1 | import time |
||
2 | import pytest |
||
3 | import numpy as np |
||
4 | |||
5 | |||
6 | from hyperactive import Hyperactive |
||
7 | from hyperactive.optimizers.strategies import CustomOptimizationStrategy |
||
8 | from hyperactive.optimizers import GridSearchOptimizer |
||
9 | |||
10 | from ._parametrize import optimizers_non_smbo |
||
11 | |||
12 | |||
13 | def objective_function(opt): |
||
14 | time.sleep(0.01) |
||
15 | score = -(opt["x1"] * opt["x1"]) |
||
16 | return score |
||
17 | |||
18 | |||
19 | search_space = { |
||
20 | "x1": list(np.arange(0, 100, 1)), |
||
21 | } |
||
22 | |||
23 | |||
24 | def test_memory_Warm_start_0(): |
||
25 | optimizer1 = GridSearchOptimizer() |
||
26 | optimizer2 = GridSearchOptimizer() |
||
27 | |||
28 | opt_strat = CustomOptimizationStrategy() |
||
29 | opt_strat.add_optimizer(optimizer1, duration=0.2) |
||
30 | opt_strat.add_optimizer(optimizer2, duration=0.8) |
||
31 | |||
32 | n_iter = 1000 |
||
33 | |||
34 | c_time = time.time() |
||
35 | |||
36 | hyper = Hyperactive() |
||
37 | hyper.add_search( |
||
38 | objective_function, |
||
39 | search_space, |
||
40 | optimizer=opt_strat, |
||
41 | n_iter=n_iter, |
||
42 | memory=True, |
||
43 | ) |
||
44 | hyper.run() |
||
45 | |||
46 | d_time = time.time() - c_time |
||
47 | |||
48 | search_data = hyper.search_data(objective_function) |
||
49 | |||
50 | optimizer1 = hyper.opt_pros[0].optimizer_setup_l[0]["optimizer"] |
||
51 | optimizer2 = hyper.opt_pros[0].optimizer_setup_l[1]["optimizer"] |
||
52 | |||
53 | assert len(search_data) == n_iter |
||
54 | |||
55 | assert len(optimizer1.search_data) == 200 |
||
56 | assert len(optimizer2.search_data) == 800 |
||
57 | |||
58 | assert optimizer1.best_score <= optimizer2.best_score |
||
59 | |||
60 | print("\n d_time", d_time) |
||
61 | |||
62 | assert d_time < 3 |
||
63 | |||
64 | |||
65 | View Code Duplication | def test_memory_Warm_start_1(): |
|
0 ignored issues
–
show
Duplication
introduced
by
Loading history...
|
|||
66 | optimizer1 = GridSearchOptimizer() |
||
67 | optimizer2 = GridSearchOptimizer() |
||
68 | |||
69 | opt_strat = CustomOptimizationStrategy() |
||
70 | opt_strat.add_optimizer(optimizer1, duration=0.2) |
||
71 | opt_strat.add_optimizer(optimizer2, duration=0.8) |
||
72 | |||
73 | n_iter = 100 |
||
74 | |||
75 | search_space = { |
||
76 | "x1": list(np.arange(0, 1, 1)), |
||
77 | } |
||
78 | |||
79 | c_time = time.time() |
||
80 | |||
81 | hyper = Hyperactive() |
||
82 | hyper.add_search( |
||
83 | objective_function, |
||
84 | search_space, |
||
85 | optimizer=opt_strat, |
||
86 | n_iter=n_iter, |
||
87 | memory=False, |
||
88 | ) |
||
89 | hyper.run() |
||
90 | |||
91 | d_time = time.time() - c_time |
||
92 | |||
93 | search_data = hyper.search_data(objective_function) |
||
94 | |||
95 | optimizer1 = hyper.opt_pros[0].optimizer_setup_l[0]["optimizer"] |
||
96 | optimizer2 = hyper.opt_pros[0].optimizer_setup_l[1]["optimizer"] |
||
97 | |||
98 | assert len(search_data) == n_iter |
||
99 | |||
100 | assert len(optimizer1.search_data) == 20 |
||
101 | assert len(optimizer2.search_data) == 80 |
||
102 | |||
103 | assert optimizer1.best_score <= optimizer2.best_score |
||
104 | |||
105 | print("\n d_time", d_time) |
||
106 | |||
107 | assert d_time > 0.95 |
||
108 | |||
109 | |||
110 | |||
111 | View Code Duplication | @pytest.mark.parametrize(*optimizers_non_smbo) |
|
0 ignored issues
–
show
|
|||
112 | def test_memory_Warm_start_2(Optimizer_non_smbo): |
||
113 | optimizer1 = GridSearchOptimizer() |
||
114 | optimizer2 = Optimizer_non_smbo() |
||
115 | |||
116 | opt_strat = CustomOptimizationStrategy() |
||
117 | opt_strat.add_optimizer(optimizer1, duration=0.5) |
||
118 | opt_strat.add_optimizer(optimizer2, duration=0.5) |
||
119 | |||
120 | search_space = { |
||
121 | "x1": list(np.arange(0, 50, 1)), |
||
122 | } |
||
123 | |||
124 | n_iter = 100 |
||
125 | |||
126 | c_time = time.time() |
||
127 | |||
128 | hyper = Hyperactive() |
||
129 | hyper.add_search( |
||
130 | objective_function, |
||
131 | search_space, |
||
132 | optimizer=opt_strat, |
||
133 | n_iter=n_iter, |
||
134 | memory=True, |
||
135 | ) |
||
136 | hyper.run() |
||
137 | |||
138 | d_time = time.time() - c_time |
||
139 | |||
140 | search_data = hyper.search_data(objective_function) |
||
141 | |||
142 | optimizer1 = hyper.opt_pros[0].optimizer_setup_l[0]["optimizer"] |
||
143 | optimizer2 = hyper.opt_pros[0].optimizer_setup_l[1]["optimizer"] |
||
144 | |||
145 | assert len(search_data) == n_iter |
||
146 | |||
147 | assert len(optimizer1.search_data) == 50 |
||
148 | assert len(optimizer2.search_data) == 50 |
||
149 | |||
150 | assert optimizer1.best_score <= optimizer2.best_score |
||
151 | |||
152 | print("\n d_time", d_time) |
||
153 | |||
154 | assert d_time < 0.9 |
||
155 |