1// This file is part of Eigen, a lightweight C++ template library
2// for linear algebra.
3//
4// Copyright (C) 2009 Gael Guennebaud <gael.guennebaud@inria.fr>
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#ifndef EIGEN_AUTODIFF_VECTOR_H
11#define EIGEN_AUTODIFF_VECTOR_H
12
13namespace Eigen {
14
15/* \class AutoDiffScalar
16  * \brief A scalar type replacement with automatic differentation capability
17  *
18  * \param DerType the vector type used to store/represent the derivatives (e.g. Vector3f)
19  *
20  * This class represents a scalar value while tracking its respective derivatives.
21  *
22  * It supports the following list of global math function:
23  *  - std::abs, std::sqrt, std::pow, std::exp, std::log, std::sin, std::cos,
24  *  - internal::abs, internal::sqrt, numext::pow, internal::exp, internal::log, internal::sin, internal::cos,
25  *  - internal::conj, internal::real, internal::imag, numext::abs2.
26  *
27  * AutoDiffScalar can be used as the scalar type of an Eigen::Matrix object. However,
28  * in that case, the expression template mechanism only occurs at the top Matrix level,
29  * while derivatives are computed right away.
30  *
31  */
32template<typename ValueType, typename JacobianType>
33class AutoDiffVector
34{
35  public:
36    //typedef typename internal::traits<ValueType>::Scalar Scalar;
37    typedef typename internal::traits<ValueType>::Scalar BaseScalar;
38    typedef AutoDiffScalar<Matrix<BaseScalar,JacobianType::RowsAtCompileTime,1> > ActiveScalar;
39    typedef ActiveScalar Scalar;
40    typedef AutoDiffScalar<typename JacobianType::ColXpr> CoeffType;
41    typedef typename JacobianType::Index Index;
42
43    inline AutoDiffVector() {}
44
45    inline AutoDiffVector(const ValueType& values)
46      : m_values(values)
47    {
48      m_jacobian.setZero();
49    }
50
51
52    CoeffType operator[] (Index i) { return CoeffType(m_values[i], m_jacobian.col(i)); }
53    const CoeffType operator[] (Index i) const { return CoeffType(m_values[i], m_jacobian.col(i)); }
54
55    CoeffType operator() (Index i) { return CoeffType(m_values[i], m_jacobian.col(i)); }
56    const CoeffType operator() (Index i) const { return CoeffType(m_values[i], m_jacobian.col(i)); }
57
58    CoeffType coeffRef(Index i) { return CoeffType(m_values[i], m_jacobian.col(i)); }
59    const CoeffType coeffRef(Index i) const { return CoeffType(m_values[i], m_jacobian.col(i)); }
60
61    Index size() const { return m_values.size(); }
62
63    // FIXME here we could return an expression of the sum
64    Scalar sum() const { /*std::cerr << "sum \n\n";*/ /*std::cerr << m_jacobian.rowwise().sum() << "\n\n";*/ return Scalar(m_values.sum(), m_jacobian.rowwise().sum()); }
65
66
67    inline AutoDiffVector(const ValueType& values, const JacobianType& jac)
68      : m_values(values), m_jacobian(jac)
69    {}
70
71    template<typename OtherValueType, typename OtherJacobianType>
72    inline AutoDiffVector(const AutoDiffVector<OtherValueType, OtherJacobianType>& other)
73      : m_values(other.values()), m_jacobian(other.jacobian())
74    {}
75
76    inline AutoDiffVector(const AutoDiffVector& other)
77      : m_values(other.values()), m_jacobian(other.jacobian())
78    {}
79
80    template<typename OtherValueType, typename OtherJacobianType>
81    inline AutoDiffVector& operator=(const AutoDiffVector<OtherValueType, OtherJacobianType>& other)
82    {
83      m_values = other.values();
84      m_jacobian = other.jacobian();
85      return *this;
86    }
87
88    inline AutoDiffVector& operator=(const AutoDiffVector& other)
89    {
90      m_values = other.values();
91      m_jacobian = other.jacobian();
92      return *this;
93    }
94
95    inline const ValueType& values() const { return m_values; }
96    inline ValueType& values() { return m_values; }
97
98    inline const JacobianType& jacobian() const { return m_jacobian; }
99    inline JacobianType& jacobian() { return m_jacobian; }
100
101    template<typename OtherValueType,typename OtherJacobianType>
102    inline const AutoDiffVector<
103      typename MakeCwiseBinaryOp<internal::scalar_sum_op<BaseScalar>,ValueType,OtherValueType>::Type,
104      typename MakeCwiseBinaryOp<internal::scalar_sum_op<BaseScalar>,JacobianType,OtherJacobianType>::Type >
105    operator+(const AutoDiffVector<OtherValueType,OtherJacobianType>& other) const
106    {
107      return AutoDiffVector<
108      typename MakeCwiseBinaryOp<internal::scalar_sum_op<BaseScalar>,ValueType,OtherValueType>::Type,
109      typename MakeCwiseBinaryOp<internal::scalar_sum_op<BaseScalar>,JacobianType,OtherJacobianType>::Type >(
110        m_values + other.values(),
111        m_jacobian + other.jacobian());
112    }
