1c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath// This file is part of Eigen, a lightweight C++ template library
2c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath// for linear algebra.
3c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath//
4c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath// Copyright (C) 2011 Gael Guennebaud <gael.guennebaud@inria.fr>
5c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath//
6c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath// This Source Code Form is subject to the terms of the Mozilla
7c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath// Public License v. 2.0. If a copy of the MPL was not distributed
8c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath// with this file, You can obtain one at http://mozilla.org/MPL/2.0/.
9c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath
10c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath#ifndef EIGEN_INCOMPLETE_LU_H
11c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath#define EIGEN_INCOMPLETE_LU_H
12c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath
13c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamathnamespace Eigen {
14c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath
15c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamathtemplate <typename _Scalar>
16c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamathclass IncompleteLU
17c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath{
18c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath    typedef _Scalar Scalar;
19c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath    typedef Matrix<Scalar,Dynamic,1> Vector;
20c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath    typedef typename Vector::Index Index;
21c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath    typedef SparseMatrix<Scalar,RowMajor> FactorType;
22c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath
23c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath  public:
24c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath    typedef Matrix<Scalar,Dynamic,Dynamic> MatrixType;
25c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath
26c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath    IncompleteLU() : m_isInitialized(false) {}
27c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath
28c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath    template<typename MatrixType>
29c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath    IncompleteLU(const MatrixType& mat) : m_isInitialized(false)
30c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath    {
31c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath      compute(mat);
32c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath    }
33c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath
34c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath    Index rows() const { return m_lu.rows(); }
35c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath    Index cols() const { return m_lu.cols(); }
36c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath
37c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath    template<typename MatrixType>
38c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath    IncompleteLU& compute(const MatrixType& mat)
39c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath    {
40c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath      m_lu = mat;
41c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath      int size = mat.cols();
42c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath      Vector diag(size);
43c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath      for(int i=0; i<size; ++i)
44c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath      {
45c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath        typename FactorType::InnerIterator k_it(m_lu,i);
46c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath        for(; k_it && k_it.index()<i; ++k_it)
47c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath        {
48c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath          int k = k_it.index();
49c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath          k_it.valueRef() /= diag(k);
50c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath
51c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath          typename FactorType::InnerIterator j_it(k_it);
52c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath          typename FactorType::InnerIterator kj_it(m_lu, k);
53c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath          while(kj_it && kj_it.index()<=k) ++kj_it;
54c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath          for(++j_it; j_it; )
55c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath          {
56c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath            if(kj_it.index()==j_it.index())
57c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath            {
58c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath              j_it.valueRef() -= k_it.value() * kj_it.value();
59c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath              ++j_it;
60c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath              ++kj_it;
61c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath            }
62c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath            else if(kj_it.index()<j_it.index()) ++kj_it;
63c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath            else                                ++j_it;
64c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath          }
65c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath        }
66c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath        if(k_it && k_it.index()==i) diag(i) = k_it.value();
67c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath        else                        diag(i) = 1;
68c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath      }
69c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath      m_isInitialized = true;
70c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath      return *this;
71c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath    }
72c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath
73c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath    template<typename Rhs, typename Dest>
74c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath    void _solve(const Rhs& b, Dest& x) const
75c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath    {
76c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath      x = m_lu.template triangularView<UnitLower>().solve(b);
77c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath      x = m_lu.template triangularView<Upper>().solve(x);
78c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath    }
79c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath
80c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath    template<typename Rhs> inline const internal::solve_retval<IncompleteLU, Rhs>
81c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath    solve(const MatrixBase<Rhs>& b) const
82c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath    {
83c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath      eigen_assert(m_isInitialized && "IncompleteLU is not initialized.");
84c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath      eigen_assert(cols()==b.rows()
85c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath                && "IncompleteLU::solve(): invalid number of rows of the right hand side matrix b");
86c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath      return internal::solve_retval<IncompleteLU, Rhs>(*this, b.derived());
87c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath    }
88c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath
89c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath  protected:
90c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath    FactorType m_lu;
91c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath    bool m_isInitialized;
92c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath};
93c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath
94c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamathnamespace internal {
95c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath
96c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamathtemplate<typename _MatrixType, typename Rhs>
97c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamathstruct solve_retval<IncompleteLU<_MatrixType>, Rhs>
98c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath  : solve_retval_base<IncompleteLU<_MatrixType>, Rhs>
99c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath{
100c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath  typedef IncompleteLU<_MatrixType> Dec;
101c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath  EIGEN_MAKE_SOLVE_HELPERS(Dec,Rhs)
102c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath
103c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath  template<typename Dest> void evalTo(Dest& dst) const
104c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath  {
105c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath    dec()._solve(rhs(),dst);
106c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath  }
107c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath};
108c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath
109c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath} // end namespace internal
110c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath
111c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath} // end namespace Eigen
112c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath
113c981c48f5bc9aefeffc0bcb0cc3934c2fae179ddNarayan Kamath#endif // EIGEN_INCOMPLETE_LU_H
114