• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 // This file is part of Eigen, a lightweight C++ template library
2 // for linear algebra.
3 //
4 // Copyright (C) 2011-2014 Gael Guennebaud <gael.guennebaud@inria.fr>
5 // Copyright (C) 2012 Désiré Nuentsa-Wakam <desire.nuentsa_wakam@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 #ifndef EIGEN_BICGSTAB_H
12 #define EIGEN_BICGSTAB_H
13 
14 namespace Eigen {
15 
16 namespace internal {
17 
18 /** \internal Low-level bi conjugate gradient stabilized algorithm
19   * \param mat The matrix A
20   * \param rhs The right hand side vector b
21   * \param x On input and initial solution, on output the computed solution.
22   * \param precond A preconditioner being able to efficiently solve for an
23   *                approximation of Ax=b (regardless of b)
24   * \param iters On input the max number of iteration, on output the number of performed iterations.
25   * \param tol_error On input the tolerance error, on output an estimation of the relative error.
26   * \return false in the case of numerical issue, for example a break down of BiCGSTAB.
27   */
28 template<typename MatrixType, typename Rhs, typename Dest, typename Preconditioner>
bicgstab(const MatrixType & mat,const Rhs & rhs,Dest & x,const Preconditioner & precond,Index & iters,typename Dest::RealScalar & tol_error)29 bool bicgstab(const MatrixType& mat, const Rhs& rhs, Dest& x,
30               const Preconditioner& precond, Index& iters,
31               typename Dest::RealScalar& tol_error)
32 {
33   using std::sqrt;
34   using std::abs;
35   typedef typename Dest::RealScalar RealScalar;
36   typedef typename Dest::Scalar Scalar;
37   typedef Matrix<Scalar,Dynamic,1> VectorType;
38   RealScalar tol = tol_error;
39   Index maxIters = iters;
40 
41   Index n = mat.cols();
42   VectorType r  = rhs - mat * x;
43   VectorType r0 = r;
44 
45   RealScalar r0_sqnorm = r0.squaredNorm();
46   RealScalar rhs_sqnorm = rhs.squaredNorm();
47   if(rhs_sqnorm == 0)
48   {
49     x.setZero();
50     return true;
51   }
52   Scalar rho    = 1;
53   Scalar alpha  = 1;
54   Scalar w      = 1;
55 
56   VectorType v = VectorType::Zero(n), p = VectorType::Zero(n);
57   VectorType y(n),  z(n);
58   VectorType kt(n), ks(n);
59 
60   VectorType s(n), t(n);
61 
62   RealScalar tol2 = tol*tol*rhs_sqnorm;
63   RealScalar eps2 = NumTraits<Scalar>::epsilon()*NumTraits<Scalar>::epsilon();
64   Index i = 0;
65   Index restarts = 0;
66 
67   while ( r.squaredNorm() > tol2 && i<maxIters )
68   {
69     Scalar rho_old = rho;
70 
71     rho = r0.dot(r);
72     if (abs(rho) < eps2*r0_sqnorm)
73     {
74       // The new residual vector became too orthogonal to the arbitrarily chosen direction r0
75       // Let's restart with a new r0:
76       r  = rhs - mat * x;
77       r0 = r;
78       rho = r0_sqnorm = r.squaredNorm();
79       if(restarts++ == 0)
80         i = 0;
81     }
82     Scalar beta = (rho/rho_old) * (alpha / w);
83     p = r + beta * (p - w * v);
84 
85     y = precond.solve(p);
86 
87     v.noalias() = mat * y;
88 
89     alpha = rho / r0.dot(v);
90     s = r - alpha * v;
91 
92     z = precond.solve(s);
93     t.noalias() = mat * z;
94 
95     RealScalar tmp = t.squaredNorm();
96     if(tmp>RealScalar(0))
97       w = t.dot(s) / tmp;
98     else
99       w = Scalar(0);
100     x += alpha * y + w * z;
101     r = s - w * t;
102     ++i;
103   }
104   tol_error = sqrt(r.squaredNorm()/rhs_sqnorm);
105   iters = i;
106   return true;
107 }
108 
109 }
110 
111 template< typename _MatrixType,
112           typename _Preconditioner = DiagonalPreconditioner<typename _MatrixType::Scalar> >
113 class BiCGSTAB;
114 
115 namespace internal {
116 
117 template< typename _MatrixType, typename _Preconditioner>
118 struct traits<BiCGSTAB<_MatrixType,_Preconditioner> >
119 {
120   typedef _MatrixType MatrixType;
121   typedef _Preconditioner Preconditioner;
122 };
123 
124 }
125 
126 /** \ingroup IterativeLinearSolvers_Module
127   * \brief A bi conjugate gradient stabilized solver for sparse square problems
128   *
129   * This class allows to solve for A.x = b sparse linear problems using a bi conjugate gradient
130   * stabilized algorithm. The vectors x and b can be either dense or sparse.
