• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 // Ceres Solver - A fast non-linear least squares minimizer
2 // Copyright 2013 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: mierle@gmail.com (Keir Mierle)
30 
31 #include "ceres/c_api.h"
32 
33 #include <cmath>
34 
35 #include "glog/logging.h"
36 #include "gtest/gtest.h"
37 
38 // Duplicated from curve_fitting.cc.
39 int num_observations = 67;
40 double data[] = {
41   0.000000e+00, 1.133898e+00,
42   7.500000e-02, 1.334902e+00,
43   1.500000e-01, 1.213546e+00,
44   2.250000e-01, 1.252016e+00,
45   3.000000e-01, 1.392265e+00,
46   3.750000e-01, 1.314458e+00,
47   4.500000e-01, 1.472541e+00,
48   5.250000e-01, 1.536218e+00,
49   6.000000e-01, 1.355679e+00,
50   6.750000e-01, 1.463566e+00,
51   7.500000e-01, 1.490201e+00,
52   8.250000e-01, 1.658699e+00,
53   9.000000e-01, 1.067574e+00,
54   9.750000e-01, 1.464629e+00,
55   1.050000e+00, 1.402653e+00,
56   1.125000e+00, 1.713141e+00,
57   1.200000e+00, 1.527021e+00,
58   1.275000e+00, 1.702632e+00,
59   1.350000e+00, 1.423899e+00,
60   1.425000e+00, 1.543078e+00,
61   1.500000e+00, 1.664015e+00,
62   1.575000e+00, 1.732484e+00,
63   1.650000e+00, 1.543296e+00,
64   1.725000e+00, 1.959523e+00,
65   1.800000e+00, 1.685132e+00,
66   1.875000e+00, 1.951791e+00,
67   1.950000e+00, 2.095346e+00,
68   2.025000e+00, 2.361460e+00,
69   2.100000e+00, 2.169119e+00,
70   2.175000e+00, 2.061745e+00,
71   2.250000e+00, 2.178641e+00,
72   2.325000e+00, 2.104346e+00,
73   2.400000e+00, 2.584470e+00,
74   2.475000e+00, 1.914158e+00,
75   2.550000e+00, 2.368375e+00,
76   2.625000e+00, 2.686125e+00,
77   2.700000e+00, 2.712395e+00,
78   2.775000e+00, 2.499511e+00,
79   2.850000e+00, 2.558897e+00,
80   2.925000e+00, 2.309154e+00,
81   3.000000e+00, 2.869503e+00,
82   3.075000e+00, 3.116645e+00,
83   3.150000e+00, 3.094907e+00,
84   3.225000e+00, 2.471759e+00,
85   3.300000e+00, 3.017131e+00,
86   3.375000e+00, 3.232381e+00,
87   3.450000e+00, 2.944596e+00,
88   3.525000e+00, 3.385343e+00,
89   3.600000e+00, 3.199826e+00,
90   3.675000e+00, 3.423039e+00,
91   3.750000e+00, 3.621552e+00,
92   3.825000e+00, 3.559255e+00,
93   3.900000e+00, 3.530713e+00,
94   3.975000e+00, 3.561766e+00,
95   4.050000e+00, 3.544574e+00,
96   4.125000e+00, 3.867945e+00,
97   4.200000e+00, 4.049776e+00,
98   4.275000e+00, 3.885601e+00,
99   4.350000e+00, 4.110505e+00,
100   4.425000e+00, 4.345320e+00,
101   4.500000e+00, 4.161241e+00,
102   4.575000e+00, 4.363407e+00,
103   4.650000e+00, 4.161576e+00,
104   4.725000e+00, 4.619728e+00,
105   4.800000e+00, 4.737410e+00,
106   4.875000e+00, 4.727863e+00,
107   4.950000e+00, 4.669206e+00,
108 };
109 
110 // A test cost function, similar to the one in curve_fitting.c.
