1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
|
/*
* Copyright (c) 2022, Miroslav Stoyanov & Weiwei Kong
*
* This file is part of
* Toolkit for Adaptive Stochastic Modeling And Non-Intrusive ApproximatioN: TASMANIAN
*
* Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following
* conditions are met:
*
* 1. Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer.
*
* 2. Redistributions in binary form must reproduce the above copyright notice, this list of conditions
* and the following disclaimer in the documentation and/or other materials provided with the distribution.
*
* 3. Neither the name of the copyright holder nor the names of its contributors may be used to endorse
* or promote products derived from this software without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES,
* INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
* IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY,
* OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA,
* OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
* OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
* POSSIBILITY OF SUCH DAMAGE.
*
* UT-BATTELLE, LLC AND THE UNITED STATES GOVERNMENT MAKE NO REPRESENTATIONS AND DISCLAIM ALL WARRANTIES, BOTH EXPRESSED AND
* IMPLIED. THERE ARE NO EXPRESS OR IMPLIED WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE, OR THAT THE USE OF
* THE SOFTWARE WILL NOT INFRINGE ANY PATENT, COPYRIGHT, TRADEMARK, OR OTHER PROPRIETARY RIGHTS, OR THAT THE SOFTWARE WILL
* ACCOMPLISH THE INTENDED RESULTS OR THAT THE SOFTWARE OR ITS USE WILL NOT RESULT IN INJURY OR DAMAGE. THE USER ASSUMES
* RESPONSIBILITY FOR ALL LIABILITIES, PENALTIES, FINES, CLAIMS, CAUSES OF ACTION, AND COSTS AND EXPENSES, CAUSED BY, RESULTING
* FROM OR ARISING OUT OF, IN WHOLE OR IN PART THE USE, STORAGE OR DISPOSAL OF THE SOFTWARE.
*/
#ifndef __TASMANIAN_PARTICLE_SWARM_CPP
#define __TASMANIAN_PARTICLE_SWARM_CPP
#include "tsgParticleSwarm.hpp"
namespace TasOptimization {
ParticleSwarmState::ParticleSwarmState(int cnum_dimensions, int cnum_particles):
positions_initialized(false), velocities_initialized(false), best_positions_initialized(false), cache_initialized(false),
num_dimensions(cnum_dimensions), num_particles(cnum_particles),
particle_positions(std::vector<double>(num_particles * num_dimensions)),
particle_velocities(std::vector<double>(num_particles * num_dimensions)),
best_particle_positions(std::vector<double>((num_particles + 1) * num_dimensions)),
cache_particle_fvals(std::vector<double>(num_particles, std::numeric_limits<double>::max())),
cache_best_particle_fvals(std::vector<double>(num_particles + 1, std::numeric_limits<double>::max())),
cache_particle_inside(std::vector<bool>(num_particles, false)),
cache_best_particle_inside(std::vector<bool>(num_particles + 1, false)) {}
ParticleSwarmState::ParticleSwarmState(int cnum_dimensions, std::vector<double> &&pp, std::vector<double> &&pv):
positions_initialized(true), velocities_initialized(true), best_positions_initialized(false), cache_initialized(false),
num_dimensions(cnum_dimensions), num_particles(pp.size() / num_dimensions),
particle_positions(std::move(pp)), particle_velocities(std::move(pv)),
best_particle_positions(std::vector<double>((num_particles + 1) * num_dimensions)),
cache_particle_fvals(std::vector<double>(num_particles, std::numeric_limits<double>::max())),
cache_best_particle_fvals(std::vector<double>(num_particles + 1, std::numeric_limits<double>::max())),
cache_particle_inside(std::vector<bool>(num_particles, false)),
cache_best_particle_inside(std::vector<bool>(num_particles + 1, false)) {}
void ParticleSwarmState::initializeParticlesInsideBox(const double box_lower[], const double box_upper[],
const std::function<double(void)> get_random01) {
for (int i=0; i<num_particles * num_dimensions; i++) {
double range = std::fabs(box_upper[i % num_dimensions] - box_lower[i % num_dimensions]);
particle_positions[i] = range * get_random01() + box_lower[i % num_dimensions];
particle_velocities[i] = 2 * range * get_random01() - range;
}
positions_initialized = true;
velocities_initialized = true;
}
void ParticleSwarmState::initializeParticlesInsideBox(const std::vector<double> &box_lower, const std::vector<double> &box_upper,
const std::function<double(void)> get_random01) {
checkVarSize("ParticleSwarmState::initializeParticlesInsideBox", "box lower bounds", box_lower.size(), num_dimensions);
checkVarSize("ParticleSwarmState::initializeParticlesInsideBox", "box upper bounds", box_upper.size(), num_dimensions);
ParticleSwarmState::initializeParticlesInsideBox(box_lower.data(), box_upper.data(), get_random01);
}
void ParticleSwarm(const ObjectiveFunction f, const TasDREAM::DreamDomain inside, const double inertia_weight,
const double cognitive_coeff, const double social_coeff, const int num_iterations,
ParticleSwarmState &state, const std::function<double(void)> get_random01) {
// Only run the algorithm on properly initialized states.
if (!state.positions_initialized) {
throw std::runtime_error("Particle positions have not been initialized in the input state object");
}
if (!state.velocities_initialized) {
throw std::runtime_error("Particle velocities have not been initialized in the input state object");
}
// Initialize helper variables and functions.
