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  * This is a port of curve_fitting.cc to the minimal C API for Ceres.
32  */
33 
34 #include <math.h>
35 #include <stdio.h>
36 #include <string.h>  // For NULL
37 #include "ceres/c_api.h"
38 
39 /* Data generated using the following octave code.
40  *
41  *   randn('seed', 23497);
42  *   m = 0.3;
43  *   c = 0.1;
44  *   x=[0:0.075:5];
45  *   y = exp(m * x + c);
46  *   noise = randn(size(x)) * 0.2;
47  *   y_observed = y + noise;
48  *   data = [x', y_observed'];
49  *
50  */
51 
52 int num_observations = 67;
53 double data[] = {
54   0.000000e+00, 1.133898e+00,
55   7.500000e-02, 1.334902e+00,
56   1.500000e-01, 1.213546e+00,
57   2.250000e-01, 1.252016e+00,
58   3.000000e-01, 1.392265e+00,
59   3.750000e-01, 1.314458e+00,
60   4.500000e-01, 1.472541e+00,
61   5.250000e-01, 1.536218e+00,
62   6.000000e-01, 1.355679e+00,
63   6.750000e-01, 1.463566e+00,
64   7.500000e-01, 1.490201e+00,
65   8.250000e-01, 1.658699e+00,
66   9.000000e-01, 1.067574e+00,
67   9.750000e-01, 1.464629e+00,
68   1.050000e+00, 1.402653e+00,
69   1.125000e+00, 1.713141e+00,
70   1.200000e+00, 1.527021e+00,
71   1.275000e+00, 1.702632e+00,
72   1.350000e+00, 1.423899e+00,
73   1.425000e+00, 1.543078e+00,
74   1.500000e+00, 1.664015e+00,
75   1.575000e+00, 1.732484e+00,
76   1.650000e+00, 1.543296e+00,
77   1.725000e+00, 1.959523e+00,
78   1.800000e+00, 1.685132e+00,
79   1.875000e+00, 1.951791e+00,
80   1.950000e+00, 2.095346e+00,
81   2.025000e+00, 2.361460e+00,
82   2.100000e+00, 2.169119e+00,
83   2.175000e+00, 2.061745e+00,
84   2.250000e+00, 2.178641e+00,
85   2.325000e+00, 2.104346e+00,
86   2.400000e+00, 2.584470e+00,
87   2.475000e+00, 1.914158e+00,
88   2.550000e+00, 2.368375e+00,
89   2.625000e+00, 2.686125e+00,
90   2.700000e+00, 2.712395e+00,
91   2.775000e+00, 2.499511e+00,
92   2.850000e+00, 2.558897e+00,
93   2.925000e+00, 2.309154e+00,
94   3.000000e+00, 2.869503e+00,
95   3.075000e+00, 3.116645e+00,
96   3.150000e+00, 3.094907e+00,
97   3.225000e+00, 2.471759e+00,
98   3.300000e+00, 3.017131e+00,
99   3.375000e+00, 3.232381e+00,
100   3.450000e+00, 2.944596e+00,
101   3.525000e+00, 3.385343e+00,
102   3.600000e+00, 3.199826e+00,
103   3.675000e+00, 3.423039e+00,
104   3.750000e+00, 3.621552e+00,
105   3.825000e+00, 3.559255e+00,
106   3.900000e+00, 3.530713e+00,
107   3.975000e+00, 3.561766e+00,
108   4.050000e+00, 3.544574e+00,
109   4.125000e+00, 3.867945e+00,
110   4.200000e+00, 4.049776e+00,
111   4.275000e+00, 3.885601e+00,
112   4.350000e+00, 4.110505e+00,
113   4.425000e+00, 4.345320e+00,
114   4.500000e+00, 4.161241e+00,
115   4.575000e+00, 4.363407e+00,
116   4.650000e+00, 4.161576e+00,
117   4.725000e+00, 4.619728e+00,
118   4.800000e+00, 4.737410e+00,
119   4.875000e+00, 4.727863e+00,
120   4.950000e+00, 4.669206e+00,
121 };
122 
123 /* This is the equivalent of a use-defined CostFunction in the C++ Ceres API.
124  * This is passed as a callback to the Ceres C API, which internally converts
125  * the callback into a CostFunction. */
exponential_residual(void * user_data,double ** parameters,double * residuals,double ** jacobians)126 int exponential_residual(void* user_data,
127                          double** parameters,
128                          double* residuals,
129                          double** jacobians) {
130   double* measurement = (double*) user_data;
131   double x = measurement[0];
132   double y = measurement[1];
133   double m = parameters[0][0];
134   double c = parameters[1][0];
135 
136   residuals[0] = y - exp(m * x + c);
137   if (jacobians == NULL) {
138     return 1;
139   }
140   if (jacobians[0] != NULL) {
141     jacobians[0][0] = - x * exp(m * x + c);  /* dr/dm */
142   }
143   if (jacobians[1] != NULL) {
144     jacobians[1][0] =     - exp(m * x + c);  /* dr/dc */
145   }
146   return 1;
147 }
148 
main(int argc,char ** argv)149 int main(int argc, char** argv) {
150   /* Note: Typically it is better to compact m and c into one block,
151    * but in this case use separate blocks to illustrate the use of
152    * multiple parameter blocks. */
153   double m = 0.0;
154   double c = 0.0;
155 
156   double *parameter_pointers[] = { &m, &c };
157   int parameter_sizes[] = { 1, 1 };
158   int i;
159 
160   ceres_problem_t* problem;
161 
162   /* Ceres has some internal stuff that needs to get initialized. */
163   ceres_init();
164 
165   problem = ceres_create_problem();
166 
167   /* Add all the residuals. */
168   for (i = 0; i < num_observations; ++i) {
169     ceres_problem_add_residual_block(
170         problem,
171         exponential_residual,  /* Cost function */
172         &data[2 * i],          /* Points to the (x,y) measurement */
173         NULL,                  /* No loss function */
174         NULL,                  /* No loss function user data */
175         1,                     /* Number of residuals */
176         2,                     /* Number of parameter blocks */
177         parameter_sizes,
178         parameter_pointers);
179   }
180 
181   ceres_solve(problem);
182   ceres_free_problem(problem);
183 
184   printf("Initial m: 0.0, c: 0.0\n");
185   printf("Final m: %g, c: %g\n", m, c);
186   return 0;
187 }
188