exponential_residual(void * user_data,double ** parameters,double * residuals,double ** jacobians)111 int exponential_residual(void* user_data,
112                          double** parameters,
113                          double* residuals,
114                          double** jacobians) {
115   double* measurement = (double*) user_data;
116   double x = measurement[0];
117   double y = measurement[1];
118   double m = parameters[0][0];
119   double c = parameters[1][0];
120 
121   residuals[0] = y - exp(m * x + c);
122   if (jacobians == NULL) {
123     return 1;
124   }
125   if (jacobians[0] != NULL) {
126     jacobians[0][0] = - x * exp(m * x + c);  // dr/dm
127   }
128   if (jacobians[1] != NULL) {
129     jacobians[1][0] =     - exp(m * x + c);  // dr/dc
130   }
131   return 1;
132 }
133 
134 namespace ceres {
135 namespace internal {
136 
TEST(C_API,SimpleEndToEndTest)137 TEST(C_API, SimpleEndToEndTest) {
138   double m = 0.0;
139   double c = 0.0;
140   double *parameter_pointers[] = { &m, &c };
141   int parameter_sizes[] = { 1, 1 };
142 
143   ceres_problem_t* problem = ceres_create_problem();
144   for (int i = 0; i < num_observations; ++i) {
145     ceres_problem_add_residual_block(
146         problem,
147         exponential_residual,  // Cost function
148         &data[2 * i],          // Points to the (x,y) measurement
149         NULL,                  // Loss function
150         NULL,                  // Loss function user data
151         1,                     // Number of residuals
152         2,                     // Number of parameter blocks
153         parameter_sizes,
154         parameter_pointers);
155   }
156 
157   ceres_solve(problem);
158 
159   EXPECT_NEAR(0.3, m, 0.02);
160   EXPECT_NEAR(0.1, c, 0.04);
161 
162   ceres_free_problem(problem);
163 }
164 
165 template<typename T>
166 class ScopedSetValue {
167  public:
ScopedSetValue(T * variable,T new_value)168   ScopedSetValue(T* variable, T new_value)
169       : variable_(variable), old_value_(*variable) {
170     *variable = new_value;
171   }
~ScopedSetValue()172   ~ScopedSetValue() {
173     *variable_ = old_value_;
174   }
175 
176  private:
177   T* variable_;
178   T old_value_;
179 };
180 
TEST(C_API,LossFunctions)181 TEST(C_API, LossFunctions) {
182   double m = 0.2;
183   double c = 0.03;
184   double *parameter_pointers[] = { &m, &c };
185   int parameter_sizes[] = { 1, 1 };
186 
187   // Create two outliers, but be careful to leave the data intact.
188   ScopedSetValue<double> outlier1x(&data[12], 2.5);
189   ScopedSetValue<double> outlier1y(&data[13], 1.0e3);
190   ScopedSetValue<double> outlier2x(&data[14], 3.2);
191   ScopedSetValue<double> outlier2y(&data[15], 30e3);
192 
193   // Create a cauchy cost function, and reuse it many times.
194   void* cauchy_loss_data =
195       ceres_create_cauchy_loss_function_data(5.0);
196 
197   ceres_problem_t* problem = ceres_create_problem();
198   for (int i = 0; i < num_observations; ++i) {
199     ceres_problem_add_residual_block(
200         problem,
201         exponential_residual,  // Cost function
202         &data[2 * i],          // Points to the (x,y) measurement
203         ceres_stock_loss_function,
204         cauchy_loss_data,      // Loss function user data
205         1,                     // Number of residuals
206         2,                     // Number of parameter blocks
207         parameter_sizes,
208         parameter_pointers);
209   }
210 
211   ceres_solve(problem);
212 
213   EXPECT_NEAR(0.3, m, 0.02);
214   EXPECT_NEAR(0.1, c, 0.04);
215 
216   ceres_free_stock_loss_function_data(cauchy_loss_data);
217   ceres_free_problem(problem);
218 }
219 
220 }  // namespace internal
221 }  // namespace ceres
222