1 /* Copyright (c) 2007-2014 Massachusetts Institute of Technology
3 * Permission is hereby granted, free of charge, to any person obtaining
4 * a copy of this software and associated documentation files (the
5 * "Software"), to deal in the Software without restriction, including
6 * without limitation the rights to use, copy, modify, merge, publish,
7 * distribute, sublicense, and/or sell copies of the Software, and to
8 * permit persons to whom the Software is furnished to do so, subject to
9 * the following conditions:
11 * The above copyright notice and this permission notice shall be
12 * included in all copies or substantial portions of the Software.
14 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
15 * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
16 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
17 * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE
18 * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
19 * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
20 * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
23 #include <octave/oct.h>
24 #include <octave/oct-map.h>
25 #include <octave/ov.h>
30 #include "nlopt_optimize_usage.h"
32 static int struct_val_default(Octave_map &m, const std::string& k,
36 if (m.contents(k).length() == 1 && (m.contents(k))(0).is_real_scalar())
37 return (m.contents(k))(0).int_value();
42 static double struct_val_default(Octave_map &m, const std::string& k,
46 if (m.contents(k).length() == 1 && (m.contents(k))(0).is_real_scalar())
47 return (m.contents(k))(0).double_value();
52 static Matrix struct_val_default(Octave_map &m, const std::string& k,
56 if ((m.contents(k)).length() == 1) {
57 if ((m.contents(k))(0).is_real_scalar())
58 return Matrix(1, dflt.length(), (m.contents(k))(0).double_value());
59 else if ((m.contents(k))(0).is_real_matrix())
60 return (m.contents(k))(0).matrix_value();
72 static double user_function(unsigned n, const double *x,
73 double *gradient, /* NULL if not needed */
76 user_function_data *data = (user_function_data *) data_;
77 octave_value_list args(1, 0);
79 for (unsigned i = 0; i < n; ++i)
82 octave_value_list res = data->f->do_multi_index_op(gradient ? 2 : 1, args);
83 if (res.length() < (gradient ? 2 : 1))
84 gripe_user_supplied_eval("nlopt_optimize");
85 else if (!res(0).is_real_scalar()
86 || (gradient && !res(1).is_real_matrix()
87 && !(n == 1 && res(1).is_real_scalar())))
88 gripe_user_returned_invalid("nlopt_optimize");
91 if (n == 1 && res(1).is_real_scalar())
92 gradient[0] = res(1).double_value();
94 Matrix grad = res(1).matrix_value();
95 for (unsigned i = 0; i < n; ++i)
96 gradient[i] = grad(i);
100 if (data->verbose) printf("nlopt_optimize eval #%d: %g\n",
101 data->neval, res(0).double_value());
102 double f = res(0).double_value();
103 if (f != f /* isnan(f) */) nlopt_force_stop(data->opt);
109 static double user_function1(unsigned n, const double *x,
110 double *gradient, /* NULL if not needed */
113 octave_function *f = (octave_function *) data_;
114 octave_value_list args(1, 0);
116 for (unsigned i = 0; i < n; ++i)
119 octave_value_list res = f->do_multi_index_op(gradient ? 2 : 1, args);
120 if (res.length() < (gradient ? 2 : 1))
121 gripe_user_supplied_eval("nlopt_optimize");
122 else if (!res(0).is_real_scalar()
123 || (gradient && !res(1).is_real_matrix()
124 && !(n == 1 && res(1).is_real_scalar())))
125 gripe_user_returned_invalid("nlopt_optimize");
128 if (n == 1 && res(1).is_real_scalar())
129 gradient[0] = res(1).double_value();
131 Matrix grad = res(1).matrix_value();
132 for (unsigned i = 0; i < n; ++i)
133 gradient[i] = grad(i);
136 return res(0).double_value();
141 #define CHECK1(cond, msg) if (!(cond)) { fprintf(stderr, msg "\n\n"); nlopt_destroy(opt); nlopt_destroy(local_opt); return NULL; }
143 nlopt_opt make_opt(Octave_map &opts, int n)
145 nlopt_opt opt = NULL, local_opt = NULL;
147 nlopt_algorithm algorithm =
148 nlopt_algorithm(struct_val_default(opts, "algorithm",
149 NLOPT_NUM_ALGORITHMS));
150 CHECK1(((int)algorithm) >= 0 && algorithm < NLOPT_NUM_ALGORITHMS,
151 "invalid opt.algorithm");
153 opt = nlopt_create(algorithm, n);
154 CHECK1(opt, "nlopt: out of memory");
156 Matrix m_inf(1, n, -HUGE_VAL);
157 Matrix lb = struct_val_default(opts, "lower_bounds", m_inf);
158 CHECK1(n == lb.length(), "wrong length of opt.lower_bounds");
159 CHECK1(nlopt_set_lower_bounds(opt, lb.data()) > 0, "nlopt: out of memory");
161 Matrix p_inf(1, n, +HUGE_VAL);
162 Matrix ub = struct_val_default(opts, "upper_bounds", p_inf);
163 CHECK1(n == ub.length(), "wrong length of opt.upper_bounds");
164 CHECK1(nlopt_set_upper_bounds(opt, ub.data()) > 0, "nlopt: out of memory");
166 nlopt_set_stopval(opt, struct_val_default(opts, "stopval", -HUGE_VAL));
167 nlopt_set_ftol_rel(opt, struct_val_default(opts, "ftol_rel", 0.0));
168 nlopt_set_ftol_abs(opt, struct_val_default(opts, "ftol_abs", 0.0));
169 nlopt_set_xtol_rel(opt, struct_val_default(opts, "xtol_rel", 0.0));
172 Matrix zeros(1, n, 0.0);
173 Matrix xtol_abs = struct_val_default(opts, "xtol_abs", zeros);
174 CHECK1(n == xtol_abs.length(), "stop.xtol_abs must have same length as x");
175 CHECK1(nlopt_set_xtol_abs(opt, xtol_abs.data())>0, "nlopt: out of memory");
178 nlopt_set_maxeval(opt, struct_val_default(opts, "maxeval", 0) < 0 ?
