|
1
|
|
|
# Author: Simon Blanke |
|
2
|
|
|
# Email: [email protected] |
|
3
|
|
|
# License: MIT License |
|
4
|
|
|
|
|
5
|
|
|
import pytest |
|
6
|
|
|
import numpy as np |
|
7
|
|
|
|
|
8
|
|
|
from gradient_free_optimizers import ( |
|
9
|
|
|
HillClimbingOptimizer, |
|
10
|
|
|
StochasticHillClimbingOptimizer, |
|
11
|
|
|
TabuOptimizer, |
|
12
|
|
|
RandomRestartHillClimbingOptimizer, |
|
13
|
|
|
RandomAnnealingOptimizer, |
|
14
|
|
|
SimulatedAnnealingOptimizer, |
|
15
|
|
|
) |
|
16
|
|
|
|
|
17
|
|
|
|
|
18
|
|
|
def objective_function(para): |
|
19
|
|
|
score = -para["x1"] * para["x1"] |
|
20
|
|
|
return score |
|
21
|
|
|
|
|
22
|
|
|
|
|
23
|
|
|
search_space = {"x1": np.arange(-100, 101, 1)} |
|
24
|
|
|
|
|
25
|
|
|
|
|
26
|
|
|
HillClimbing_para = [ |
|
27
|
|
|
({"epsilon": 0.0001}), |
|
28
|
|
|
({"epsilon": 1}), |
|
29
|
|
|
({"epsilon": 10}), |
|
30
|
|
|
({"epsilon": 10000}), |
|
31
|
|
|
({"distribution": "normal"}), |
|
32
|
|
|
({"distribution": "laplace"}), |
|
33
|
|
|
({"distribution": "logistic"}), |
|
34
|
|
|
({"distribution": "gumbel"}), |
|
35
|
|
|
({"n_neighbours": 1}), |
|
36
|
|
|
({"n_neighbours": 10}), |
|
37
|
|
|
({"n_neighbours": 100}), |
|
38
|
|
|
({"rand_rest_p": 0}), |
|
39
|
|
|
({"rand_rest_p": 0.5}), |
|
40
|
|
|
({"rand_rest_p": 1}), |
|
41
|
|
|
({"rand_rest_p": 10}), |
|
42
|
|
|
] |
|
43
|
|
|
|
|
44
|
|
|
|
|
45
|
|
|
pytest_wrapper = ("para", HillClimbing_para) |
|
46
|
|
|
|
|
47
|
|
|
optimizers_local = ( |
|
48
|
|
|
"Optimizer", |
|
49
|
|
|
[ |
|
50
|
|
|
(HillClimbingOptimizer), |
|
51
|
|
|
(StochasticHillClimbingOptimizer), |
|
52
|
|
|
(TabuOptimizer), |
|
53
|
|
|
(SimulatedAnnealingOptimizer), |
|
54
|
|
|
(RandomRestartHillClimbingOptimizer), |
|
55
|
|
|
(RandomAnnealingOptimizer), |
|
56
|
|
|
], |
|
57
|
|
|
) |
|
58
|
|
|
|
|
59
|
|
|
|
|
60
|
|
|
@pytest.mark.parametrize(*optimizers_local) |
|
61
|
|
|
@pytest.mark.parametrize(*pytest_wrapper) |
|
62
|
|
|
def test_HillClimbing_para(Optimizer, para): |
|
63
|
|
|
opt = Optimizer(search_space, **para) |
|
64
|
|
|
opt.search( |
|
65
|
|
|
objective_function, |
|
66
|
|
|
n_iter=10, |
|
67
|
|
|
memory=False, |
|
68
|
|
|
verbosity=False, |
|
69
|
|
|
initialize={"vertices": 1}, |
|
70
|
|
|
) |
|
71
|
|
|
|
|
72
|
|
|
for optimizer in opt.optimizers: |
|
73
|
|
|
para_key = list(para.keys())[0] |
|
74
|
|
|
para_value = getattr(optimizer, para_key) |
|
75
|
|
|
|
|
76
|
|
|
assert para_value == para[para_key] |
|
77
|
|
|
|
SimonBlanke /
Gradient-Free-Optimizers