113
114    template<typename OtherValueType, typename OtherJacobianType>
115    inline AutoDiffVector&
116    operator+=(const AutoDiffVector<OtherValueType,OtherJacobianType>& other)
117    {
118      m_values += other.values();
119      m_jacobian += other.jacobian();
120      return *this;
121    }
122
123    template<typename OtherValueType,typename OtherJacobianType>
124    inline const AutoDiffVector<
125      typename MakeCwiseBinaryOp<internal::scalar_difference_op<Scalar>,ValueType,OtherValueType>::Type,
126      typename MakeCwiseBinaryOp<internal::scalar_difference_op<Scalar>,JacobianType,OtherJacobianType>::Type >
127    operator-(const AutoDiffVector<OtherValueType,OtherJacobianType>& other) const
128    {
129      return AutoDiffVector<
130        typename MakeCwiseBinaryOp<internal::scalar_difference_op<Scalar>,ValueType,OtherValueType>::Type,
131        typename MakeCwiseBinaryOp<internal::scalar_difference_op<Scalar>,JacobianType,OtherJacobianType>::Type >(
132          m_values - other.values(),
133          m_jacobian - other.jacobian());
134    }
135
136    template<typename OtherValueType, typename OtherJacobianType>
137    inline AutoDiffVector&
138    operator-=(const AutoDiffVector<OtherValueType,OtherJacobianType>& other)
139    {
140      m_values -= other.values();
141      m_jacobian -= other.jacobian();
142      return *this;
143    }
144
145    inline const AutoDiffVector<
146      typename MakeCwiseUnaryOp<internal::scalar_opposite_op<Scalar>, ValueType>::Type,
147      typename MakeCwiseUnaryOp<internal::scalar_opposite_op<Scalar>, JacobianType>::Type >
148    operator-() const
149    {
150      return AutoDiffVector<
151        typename MakeCwiseUnaryOp<internal::scalar_opposite_op<Scalar>, ValueType>::Type,
152        typename MakeCwiseUnaryOp<internal::scalar_opposite_op<Scalar>, JacobianType>::Type >(
153          -m_values,
154          -m_jacobian);
155    }
156
157    inline const AutoDiffVector<
158      typename MakeCwiseUnaryOp<internal::scalar_multiple_op<Scalar>, ValueType>::Type,
159      typename MakeCwiseUnaryOp<internal::scalar_multiple_op<Scalar>, JacobianType>::Type>
160    operator*(const BaseScalar& other) const
161    {
162      return AutoDiffVector<
163        typename MakeCwiseUnaryOp<internal::scalar_multiple_op<Scalar>, ValueType>::Type,
164        typename MakeCwiseUnaryOp<internal::scalar_multiple_op<Scalar>, JacobianType>::Type >(
165          m_values * other,
166          m_jacobian * other);
167    }
168
169    friend inline const AutoDiffVector<
170      typename MakeCwiseUnaryOp<internal::scalar_multiple_op<Scalar>, ValueType>::Type,
171      typename MakeCwiseUnaryOp<internal::scalar_multiple_op<Scalar>, JacobianType>::Type >
172    operator*(const Scalar& other, const AutoDiffVector& v)
173    {
174      return AutoDiffVector<
175        typename MakeCwiseUnaryOp<internal::scalar_multiple_op<Scalar>, ValueType>::Type,
176        typename MakeCwiseUnaryOp<internal::scalar_multiple_op<Scalar>, JacobianType>::Type >(
177          v.values() * other,
178          v.jacobian() * other);
179    }
180
181//     template<typename OtherValueType,typename OtherJacobianType>
182//     inline const AutoDiffVector<
183//       CwiseBinaryOp<internal::scalar_multiple_op<Scalar>, ValueType, OtherValueType>
184//       CwiseBinaryOp<internal::scalar_sum_op<Scalar>,
185//         CwiseUnaryOp<internal::scalar_multiple_op<Scalar>, JacobianType>,
186//         CwiseUnaryOp<internal::scalar_multiple_op<Scalar>, OtherJacobianType> > >
187//     operator*(const AutoDiffVector<OtherValueType,OtherJacobianType>& other) const
188//     {
189//       return AutoDiffVector<
190//         CwiseBinaryOp<internal::scalar_multiple_op<Scalar>, ValueType, OtherValueType>
191//         CwiseBinaryOp<internal::scalar_sum_op<Scalar>,
192//           CwiseUnaryOp<internal::scalar_multiple_op<Scalar>, JacobianType>,
193//           CwiseUnaryOp<internal::scalar_multiple_op<Scalar>, OtherJacobianType> > >(
194//             m_values.cwise() * other.values(),
195//             (m_jacobian * other.values()) + (m_values * other.jacobian()));
196//     }
197
198    inline AutoDiffVector& operator*=(const Scalar& other)
199    {
200      m_values *= other;
201      m_jacobian *= other;
202      return *this;
203    }
204
205    template<typename OtherValueType,typename OtherJacobianType>
206    inline AutoDiffVector& operator*=(const AutoDiffVector<OtherValueType,OtherJacobianType>& other)
207    {
208      *this = *this * other;
209      return *this;
210    }
211
212  protected:
213    ValueType m_values;
214    JacobianType m_jacobian;
215
216};
217
218}
219
220#endif // EIGEN_AUTODIFF_VECTOR_H
221