|
1
|
|
|
def warn(*args, **kwargs): |
|
2
|
|
|
pass |
|
3
|
|
|
|
|
4
|
|
|
|
|
5
|
|
|
import warnings |
|
6
|
|
|
|
|
7
|
|
|
warnings.warn = warn |
|
8
|
|
|
|
|
9
|
|
|
import numpy as np |
|
10
|
|
|
from gradient_free_optimizers import ( |
|
11
|
|
|
HillClimbingOptimizer, |
|
12
|
|
|
StochasticHillClimbingOptimizer, |
|
13
|
|
|
RepulsingHillClimbingOptimizer, |
|
14
|
|
|
SimulatedAnnealingOptimizer, |
|
15
|
|
|
DownhillSimplexOptimizer, |
|
16
|
|
|
RandomSearchOptimizer, |
|
17
|
|
|
PowellsMethod, |
|
18
|
|
|
GridSearchOptimizer, |
|
19
|
|
|
RandomRestartHillClimbingOptimizer, |
|
20
|
|
|
RandomAnnealingOptimizer, |
|
21
|
|
|
PatternSearch, |
|
22
|
|
|
DirectAlgorithm, |
|
23
|
|
|
ParallelTemperingOptimizer, |
|
24
|
|
|
ParticleSwarmOptimizer, |
|
25
|
|
|
SpiralOptimization, |
|
26
|
|
|
EvolutionStrategyOptimizer, |
|
27
|
|
|
LipschitzOptimizer, |
|
28
|
|
|
BayesianOptimizer, |
|
29
|
|
|
TreeStructuredParzenEstimators, |
|
30
|
|
|
ForestOptimizer, |
|
31
|
|
|
) |
|
32
|
|
|
|
|
33
|
|
|
optimizers = [ |
|
34
|
|
|
HillClimbingOptimizer, |
|
35
|
|
|
StochasticHillClimbingOptimizer, |
|
36
|
|
|
RepulsingHillClimbingOptimizer, |
|
37
|
|
|
SimulatedAnnealingOptimizer, |
|
38
|
|
|
DownhillSimplexOptimizer, |
|
39
|
|
|
RandomSearchOptimizer, |
|
40
|
|
|
PowellsMethod, |
|
41
|
|
|
GridSearchOptimizer, |
|
42
|
|
|
RandomRestartHillClimbingOptimizer, |
|
43
|
|
|
RandomAnnealingOptimizer, |
|
44
|
|
|
PatternSearch, |
|
45
|
|
|
DirectAlgorithm, |
|
46
|
|
|
ParallelTemperingOptimizer, |
|
47
|
|
|
ParticleSwarmOptimizer, |
|
48
|
|
|
SpiralOptimization, |
|
49
|
|
|
EvolutionStrategyOptimizer, |
|
50
|
|
|
LipschitzOptimizer, |
|
51
|
|
|
BayesianOptimizer, |
|
52
|
|
|
TreeStructuredParzenEstimators, |
|
53
|
|
|
ForestOptimizer, |
|
54
|
|
|
] |
|
55
|
|
|
|
|
56
|
|
|
|
|
57
|
|
View Code Duplication |
def ackley_function(para): |
|
|
|
|
|
|
58
|
|
|
x, y = para["x"], para["y"] |
|
59
|
|
|
|
|
60
|
|
|
loss = ( |
|
61
|
|
|
-20 * np.exp(-0.2 * np.sqrt(0.5 * (x * x + y * y))) |
|
62
|
|
|
- np.exp(0.5 * (np.cos(2 * np.pi * x) + np.cos(2 * np.pi * y))) |
|
63
|
|
|
+ np.exp(1) |
|
64
|
|
|
+ 20 |
|
65
|
|
|
) |
|
66
|
|
|
|
|
67
|
|
|
return -loss |
|
68
|
|
|
|
|
69
|
|
|
|
|
70
|
|
|
search_space = { |
|
71
|
|
|
"x": np.arange(-10, 10, 1), |
|
72
|
|
|
"y": np.arange(-10, 10, 1), |
|
73
|
|
|
} |
|
74
|
|
|
|
|
75
|
|
|
for optimizer in optimizers: |
|
76
|
|
|
opt = optimizer(search_space) |
|
77
|
|
|
opt.search(ackley_function, n_iter=15, verbosity=False) |
|
78
|
|
|
|
SimonBlanke /
Gradient-Free-Optimizers
Loading history...