blob: 649f3f714c265023c4b4b5512daf12758230ee12 [file] [log] [blame]
Keir Mierle8ebb0732012-04-30 23:09:08 -07001// Ceres Solver - A fast non-linear least squares minimizer
2// Copyright 2010, 2011, 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: keir@google.com (Keir Mierle)
30// sameeragarwal@google.com (Sameer Agarwal)
31
32#include "ceres/residual_block.h"
33
34#include <algorithm>
35#include <cstddef>
36#include <vector>
37
Sameer Agarwal487250e2013-04-05 14:20:37 -070038#include "ceres/blas.h"
Keir Mierle8ebb0732012-04-30 23:09:08 -070039#include "ceres/corrector.h"
40#include "ceres/parameter_block.h"
41#include "ceres/residual_block_utils.h"
42#include "ceres/cost_function.h"
43#include "ceres/internal/eigen.h"
44#include "ceres/internal/fixed_array.h"
45#include "ceres/local_parameterization.h"
46#include "ceres/loss_function.h"
47
Sameer Agarwale6707b22013-04-16 15:44:23 -070048using Eigen::Dynamic;
49
Keir Mierle8ebb0732012-04-30 23:09:08 -070050namespace ceres {
51namespace internal {
52
53ResidualBlock::ResidualBlock(const CostFunction* cost_function,
54 const LossFunction* loss_function,
Keir Mierle04938ef2013-02-17 12:37:55 -080055 const vector<ParameterBlock*>& parameter_blocks,
56 int index)
Keir Mierle8ebb0732012-04-30 23:09:08 -070057 : cost_function_(cost_function),
58 loss_function_(loss_function),
59 parameter_blocks_(
60 new ParameterBlock* [
Keir Mierle04938ef2013-02-17 12:37:55 -080061 cost_function->parameter_block_sizes().size()]),
62 index_(index) {
Keir Mierle8ebb0732012-04-30 23:09:08 -070063 std::copy(parameter_blocks.begin(),
64 parameter_blocks.end(),
65 parameter_blocks_.get());
66}
67
Sameer Agarwal039ff072013-02-26 09:15:39 -080068bool ResidualBlock::Evaluate(const bool apply_loss_function,
69 double* cost,
Keir Mierle8ebb0732012-04-30 23:09:08 -070070 double* residuals,
71 double** jacobians,
72 double* scratch) const {
73 const int num_parameter_blocks = NumParameterBlocks();
74 const int num_residuals = cost_function_->num_residuals();
75
76 // Collect the parameters from their blocks. This will rarely allocate, since
77 // residuals taking more than 8 parameter block arguments are rare.
78 FixedArray<const double*, 8> parameters(num_parameter_blocks);
79 for (int i = 0; i < num_parameter_blocks; ++i) {
80 parameters[i] = parameter_blocks_[i]->state();
81 }
82
83 // Put pointers into the scratch space into global_jacobians as appropriate.
84 FixedArray<double*, 8> global_jacobians(num_parameter_blocks);
85 if (jacobians != NULL) {
86 for (int i = 0; i < num_parameter_blocks; ++i) {
87 const ParameterBlock* parameter_block = parameter_blocks_[i];
88 if (jacobians[i] != NULL &&
89 parameter_block->LocalParameterizationJacobian() != NULL) {
90 global_jacobians[i] = scratch;
91 scratch += num_residuals * parameter_block->Size();
92 } else {
93 global_jacobians[i] = jacobians[i];
94 }
95 }
96 }
97
98 // If the caller didn't request residuals, use the scratch space for them.
99 bool outputting_residuals = (residuals != NULL);
100 if (!outputting_residuals) {
101 residuals = scratch;
102 }
103
104 // Invalidate the evaluation buffers so that we can check them after
105 // the CostFunction::Evaluate call, to see if all the return values
106 // that were required were written to and that they are finite.
