhs.coeff(2) - lhs.coeff(2) * rhs.coeff(1)), |
| 37 | numext::conj(lhs.coeff(2) * rhs.coeff(0) - lhs.coeff(0) * rhs.coeff(2)), |
| 38 | numext::conj(lhs.coeff(0) * rhs.coeff(1) - lhs.coeff(1) * rhs.coeff(0)) |
| 39 | ); |
| 40 | } |
| 41 | |
| 42 | namespace internal { |
| 43 | |
| 44 | template< int Arch,typename VectorLhs,typename VectorRhs, |
| 45 | typename Scalar = typename VectorLhs::Scalar, |
| 46 | bool Vectorizable = bool((VectorLhs::Flags&VectorRhs::Flags)&PacketAccessBit)> |
| 47 | struct cross3_impl { |
| 48 | static inline typename internal::plain_matrix_type<VectorLhs>::type |
| 49 | run(const VectorLhs& lhs, const VectorRhs& rhs) |
| 50 | { |
| 51 | return typename internal::plain_matrix_type<VectorLhs>::type( |
| 52 | numext::conj(lhs.coeff(1) * rhs.coeff(2) - lhs.coeff(2) * rhs.coeff(1)), |
| 53 | numext::conj(lhs.coeff(2) * rhs.coeff(0) - lhs.coeff(0) * rhs.coeff(2)), |
| 54 | numext::conj(lhs.coeff(0) * rhs.coeff(1) - lhs.coeff(1) * rhs.coeff(0)), |
| 55 | 0 |
| 56 | ); |
| 57 | } |
| 58 | }; |
| 59 | |
| 60 | } |
| 61 | |
| 62 | /** \geometry_module |
| 63 | * |
| 64 | * \returns the cross product of \c *this and \a other using only the x, y, and z coefficients |
| 65 | * |
| 66 | * The size of \c *this and \a other must be four. This function is especially useful |
| 67 | * when using 4D vectors instead of 3D ones to get advantage of SSE/AltiVec vectorization. |
| 68 | * |
| 69 | * \sa MatrixBase::cross() |
| 70 | */ |
| 71 | template<typename Derived> |
| 72 | template<typename OtherDerived> |
| 73 | inline typename MatrixBase<Derived>::PlainObject |
| 74 | MatrixBase<Derived>::cross3(const MatrixBase<OtherDerived>& other) const |
| 75 | { |
| 76 | EIGEN_STATIC_ASSERT_VECTOR_SPECIFIC_SIZE(Derived,4) |
| 77 | EIGEN_STATIC_ASSERT_VECTOR_SPECIFIC_SIZE(OtherDerived,4) |
| 78 | |
| 79 | typedef typename internal::nested<Derived,2>::type DerivedNested; |
| 80 | typedef typename internal::nested<OtherDerived,2>::type OtherDerivedNested; |
| 81 | DerivedNested lhs(derived()); |
| 82 | OtherDerivedNested rhs(other.derived()); |
| 83 | |
| 84 | return internal::cross3_impl<Architecture::Target, |
| 85 | typename internal::remove_all<DerivedNested>::type, |
| 86 | typename internal::remove_all<OtherDerivedNested>::type>::run(lhs,rhs); |
| 87 | } |
| 88 | |
| 89 | /** \returns a matrix expression of the cross product of each column or row |
| 90 | * of the referenced expression with the \a other vector. |
| 91 | * |
| 92 | * The referenced matrix must have one dimension equal to 3. |
| 93 | * The result matrix has the same dimensions than the referenced one. |
| 94 | * |
| 95 | * \geometry_module |
| 96 | * |
| 97 | * \sa MatrixBase::cross() */ |
| 98 | template<typename ExpressionType, int Direction> |
| 99 | template<typename OtherDerived> |
| 100 | const typename VectorwiseOp<ExpressionType,Direction>::CrossReturnType |
| 101 | VectorwiseOp<ExpressionType,Direction>::cross(const MatrixBase<OtherDerived>& other) const |
| 102 | { |
| 103 | EIGEN_STATIC_ASSERT_VECTOR_SPECIFIC_SIZE(OtherDerived,3) |
| 104 | EIGEN_STATIC_ASSERT((internal::is_same<Scalar, typename OtherDerived::Scalar>::value), |
| 105 | YOU_MIXED_DIFFERENT_NUMERIC_TYPES__YOU_NEED_TO_USE_THE_CAST_METHOD_OF_MATRIXBASE_TO_CAST_NUMERIC_TYPES_EXPLICITLY) |
| 106 | |
| 107 | CrossReturnType res(_expression().rows(),_expression().cols()); |
| 108 | if(Direction==Vertical) |
| 109 | { |
| 110 | eigen_assert(CrossReturnType::RowsAtCompileTime==3 && "the matrix must have exactly 3 rows"); |
| 111 | res.row(0) = (_expression().row(1) * other.coeff(2) - _expression().row(2) * other.coeff(1)).conjugate(); |
| 112 | res.row(1) = (_expression().row(2) * other.coeff(0) - _expression().row(0) * other.coeff(2)).conjugate(); |
| 113 | res.row(2) = (_expression().row(0) * other.coeff(1) - _expression().row(1) * other.coeff(0)).conjugate(); |
| 114 | } |
| 115 | else |
| 116 | { |
| 117 | eigen_assert(CrossReturnType::ColsAtCompileTime==3 && "the matrix must have exactly 3 columns"); |
| 118 | res.col(0) = (_expression().col(1) * other.coeff(2) - _expression().col(2) * other.coeff(1)).conjugate(); |
