|
1
|
|
|
"""Hybrid bat algorithm. |
|
2
|
|
|
|
|
3
|
|
|
Date: 2018 |
|
4
|
|
|
|
|
5
|
|
|
Authors : Grega Vrbancic |
|
6
|
|
|
|
|
7
|
|
|
License: MIT |
|
8
|
|
|
|
|
9
|
|
|
Reference paper: Fister Jr., Iztok and Fister, Dusan and Yang, Xin-She. |
|
10
|
|
|
"A Hybrid Bat Algorithm". Elektrotehniski vestnik, 2013. 1-7. |
|
11
|
|
|
""" |
|
12
|
|
|
|
|
13
|
|
|
|
|
14
|
|
|
import random |
|
15
|
|
|
from NiaPy.benchmarks.utility import Utility |
|
16
|
|
|
|
|
17
|
|
|
__all__ = ['HybridBatAlgorithm'] |
|
18
|
|
|
|
|
19
|
|
|
|
|
20
|
|
|
class HybridBatAlgorithm(object): |
|
21
|
|
|
# pylint: disable=too-many-instance-attributes |
|
22
|
|
View Code Duplication |
def __init__(self, D, NP, nFES, A, r, Qmin, Qmax, benchmark): |
|
|
|
|
|
|
23
|
|
|
self.benchmark = Utility.get_benchmark(benchmark) |
|
24
|
|
|
self.D = D # dimension |
|
25
|
|
|
self.NP = NP # population size |
|
26
|
|
|
self.nFES = nFES # number of function evaluations |
|
27
|
|
|
self.A = A # loudness |
|
28
|
|
|
self.r = r # pulse rate |
|
29
|
|
|
self.Qmin = Qmin # frequency min |
|
30
|
|
|
self.Qmax = Qmax # frequency max |
|
31
|
|
|
self.Lower = self.benchmark.Lower # lower bound |
|
32
|
|
|
self.Upper = self.benchmark.Upper # upper bound |
|
33
|
|
|
self.Fun = self.benchmark.function() |
|
34
|
|
|
|
|
35
|
|
|
self.f_min = 0.0 # minimum fitness |
|
36
|
|
|
|
|
37
|
|
|
self.Lb = [0] * self.D # lower bound |
|
38
|
|
|
self.Ub = [0] * self.D # upper bound |
|
39
|
|
|
self.Q = [0] * self.NP # frequency |
|
40
|
|
|
|
|
41
|
|
|
self.v = [[0 for _i in range(self.D)] |
|
42
|
|
|
for _j in range(self.NP)] # velocity |
|
43
|
|
|
self.Sol = [[0 for _i in range(self.D)] for _j in range( |
|
44
|
|
|
self.NP)] # population of solutions |
|
45
|
|
|
self.Fitness = [0] * self.NP # fitness |
|
46
|
|
|
self.best = [0] * self.D # best solution |
|
47
|
|
|
self.evaluations = 0 # evaluations counter |
|
48
|
|
|
|
|
49
|
|
|
self.F = 0.5 |
|
50
|
|
|
self.CR = 0.9 |
|
51
|
|
View Code Duplication |
|
|
|
|
|
|
|
52
|
|
|
def best_bat(self): |
|
53
|
|
|
i = 0 |
|
54
|
|
|
j = 0 |
|
55
|
|
|
for i in range(self.NP): |
|
56
|
|
|
if self.Fitness[i] < self.Fitness[j]: |
|
57
|
|
|
j = i |
|
58
|
|
|
for i in range(self.D): |
|
59
|
|
|
self.best[i] = self.Sol[j][i] |
|
60
|
|
|
self.f_min = self.Fitness[j] |
|
61
|
|
View Code Duplication |
|
|
|
|
|
|
|
62
|
|
|
def init_bat(self): |
|
63
|
|
|
for i in range(self.D): |
|
64
|
|
|
self.Lb[i] = self.Lower |
