• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 // Ceres Solver - A fast non-linear least squares minimizer
2 // Copyright 2012 Google Inc. All rights reserved.
3 // http://code.google.com/p/ceres-solver/
4 //
5 // Redistribution and use in source and binary forms, with or without
6 // modification, are permitted provided that the following conditions are met:
7 //
8 // * Redistributions of source code must retain the above copyright notice,
9 //   this list of conditions and the following disclaimer.
10 // * Redistributions in binary form must reproduce the above copyright notice,
11 //   this list of conditions and the following disclaimer in the documentation
12 //   and/or other materials provided with the distribution.
13 // * Neither the name of Google Inc. nor the names of its contributors may be
14 //   used to endorse or promote products derived from this software without
15 //   specific prior written permission.
16 //
17 // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
18 // AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19 // IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20 // ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
21 // LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
22 // CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
23 // SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
24 // INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
25 // CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
26 // ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
27 // POSSIBILITY OF SUCH DAMAGE.
28 //
29 // Author: sameeragarwal@google.com (Sameer Agarwal)
30 
31 #ifndef CERES_INTERNAL_DOGLEG_STRATEGY_H_
32 #define CERES_INTERNAL_DOGLEG_STRATEGY_H_
33 
34 #include "ceres/linear_solver.h"
35 #include "ceres/trust_region_strategy.h"
36 
37 namespace ceres {
38 namespace internal {
39 
40 // Dogleg step computation and trust region sizing strategy based on
41 // on "Methods for Nonlinear Least Squares" by K. Madsen, H.B. Nielsen
42 // and O. Tingleff. Available to download from
43 //
44 // http://www2.imm.dtu.dk/pubdb/views/edoc_download.php/3215/pdf/imm3215.pdf
45 //
46 // One minor modification is that instead of computing the pure
47 // Gauss-Newton step, we compute a regularized version of it. This is
48 // because the Jacobian is often rank-deficient and in such cases
49 // using a direct solver leads to numerical failure.
50 //
51 // If SUBSPACE is passed as the type argument to the constructor, the
52 // DoglegStrategy follows the approach by Shultz, Schnabel, Byrd.
53 // This finds the exact optimum over the two-dimensional subspace
54 // spanned by the two Dogleg vectors.
55 class DoglegStrategy : public TrustRegionStrategy {
56 public:
57   DoglegStrategy(const TrustRegionStrategy::Options& options);
~DoglegStrategy()58   virtual ~DoglegStrategy() {}
59 
60   // TrustRegionStrategy interface
61   virtual Summary ComputeStep(const PerSolveOptions& per_solve_options,
62                               SparseMatrix* jacobian,
63                               const double* residuals,
64                               double* step);
65   virtual void StepAccepted(double step_quality);
66   virtual void StepRejected(double step_quality);
67   virtual void StepIsInvalid();
68 
69   virtual double Radius() const;
70 
71   // These functions are predominantly for testing.
gradient()72   Vector gradient() const { return gradient_; }
gauss_newton_step()73   Vector gauss_newton_step() const { return gauss_newton_step_; }
subspace_basis()74   Matrix subspace_basis() const { return subspace_basis_; }
subspace_g()75   Vector subspace_g() const { return subspace_g_; }
subspace_B()76   Matrix subspace_B() const { return subspace_B_; }
77 
78  private:
79   typedef Eigen::Matrix<double, 2, 1, Eigen::DontAlign> Vector2d;
80   typedef Eigen::Matrix<double, 2, 2, Eigen::DontAlign> Matrix2d;
81 
82   LinearSolver::Summary ComputeGaussNewtonStep(SparseMatrix* jacobian,
83                                                const double* residuals);
84   void ComputeCauchyPoint(SparseMatrix* jacobian);
85   void ComputeGradient(SparseMatrix* jacobian, const double* residuals);
86   void ComputeTraditionalDoglegStep(double* step);
87   bool ComputeSubspaceModel(SparseMatrix* jacobian);
88   void ComputeSubspaceDoglegStep(double* step);
89 
90   bool FindMinimumOnTrustRegionBoundary(Vector2d* minimum) const;
91   Vector MakePolynomialForBoundaryConstrainedProblem() const;
92   Vector2d ComputeSubspaceStepFromRoot(double lambda) const;
93   double EvaluateSubspaceModel(const Vector2d& x) const;
94 
95   LinearSolver* linear_solver_;
96   double radius_;
97   const double max_radius_;
98 
99   const double min_diagonal_;
100   const double max_diagonal_;
101 
102   // mu is used to scale the diagonal matrix used to make the
103   // Gauss-Newton solve full rank. In each solve, the strategy starts
104   // out with mu = min_mu, and tries values upto max_mu. If the user
105   // reports an invalid step, the value of mu_ is increased so that
106   // the next solve starts with a stronger regularization.
107   //
108   // If a successful step is reported, then the value of mu_ is
109   // decreased with a lower bound of min_mu_.
110   double mu_;
111   const double min_mu_;
112   const double max_mu_;
113   const double mu_increase_factor_;
114   const double increase_threshold_;
115   const double decrease_threshold_;
116 
117   Vector diagonal_;  // sqrt(diag(J^T J))
118   Vector lm_diagonal_;
119 
120   Vector gradient_;
121   Vector gauss_newton_step_;
122 
123   // cauchy_step = alpha * gradient
124   double alpha_;
125   double dogleg_step_norm_;
126 
127   // When, ComputeStep is called, reuse_ indicates whether the
128   // Gauss-Newton and Cauchy steps from the last call to ComputeStep
129   // can be reused or not.
130   //
131   // If the user called StepAccepted, then it is expected that the
132   // user has recomputed the Jacobian matrix and new Gauss-Newton
133   // solve is needed and reuse is set to false.
134   //
135   // If the user called StepRejected, then it is expected that the
136   // user wants to solve the trust region problem with the same matrix
137   // but a different trust region radius and the Gauss-Newton and
138   // Cauchy steps can be reused to compute the Dogleg, thus reuse is
139   // set to true.
140   //
141   // If the user called StepIsInvalid, then there was a numerical
142   // problem with the step computed in the last call to ComputeStep,
143   // and the regularization used to do the Gauss-Newton solve is
144   // increased and a new solve should be done when ComputeStep is
145   // called again, thus reuse is set to false.
146   bool reuse_;
147 
148   // The dogleg type determines how the minimum of the local
149   // quadratic model is found.
150   DoglegType dogleg_type_;
151 
152   // If the type is SUBSPACE_DOGLEG, the two-dimensional
153   // model 1/2 x^T B x + g^T x has to be computed and stored.
154   bool subspace_is_one_dimensional_;
155   Matrix subspace_basis_;
156   Vector2d subspace_g_;
157   Matrix2d subspace_B_;
158 };
159 
160 }  // namespace internal
161 }  // namespace ceres
162 
163 #endif  // CERES_INTERNAL_DOGLEG_STRATEGY_H_
164