| 119 | res.col(1) = (_expression().col(2) * other.coeff(0) - _expression().col(0) * other.coeff(2)).conjugate(); |
| 120 | res.col(2) = (_expression().col(0) * other.coeff(1) - _expression().col(1) * other.coeff(0)).conjugate(); |
| 121 | } |
| 122 | return res; |
| 123 | } |
| 124 | |
| 125 | namespace internal { |
| 126 | |
| 127 | template<typename Derived, int Size = Derived::SizeAtCompileTime> |
| 128 | struct unitOrthogonal_selector |
| 129 | { |
| 130 | typedef typename plain_matrix_type<Derived>::type VectorType; |
| 131 | typedef typename traits<Derived>::Scalar Scalar; |
| 132 | typedef typename NumTraits<Scalar>::Real RealScalar; |
| 133 | typedef typename Derived::Index Index; |
| 134 | typedef Matrix<Scalar,2,1> Vector2; |
| 135 | static inline VectorType run(const Derived& src) |
| 136 | { |
| 137 | VectorType perp = VectorType::Zero(src.size()); |
| 138 | Index maxi = 0; |
| 139 | Index sndi = 0; |
| 140 | src.cwiseAbs().maxCoeff(&maxi); |
| 141 | if (maxi==0) |
| 142 | sndi = 1; |
| 143 | RealScalar invnm = RealScalar(1)/(Vector2() << src.coeff(sndi),src.coeff(maxi)).finished().norm(); |
| 144 | perp.coeffRef(maxi) = -numext::conj(src.coeff(sndi)) * invnm; |
| 145 | perp.coeffRef(sndi) = numext::conj(src.coeff(maxi)) * invnm; |
| 146 | |
| 147 | return perp; |
| 148 | } |
| 149 | }; |
| 150 | |
| 151 | template<typename Derived> |
| 152 | struct unitOrthogonal_selector<Derived,3> |
| 153 | { |
| 154 | typedef typename plain_matrix_type<Derived>::type VectorType; |
| 155 | typedef typename traits<Derived>::Scalar Scalar; |
| 156 | typedef typename NumTraits<Scalar>::Real RealScalar; |
| 157 | static inline VectorType run(const Derived& src) |
| 158 | { |
| 159 | VectorType perp; |
| 160 | /* Let us compute the crossed product of *this with a vector |
| 161 | * that is not too close to being colinear to *this. |
| 162 | */ |
| 163 | |
| 164 | /* unless the x and y coords are both close to zero, we can |
| 165 | * simply take ( -y, x, 0 ) and normalize it. |
| 166 | */ |
| 167 | if((!isMuchSmallerThan(src.x(), src.z())) |
| 168 | || (!isMuchSmallerThan(src.y(), src.z()))) |
| 169 | { |
| 170 | RealScalar invnm = RealScalar(1)/src.template head<2>().norm(); |
| 171 | perp.coeffRef(0) = -numext::conj(src.y())*invnm; |
| 172 | perp.coeffRef(1) = numext::conj(src.x())*invnm; |
| 173 | perp.coeffRef(2) = 0; |
| 174 | } |
| 175 | /* if both x and y are close to zero, then the vector is close |
| 176 | * to the z-axis, so it's far from colinear to the x-axis for instance. |
| 177 | * So we take the crossed product with (1,0,0) and normalize it. |
| 178 | */ |
| 179 | else |
| 180 | { |
| 181 | RealScalar invnm = RealScalar(1)/src.template tail<2>().norm(); |
| 182 | perp.coeffRef(0) = 0; |
| 183 | perp.coeffRef(1) = -numext::conj(src.z())*invnm; |
| 184 | perp.coeffRef(2) = numext::conj(src.y())*invnm; |
| 185 | } |
| 186 | |
| 187 | return perp; |
| 188 | } |
| 189 | }; |
| 190 | |
| 191 | template<typename Derived> |
| 192 | struct unitOrthogonal_selector<Derived,2> |
| 193 | { |
| 194 | typedef typename plain_matrix_type<Derived>::type VectorType; |
| 195 | static inline VectorType run(const Derived& src) |
| 196 | { return VectorType(-numext::conj(src.y()), numext::conj(src.x())).normalized(); } |
| 197 | }; |
| 198 | |
| 199 | } // end namespace internal |
| 200 | |
| 201 | /** \returns a unit vector which is orthogonal to \c *this |
| 202 | * |
| 203 | * The size of \c *this must be at least 2. If the size is exactly 2, |
| 204 | * then the returned vector is a counter clock wise rotation of \c *this, i.e., (-y,x).normalized(). |
| 205 | * |
| 206 | * \sa cross() |
| 207 | */ |
| 208 | template<typename Derived> |
| 209 | typename MatrixBase<Derived>::PlainObject |
| 210 | MatrixBase<Derived>::unitOrthogonal() const |
| 211 | { |
| 212 | EIGEN_STATIC_ASSERT_VECTOR_ONLY(Derived) |
| 213 | return internal::unitOrthogonal_selector<Derived>::run(derived()); |
| 214 | } |
| 215 | |
| 216 | } // end namespace Eigen |
| 217 | |
| 218 | #endif // EIGEN_ORTHOMETHODS_H |