|
65
|
|
|
self.Ub[i] = self.Upper |
|
66
|
|
|
|
|
67
|
|
|
for i in range(self.NP): |
|
68
|
|
|
self.Q[i] = 0 |
|
69
|
|
|
for j in range(self.D): |
|
70
|
|
|
rnd = random.uniform(0, 1) |
|
71
|
|
|
self.v[i][j] = 0.0 |
|
72
|
|
|
self.Sol[i][j] = self.Lb[j] + (self.Ub[j] - self.Lb[j]) * rnd |
|
73
|
|
|
self.Fitness[i] = self.Fun(self.D, self.Sol[i]) |
|
74
|
|
|
self.evaluations = self.evaluations + 1 |
|
75
|
|
|
self.best_bat() |
|
76
|
|
|
|
|
77
|
|
|
@classmethod |
|
78
|
|
|
def simplebounds(cls, val, lower, upper): |
|
79
|
|
|
if val < lower: |
|
80
|
|
|
val = lower |
|
81
|
|
|
if val > upper: |
|
82
|
|
|
val = upper |
|
83
|
|
|
return val |
|
84
|
|
|
|
|
85
|
|
|
def move_bat(self): |
|
86
|
|
|
S = [[0.0 for i in range(self.D)] for j in range(self.NP)] |
|
87
|
|
|
|
|
88
|
|
|
self.init_bat() |
|
89
|
|
|
|
|
90
|
|
|
while True: |
|
91
|
|
|
if self.evaluations == self.nFES: |
|
92
|
|
|
break |
|
93
|
|
|
|
|
94
|
|
|
for i in range(self.NP): |
|
95
|
|
|
rnd = random.uniform(0, 1) |
|
96
|
|
|
self.Q[i] = self.Qmin + (self.Qmin - self.Qmax) * rnd |
|
97
|
|
|
j = None |
|
98
|
|
|
for j in range(self.D): |
|
99
|
|
|
self.v[i][j] = self.v[i][j] + (self.Sol[i][j] - |
|
100
|
|
|
self.best[j]) * self.Q[i] |
|
101
|
|
|
S[i][j] = self.Sol[i][j] + self.v[i][j] |
|
102
|
|
|
|
|
103
|
|
|
S[i][j] = self.simplebounds(S[i][j], self.Lb[j], |
|
104
|
|
|
self.Ub[j]) |
|
105
|
|
|
|
|
106
|
|
|
rnd = random.random() |
|
107
|
|
|
|
|
108
|
|
|
if rnd > self.r: |
|
109
|
|
|
if random.random() < self.CR: |
|
110
|
|
|
naklj = random.sample(range(0, self.NP), 3) |
|
111
|
|
|
S[i][j] = S[naklj[0]][j] + self.F * \ |
|
112
|
|
|
(S[naklj[1]][j] - S[naklj[2]][j]) |
|
113
|
|
|
S[i][j] = self.simplebounds(S[i][j], self.Lb[j], |
|
114
|
|
|
self.Ub[j]) |
|
115
|
|
|
|
|
116
|
|
|
Fnew = self.Fun(self.D, S[i]) |
|
117
|
|
|
self.evaluations = self.evaluations + 1 |
|
118
|
|
|
|
|
119
|
|
|
rnd = random.random() |
|
120
|
|
|
|
|
121
|
|
|
if (Fnew <= self.Fitness[i]) and (rnd < self.A): |
|
122
|
|
|
for j in range(self.D): |
|
123
|
|
|
self.Sol[i][j] = S[i][j] |
|
124
|
|
|
self.Fitness[i] = Fnew |
|
125
|
|
|
|
|
126
|
|
|
if Fnew <= self.f_min: |
|
127
|
|
|
for j in range(self.D): |
|
128
|
|
|
self.best[j] = S[i][j] |
|
129
|
|
|
self.f_min = Fnew |
|
130
|
|
|
|
|
131
|
|
|
return self.f_min |
|
132
|
|
|
|
|
133
|
|
|
def run(self): |
|
134
|
|
|
return self.move_bat() |
|
135
|
|
|
|
NiaOrg /
NiaPy
Loading history...