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
|
// SPDX-License-Identifier: LGPL-3.0-or-later
// Author: Kristian Lytje
#include <mini/Golden.h>
#include <mini/detail/Parameter.h>
#include <mini/detail/FittedParameter.h>
#include <mini/detail/Evaluation.h>
#include <utility/Exceptions.h>
#include <utility/Utility.h>
using namespace ausaxs;
using namespace ausaxs::mini;
Golden::Golden(double(&func)(std::vector<double>)) : Minimizer(func) {}
Golden::Golden(std::function<double(std::vector<double>)> func) : Minimizer(std::move(func)) {}
Golden::Golden(double(&func)(std::vector<double>), const Parameter& param) : Minimizer(func) {
add_parameter(param);
}
Golden::Golden(std::function<double(std::vector<double>)> func, const Parameter& param) : Minimizer(std::move(func)) {
add_parameter(param);
}
Limit Golden::search(Limit bounds) const {
// Code adapted from the python implementation from Wikipedia: https://en.wikipedia.org/wiki/Golden-section_search
double a = bounds.min, b = bounds.max;
double temp = a + b;
// sort such that a < b
a = std::min(a, b);
b = temp - a;
double diff = b - a;
if (diff < tol) [[unlikely]] {
return Limit(a, b);
}
// expected number of steps to reach tolerance
unsigned int n = std::ceil(std::log(tol/diff)/std::log(invphi));
double c = a + invphi2*diff;
double d = a + invphi*diff;
double fc = function({c});
double fd = function({d});
for (unsigned int k = 0; k < n-1; k++) {
if (fc < fd) {
b = d;
d = c;
fd = fc;
diff = invphi*diff;
c = a + invphi2*diff;
fc = function({c});
} else {
a = c;
c = d;
fc = fd;
diff = invphi*diff;
d = a + invphi*diff;
fd = function({d});
}
}
if (fc < fd) {
return Limit(a, d);
} else {
return Limit(c, b);
}
}
Result Golden::minimize_override() {
Limit optimal_interval = search(parameters[0].bounds.value());
FittedParameter p(parameters[0], optimal_interval.center(), optimal_interval-optimal_interval.center());
return Result(p, function({p.value}), fevals);
}
void Golden::add_parameter(const Parameter& param) {
if (!param.has_bounds()) {throw except::invalid_argument("Golden::add_parameter: The parameter must be supplied with limits for this minimizer.");}
if (!parameters.empty()) {throw except::invalid_operation("Golden::add_parameter: This minimizer only supports 1D problems.");}
parameters.push_back(param);
}
|