size_t num_dimensions = (size_t) state.getNumDimensions();
size_t num_particles = (size_t) state.getNumParticles();
// Create a lambda that converts f to a constrained version that only evaluates points inside the domain. This lambda also
// writes to a bool vector whose i-th entry is true if particle i is in the domain.
auto f_constrained = [=](const std::vector<double> &x_batch, std::vector<double> &fval_batch, std::vector<bool> &inside_batch)->void {
// Collect and apply the domain information given by inside() and x_batch.
size_t num_batch(fval_batch.size()), num_inside(0);
std::vector<double> candidate(num_dimensions), inside_points;
for (size_t i=0; i<num_batch; i++) {
fval_batch[i] = std::numeric_limits<double>::max();
std::copy_n(x_batch.begin() + i * num_dimensions, num_dimensions, candidate.begin());
inside_batch[i] = inside(candidate);
if (inside_batch[i]) {
std::copy_n(candidate.begin(), num_dimensions, std::back_inserter(inside_points));
num_inside++;
}
}
// Evaluate f on the inside points and copy the resulting values to fval_batch.
std::vector<double> inside_vals(num_inside);
if (num_inside > 0)
f(inside_points, inside_vals);
int j = 0;
for (size_t i=0; i<num_batch; i++) {
fval_batch[i] = (inside_batch[i]) ? inside_vals[j++] : 0.0;
}
};
// Create a lambda that updates the best particle positions and fvals (in the cache).
auto update = [&]()->void {
for (size_t i=0; i<num_particles; i++) {
// if current particle not inside, do nothing
// if inside and best particle not inside, accept new best know position
// if both inside, but current is better than best know, update the best known
if (state.cache_particle_inside[i]
and (not state.cache_best_particle_inside[i] or state.cache_particle_fvals[i] < state.cache_best_particle_fvals[i])){
std::copy_n(state.particle_positions.begin() + i * num_dimensions, num_dimensions,
state.best_particle_positions.begin() + i * num_dimensions);
state.cache_best_particle_fvals[i] = state.cache_particle_fvals[i];
state.cache_best_particle_inside[i] = true;
if (not state.cache_best_particle_inside[num_particles] or
state.cache_best_particle_fvals[i] < state.cache_best_particle_fvals[num_particles]) {
std::copy_n(state.particle_positions.begin() + i * num_dimensions, num_dimensions,
state.best_particle_positions.begin() + num_particles * num_dimensions);
state.cache_best_particle_fvals[num_particles] = state.cache_best_particle_fvals[i];
state.cache_best_particle_inside[num_particles] = true;
}
}
}
};
// Set up the cache and best particle positions.
if (!state.cache_initialized) {
f_constrained(state.particle_positions, state.cache_particle_fvals, state.cache_particle_inside);
if (state.best_positions_initialized) {
f_constrained(state.best_particle_positions, state.cache_best_particle_fvals, state.cache_best_particle_inside);
}
state.cache_initialized = true;
}
update();
state.best_positions_initialized = true;
// Main algorithm starts here.
std::vector<double> rng_cache(2 * num_particles);
for(int iter=0; iter<num_iterations; iter++) {
if (state.cache_best_particle_inside[num_particles]) {
for(auto &r : rng_cache) r = get_random01();
#pragma omp parallel for
for (int i=0; i<static_cast<int>(num_particles); i++) {
if (state.cache_best_particle_inside[i]) {
for(size_t j=0; j<num_dimensions; j++){
state.particle_velocities[i*num_dimensions + j]
= inertia_weight * state.particle_velocities[i*num_dimensions + j]
+ cognitive_coeff * rng_cache[2*i] *
(state.best_particle_positions[i*num_dimensions + j] - state.particle_positions[i*num_dimensions + j])
+ social_coeff * rng_cache[2*i + 1] *
(state.best_particle_positions[num_particles * num_dimensions + j] - state.particle_positions[i*num_dimensions + j]);
}
} else {
for(size_t j=0; j<num_dimensions; j++){
state.particle_velocities[i*num_dimensions + j]
= inertia_weight * state.particle_velocities[i*num_dimensions + j]
+ social_coeff * rng_cache[2*i] *
(state.best_particle_positions[num_particles * num_dimensions + j] - state.particle_positions[i*num_dimensions + j]);
}
}
}
} else {
for(size_t i=0; i<num_particles; i++) rng_cache[i] = get_random01();
#pragma omp parallel for
for (int i=0; i<static_cast<int>(num_particles); i++) {
if (state.cache_best_particle_inside[i]) {
for(size_t j=0; j<num_dimensions; j++){
state.particle_velocities[i*num_dimensions + j]
= inertia_weight * state.particle_velocities[i*num_dimensions + j]
+ cognitive_coeff * rng_cache[i] *
(state.best_particle_positions[i*num_dimensions + j] - state.particle_positions[i*num_dimensions + j]);
}
} else {
for(size_t j=0; j<num_dimensions; j++){
state.particle_velocities[i*num_dimensions + j] = inertia_weight * state.particle_positions[i*num_dimensions + j];
}
}
}
}
for (size_t i=0; i< num_particles * num_dimensions; i++) {
state.particle_positions[i] += state.particle_velocities[i];
}
f_constrained(state.particle_positions, state.cache_particle_fvals, state.cache_particle_inside);
update();
}
}
} // End namespace
#endif
|