179 0 : struct_val_default(opts, "maxeval", 0));
180 nlopt_set_maxtime(opt, struct_val_default(opts, "maxtime", 0.0));
182 nlopt_set_population(opt, struct_val_default(opts, "population", 0));
183 nlopt_set_vector_storage(opt, struct_val_default(opts, "vector_storage", 0));
185 if (opts.contains("initial_step")) {
186 Matrix zeros(1, n, 0.0);
187 Matrix initial_step = struct_val_default(opts, "initial_step", zeros);
188 CHECK1(n == initial_step.length(),
189 "stop.initial_step must have same length as x");
190 CHECK1(nlopt_set_initial_step(opt, initial_step.data()) > 0,
191 "nlopt: out of memory");
194 if (opts.contains("local_optimizer")) {
195 CHECK1(opts.contents("local_optimizer").length() == 1
196 && (opts.contents("local_optimizer"))(0).is_map(),
197 "opt.local_optimizer must be a structure");
198 Octave_map local_opts = (opts.contents("local_optimizer"))(0).map_value();
199 CHECK1((local_opt = make_opt(local_opts, n)),
200 "error initializing local optimizer");
201 nlopt_set_local_optimizer(opt, local_opt);
202 nlopt_destroy(local_opt); local_opt = NULL;
208 #define CHECK(cond, msg) if (!(cond)) { fprintf(stderr, msg "\n\n"); nlopt_destroy(opt); return retval; }
210 DEFUN_DLD(nlopt_optimize, args, nargout, NLOPT_OPTIMIZE_USAGE)
212 octave_value_list retval;
214 nlopt_opt opt = NULL;
216 CHECK(args.length() == 2 && nargout <= 3, "wrong number of args");
218 CHECK(args(0).is_map(), "opt must be structure")
219 Octave_map opts = args(0).map_value();
221 CHECK(args(1).is_real_matrix() || args(1).is_real_scalar(),
222 "x must be real vector");
223 Matrix x = args(1).is_real_scalar() ?
224 Matrix(1, 1, args(1).double_value()) : args(1).matrix_value();
227 CHECK((opt = make_opt(opts, n)), "error initializing nlopt options");
229 user_function_data d;
231 d.verbose = struct_val_default(opts, "verbose", 0);
233 if (opts.contains("min_objective")) {
234 CHECK(opts.contents("min_objective").length() == 1
235 && (opts.contents("min_objective"))(0).is_function_handle(),
236 "opt.min_objective must be a function");
237 d.f = (opts.contents("min_objective"))(0).function_value();
238 nlopt_set_min_objective(opt, user_function, &d);
240 else if (opts.contains("max_objective")) {
241 CHECK(opts.contents("max_objective").length() == 1
242 && (opts.contents("max_objective"))(0).is_function_handle(),
243 "opt.max_objective must be a function");
244 d.f = (opts.contents("max_objective"))(0).function_value();
245 nlopt_set_max_objective(opt, user_function, &d);
248 CHECK(0,"either opt.min_objective or opt.max_objective must exist");
251 if (opts.contains("fc") && opts.contents("fc").length() == 1) {
252 CHECK((opts.contents("fc"))(0).is_cell(), "opt.fc must be cell array");
253 Cell fc = (opts.contents("fc"))(0).cell_value();
254 Matrix zeros(1, fc.length(), 0.0);
255 Matrix fc_tol = struct_val_default(opts, "fc_tol", zeros);
256 CHECK(fc_tol.length() == fc.length(),
257 "opt.fc must have same length as opt.fc_tol");
258 for (int i = 0; i < fc.length(); ++i) {
259 CHECK(fc(i).is_function() || fc(i).is_function_handle(),
260 "opt.fc must be a cell array of function handles");
261 CHECK(nlopt_add_inequality_constraint(opt, user_function1,
262 fc(i).function_value(),
264 "nlopt error adding inequality constraint");
268 if (opts.contains("h") && opts.contents("h").length() == 1) {
269 CHECK((opts.contents("h"))(0).is_cell(), "opt.h must be cell array");
270 Cell h = (opts.contents("h"))(0).cell_value();
271 Matrix zeros(1, h.length(), 0.0);
272 Matrix h_tol = struct_val_default(opts, "h_tol", zeros);
273 CHECK(h_tol.length() == h.length(),
274 "opt.h must have same length as opt.h_tol");
275 for (int i = 0; i < h.length(); ++i) {
276 CHECK(h(i).is_function() || h(i).is_function_handle(),
277 "opt.h must be a cell array of function handles");
278 CHECK(nlopt_add_equality_constraint(opt, user_function1,
279 h(i).function_value(),
281 "nlopt error adding equality constraint");
287 nlopt_result ret = nlopt_optimize(opt, x.fortran_vec(), &opt_f);
293 retval(2) = int(ret);