131   *
132   * \tparam _MatrixType the type of the sparse matrix A, can be a dense or a sparse matrix.
133   * \tparam _Preconditioner the type of the preconditioner. Default is DiagonalPreconditioner
134   *
135   * \implsparsesolverconcept
136   *
137   * The maximal number of iterations and tolerance value can be controlled via the setMaxIterations()
138   * and setTolerance() methods. The defaults are the size of the problem for the maximal number of iterations
139   * and NumTraits<Scalar>::epsilon() for the tolerance.
140   *
141   * The tolerance corresponds to the relative residual error: |Ax-b|/|b|
142   *
143   * \b Performance: when using sparse matrices, best performance is achied for a row-major sparse matrix format.
144   * Moreover, in this case multi-threading can be exploited if the user code is compiled with OpenMP enabled.
145   * See \ref TopicMultiThreading for details.
146   *
147   * This class can be used as the direct solver classes. Here is a typical usage example:
148   * \include BiCGSTAB_simple.cpp
149   *
150   * By default the iterations start with x=0 as an initial guess of the solution.
151   * One can control the start using the solveWithGuess() method.
152   *
153   * BiCGSTAB can also be used in a matrix-free context, see the following \link MatrixfreeSolverExample example \endlink.
154   *
155   * \sa class SimplicialCholesky, DiagonalPreconditioner, IdentityPreconditioner
156   */
157 template< typename _MatrixType, typename _Preconditioner>
158 class BiCGSTAB : public IterativeSolverBase<BiCGSTAB<_MatrixType,_Preconditioner> >
159 {
160   typedef IterativeSolverBase<BiCGSTAB> Base;
161   using Base::matrix;
162   using Base::m_error;
163   using Base::m_iterations;
164   using Base::m_info;
165   using Base::m_isInitialized;
166 public:
167   typedef _MatrixType MatrixType;
168   typedef typename MatrixType::Scalar Scalar;
169   typedef typename MatrixType::RealScalar RealScalar;
170   typedef _Preconditioner Preconditioner;
171 
172 public:
173 
174   /** Default constructor. */
175   BiCGSTAB() : Base() {}
176 
177   /** Initialize the solver with matrix \a A for further \c Ax=b solving.
178     *
179     * This constructor is a shortcut for the default constructor followed
180     * by a call to compute().
181     *
182     * \warning this class stores a reference to the matrix A as well as some
183     * precomputed values that depend on it. Therefore, if \a A is changed
184     * this class becomes invalid. Call compute() to update it with the new
185     * matrix A, or modify a copy of A.
186     */
187   template<typename MatrixDerived>
188   explicit BiCGSTAB(const EigenBase<MatrixDerived>& A) : Base(A.derived()) {}
189 
190   ~BiCGSTAB() {}
191 
192   /** \internal */
193   template<typename Rhs,typename Dest>
194   void _solve_with_guess_impl(const Rhs& b, Dest& x) const
195   {
196     bool failed = false;
197     for(Index j=0; j<b.cols(); ++j)
198     {
199       m_iterations = Base::maxIterations();
200       m_error = Base::m_tolerance;
201 
202       typename Dest::ColXpr xj(x,j);
203       if(!internal::bicgstab(matrix(), b.col(j), xj, Base::m_preconditioner, m_iterations, m_error))
204         failed = true;
205     }
206     m_info = failed ? NumericalIssue
207            : m_error <= Base::m_tolerance ? Success
208            : NoConvergence;
209     m_isInitialized = true;
210   }
211 
212   /** \internal */
213   using Base::_solve_impl;
214   template<typename Rhs,typename Dest>
215   void _solve_impl(const MatrixBase<Rhs>& b, Dest& x) const
216   {
217     x.resize(this->rows(),b.cols());
218     x.setZero();
219     _solve_with_guess_impl(b,x);
220   }
221 
222 protected:
223 
224 };
225 
226 } // end namespace Eigen
227 
228 #endif // EIGEN_BICGSTAB_H
229