|
1
|
|
|
def warn(*args, **kwargs): |
|
2
|
|
|
pass |
|
3
|
|
|
|
|
4
|
|
|
|
|
5
|
|
|
import warnings |
|
6
|
|
|
|
|
7
|
|
|
warnings.warn = warn |
|
8
|
|
|
|
|
9
|
|
|
import os |
|
10
|
|
|
import numpy as np |
|
11
|
|
|
|
|
12
|
|
|
|
|
13
|
|
|
from gradient_free_optimizers import ( |
|
14
|
|
|
HillClimbingOptimizer, |
|
15
|
|
|
StochasticHillClimbingOptimizer, |
|
16
|
|
|
RepulsingHillClimbingOptimizer, |
|
17
|
|
|
SimulatedAnnealingOptimizer, |
|
18
|
|
|
DownhillSimplexOptimizer, |
|
19
|
|
|
RandomSearchOptimizer, |
|
20
|
|
|
GridSearchOptimizer, |
|
21
|
|
|
RandomRestartHillClimbingOptimizer, |
|
22
|
|
|
PowellsMethod, |
|
23
|
|
|
PatternSearch, |
|
24
|
|
|
LipschitzOptimizer, |
|
25
|
|
|
DirectAlgorithm, |
|
26
|
|
|
RandomAnnealingOptimizer, |
|
27
|
|
|
ParallelTemperingOptimizer, |
|
28
|
|
|
ParticleSwarmOptimizer, |
|
29
|
|
|
SpiralOptimization, |
|
30
|
|
|
GeneticAlgorithmOptimizer, |
|
31
|
|
|
EvolutionStrategyOptimizer, |
|
32
|
|
|
DifferentialEvolutionOptimizer, |
|
33
|
|
|
BayesianOptimizer, |
|
34
|
|
|
TreeStructuredParzenEstimators, |
|
35
|
|
|
ForestOptimizer, |
|
36
|
|
|
) |
|
37
|
|
|
from surfaces.test_functions.mathematical import SphereFunction, AckleyFunction |
|
38
|
|
|
|
|
39
|
|
|
from search_path_gif import search_path_gif |
|
40
|
|
|
|
|
41
|
|
|
here = os.path.dirname(os.path.abspath(__file__)) |
|
42
|
|
|
|
|
43
|
|
|
search_space = { |
|
44
|
|
|
"x0": np.arange(-2, 8, 0.1), |
|
45
|
|
|
"x1": np.arange(-2, 8, 0.1), |
|
46
|
|
|
} |
|
47
|
|
|
|
|
48
|
|
|
initialize_1 = { |
|
49
|
|
|
"warm_start": [{"x0": 7, "x1": 7}], |
|
50
|
|
|
} |
|
51
|
|
|
initialize_pop = {"vertices": 4} |
|
52
|
|
|
random_state = 23 |
|
53
|
|
|
|
|
54
|
|
|
sphere_function = SphereFunction(n_dim=2) |
|
55
|
|
|
ackley_function = AckleyFunction() |
|
56
|
|
|
|
|
57
|
|
|
test_function_l = [sphere_function, ackley_function] |
|
58
|
|
|
|
|
59
|
|
|
optimizer_l = [ |
|
60
|
|
|
HillClimbingOptimizer, |
|
61
|
|
|
StochasticHillClimbingOptimizer, |
|
62
|
|
|
RepulsingHillClimbingOptimizer, |
|
63
|
|
|
SimulatedAnnealingOptimizer, |
|
64
|
|
|
DownhillSimplexOptimizer, |
|
65
|
|
|
RandomSearchOptimizer, |
|
66
|
|
|
GridSearchOptimizer, |
|
67
|
|
|
RandomRestartHillClimbingOptimizer, |
|
68
|
|
|
PowellsMethod, |
|
69
|
|
|
PatternSearch, |
|
70
|
|
|
LipschitzOptimizer, |
|
71
|
|
|
DirectAlgorithm, |
|
72
|
|
|
RandomAnnealingOptimizer, |
|
73
|
|
|
ParallelTemperingOptimizer, |
|
74
|
|
|
ParticleSwarmOptimizer, |
|
75
|
|
|
SpiralOptimization, |
|
76
|
|
|
GeneticAlgorithmOptimizer, |
|
77
|
|
|
EvolutionStrategyOptimizer, |
|
78
|
|
|
DifferentialEvolutionOptimizer, |
|
79
|
|
|
BayesianOptimizer, |
|
80
|
|
|
TreeStructuredParzenEstimators, |
|
81
|
|
|
ForestOptimizer, |
|
82
|
|
|
] |
|
83
|
|
|
""" |
|
84
|
|
|
|
|
85
|
|
|
""" |
|
86
|
|
|
|
|
87
|
|
|
for test_function_ in test_function_l: |
|
88
|
|
|
objective_function = test_function_.objective_function |
|
89
|
|
|
for optimizer in optimizer_l: |
|
90
|
|
|
if optimizer.computationally_expensive: |
|
91
|
|
|
n_iter = 50 |
|
92
|
|
|
else: |
|
93
|
|
|
n_iter = 150 |
|
94
|
|
|
|
|
95
|
|
|
if optimizer.optimizer_type == "population": |
|
96
|
|
|
initialize = initialize_pop |
|
97
|
|
|
else: |
|
98
|
|
|
initialize = initialize_1 |
|
99
|
|
|
|
|
100
|
|
|
para_dict = { |
|
101
|
|
|
"path": os.path.join(here, "gifs"), |
|
102
|
|
|
"optimizer": optimizer, |
|
103
|
|
|
"opt_para": {}, |
|
104
|
|
|
"name": optimizer._name_ |
|
105
|
|
|
+ "_" |
|
106
|
|
|
+ test_function_._name_ |
|
107
|
|
|
+ "_" |
|
108
|
|
|
+ ".gif", |
|
109
|
|
|
"n_iter": n_iter, |
|
110
|
|
|
"objective_function": objective_function, |
|
111
|
|
|
"search_space": search_space, |
|
112
|
|
|
"initialize": initialize, |
|
113
|
|
|
"random_state": random_state, |
|
114
|
|
|
} |
|
115
|
|
|
search_path_gif(**para_dict) |
|
116
|
|
|
|
SimonBlanke /
Gradient-Free-Optimizers