107 double** eval_jacobians = (jacobians != NULL) ? global_jacobians.get() : NULL;
108
109 InvalidateEvaluation(*this, cost, residuals, eval_jacobians);
110
Sameer Agarwal552f9f82012-08-31 07:27:22 -0700111 if (!cost_function_->Evaluate(parameters.get(), residuals, eval_jacobians)) {
112 return false;
113 }
114
115 if (!IsEvaluationValid(*this,
Keir Mierle8ebb0732012-04-30 23:09:08 -0700116 parameters.get(),
117 cost,
118 residuals,
119 eval_jacobians)) {
120 string message =
121 "\n\n"
122 "Error in evaluating the ResidualBlock.\n\n"
123 "There are two possible reasons. Either the CostFunction did not evaluate and fill all \n" // NOLINT
124 "residual and jacobians that were requested or there was a non-finite value (nan/infinite)\n" // NOLINT
125 "generated during the or jacobian computation. \n\n" +
126 EvaluationToString(*this,
127 parameters.get(),
128 cost,
129 residuals,
130 eval_jacobians);
131 LOG(WARNING) << message;
132 return false;
133 }
134
135 double squared_norm = VectorRef(residuals, num_residuals).squaredNorm();
136
137 // Update the jacobians with the local parameterizations.
138 if (jacobians != NULL) {
139 for (int i = 0; i < num_parameter_blocks; ++i) {
140 if (jacobians[i] != NULL) {
141 const ParameterBlock* parameter_block = parameter_blocks_[i];
142
143 // Apply local reparameterization to the jacobians.
144 if (parameter_block->LocalParameterizationJacobian() != NULL) {
Sameer Agarwale6707b22013-04-16 15:44:23 -0700145 // jacobians[i] = global_jacobians[i] * global_to_local_jacobian.
146 MatrixMatrixMultiply<Dynamic, Dynamic, Dynamic, Dynamic, 0>(
Sameer Agarwal487250e2013-04-05 14:20:37 -0700147 global_jacobians[i],
148 num_residuals,
149 parameter_block->Size(),
Keir Mierle8ebb0732012-04-30 23:09:08 -0700150 parameter_block->LocalParameterizationJacobian(),
151 parameter_block->Size(),
Sameer Agarwal487250e2013-04-05 14:20:37 -0700152 parameter_block->LocalSize(),
153 jacobians[i], 0, 0, num_residuals, parameter_block->LocalSize());
Keir Mierle8ebb0732012-04-30 23:09:08 -0700154 }
155 }
156 }
157 }
158
Sameer Agarwal039ff072013-02-26 09:15:39 -0800159 if (loss_function_ == NULL || !apply_loss_function) {
Keir Mierle8ebb0732012-04-30 23:09:08 -0700160 *cost = 0.5 * squared_norm;
161 return true;
162 }
163
164 double rho[3];
165 loss_function_->Evaluate(squared_norm, rho);
166 *cost = 0.5 * rho[0];
167
168 // No jacobians and not outputting residuals? All done. Doing an early exit
169 // here avoids constructing the "Corrector" object below in a common case.
170 if (jacobians == NULL && !outputting_residuals) {
171 return true;
172 }
173
174 // Correct for the effects of the loss function. The jacobians need to be
175 // corrected before the residuals, since they use the uncorrected residuals.
176 Corrector correct(squared_norm, rho);
177 if (jacobians != NULL) {
178 for (int i = 0; i < num_parameter_blocks; ++i) {
179 if (jacobians[i] != NULL) {
180 const ParameterBlock* parameter_block = parameter_blocks_[i];
181
182 // Correct the jacobians for the loss function.
183 correct.CorrectJacobian(num_residuals,
184 parameter_block->LocalSize(),
185 residuals,
186 jacobians[i]);
187 }
188 }
189 }
190
191 // Correct the residuals with the loss function.
192 if (outputting_residuals) {
193 correct.CorrectResiduals(num_residuals, residuals);
194 }
195 return true;
196}
197
198int ResidualBlock::NumScratchDoublesForEvaluate() const {
199 // Compute the amount of scratch space needed to store the full-sized
200 // jacobians. For parameters that have no local parameterization no storage
201 // is needed and the passed-in jacobian array is used directly. Also include
202 // space to store the residuals, which is needed for cost-only evaluations.
203 // This is slightly pessimistic, since both won't be needed all the time, but
204 // the amount of excess should not cause problems for the caller.
205 int num_parameters = NumParameterBlocks();
206 int scratch_doubles = 1;
207 for (int i = 0; i < num_parameters; ++i) {
208 const ParameterBlock* parameter_block = parameter_blocks_[i];
209 if (!parameter_block->IsConstant() &&
210 parameter_block->LocalParameterizationJacobian() != NULL) {
211 scratch_doubles += parameter_block->Size();
212 }
213 }
214 scratch_doubles *= NumResiduals();
215 return scratch_doubles;
216}
217
218} // namespace internal
219} // namespace ceres