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
|
#ifndef BRUTEFORCEOPTIMIZER_HPP
#define BRUTEFORCEOPTIMIZER_HPP
#include <libslic3r/Optimize/Optimizer.hpp>
namespace Slic3r { namespace opt {
namespace detail {
// Implementing a bruteforce optimizer
// Return the number of iterations needed to reach a specific grid position (idx)
template<size_t N>
long num_iter(const std::array<size_t, N> &idx, size_t gridsz)
{
long ret = 0;
for (size_t i = 0; i < N; ++i) ret += idx[i] * std::pow(gridsz, i);
return ret;
}
// Implementation of a grid search where the search interval is sampled in
// equidistant points for each dimension. Grid size determines the number of
// samples for one dimension so the number of function calls is gridsize ^ dimension.
struct AlgBurteForce {
bool to_min;
StopCriteria stc;
size_t gridsz;
AlgBurteForce(const StopCriteria &cr, size_t gs): stc{cr}, gridsz{gs} {}
// This function is called recursively for each dimension and generates
// the grid values for the particular dimension. If D is less than zero,
// the object function input values are generated for each dimension and it
// can be evaluated. The current best score is compared with the newly
// returned score and changed appropriately.
template<int D, size_t N, class Fn, class Cmp>
bool run(std::array<size_t, N> &idx,
Result<N> &result,
const Bounds<N> &bounds,
Fn &&fn,
Cmp &&cmp)
{
if (stc.stop_condition()) return false;
if constexpr (D < 0) { // Let's evaluate fn
Input<N> inp;
auto max_iter = stc.max_iterations();
if (max_iter && num_iter(idx, gridsz) >= max_iter)
return false;
for (size_t d = 0; d < N; ++d) {
const Bound &b = bounds[d];
double step = (b.max() - b.min()) / (gridsz - 1);
inp[d] = b.min() + idx[d] * step;
}
auto score = fn(inp);
if (cmp(score, result.score)) { // Change current score to the new
double absdiff = std::abs(score - result.score);
result.score = score;
result.optimum = inp;
// Check if the required precision is reached.
if (absdiff < stc.abs_score_diff() ||
absdiff < stc.rel_score_diff() * std::abs(score))
return false;
}
} else {
for (size_t i = 0; i < gridsz; ++i) {
idx[D] = i; // Mark the current grid position and dig down
if (!run<D - 1>(idx, result, bounds, std::forward<Fn>(fn),
std::forward<Cmp>(cmp)))
return false;
}
}
return true;
}
template<class Fn, size_t N>
Result<N> optimize(Fn&& fn,
const Input<N> &/*initvals*/,
const Bounds<N>& bounds)
{
std::array<size_t, N> idx = {};
Result<N> result;
if (to_min) {
result.score = std::numeric_limits<double>::max();
run<int(N) - 1>(idx, result, bounds, std::forward<Fn>(fn),
std::less<double>{});
}
else {
result.score = std::numeric_limits<double>::lowest();
run<int(N) - 1>(idx, result, bounds, std::forward<Fn>(fn),
std::greater<double>{});
}
return result;
}
};
} // namespace detail
using AlgBruteForce = detail::AlgBurteForce;
template<>
class Optimizer<AlgBruteForce> {
AlgBruteForce m_alg;
public:
Optimizer(const StopCriteria &cr = {}, size_t gridsz = 100)
: m_alg{cr, gridsz}
{}
Optimizer& to_max() { m_alg.to_min = false; return *this; }
Optimizer& to_min() { m_alg.to_min = true; return *this; }
template<class Func, size_t N>
Result<N> optimize(Func&& func,
const Input<N> &initvals,
const Bounds<N>& bounds)
{
return m_alg.optimize(std::forward<Func>(func), initvals, bounds);
}
Optimizer &set_criteria(const StopCriteria &cr)
{
m_alg.stc = cr; return *this;
}
const StopCriteria &get_criteria() const { return m_alg.stc; }
};
}} // namespace Slic3r::opt
#endif // BRUTEFORCEOPTIMIZER_HPP
|
