|
1
|
|
|
import random as rnd |
|
2
|
|
|
import copy |
|
3
|
|
|
import numpy as npx |
|
4
|
|
|
|
|
5
|
|
|
__all__ = ['CuckooSearchAlgorithm'] |
|
6
|
|
|
|
|
7
|
|
|
|
|
8
|
|
|
class Cuckoo(object): |
|
9
|
|
|
"""Defines cuckoo for population.""" |
|
10
|
|
|
|
|
11
|
|
|
def __init__(self, D, LB, UB): |
|
12
|
|
|
self.D = D # dimension of the problem |
|
13
|
|
|
self.LB = LB # lower bound |
|
14
|
|
|
self.UB = UB # upper bound |
|
15
|
|
|
self.Solution = [] |
|
16
|
|
|
|
|
17
|
|
|
self.Fitness = float('inf') |
|
18
|
|
|
self.generateCuckoo() |
|
19
|
|
|
|
|
20
|
|
|
def generateCuckoo(self): |
|
21
|
|
View Code Duplication |
self.Solution = [self.LB + (self.UB - self.LB) * rnd.random() |
|
|
|
|
|
|
22
|
|
|
for _i in range(self.D)] |
|
23
|
|
|
|
|
24
|
|
|
def evaluate(self): |
|
25
|
|
|
self.Fitness = Cuckoo.FuncEval(self.D, self.Solution) |
|
26
|
|
|
|
|
27
|
|
|
def simpleBound(self): |
|
28
|
|
|
for i in range(self.D): |
|
29
|
|
|
if self.Solution[i] < self.LB: |
|
30
|
|
|
self.Solution[i] = self.LB |
|
31
|
|
|
if self.Solution[i] > self.UB: |
|
32
|
|
|
self.Solution[i] = self.UB |
|
33
|
|
|
|
|
34
|
|
|
def toString(self): |
|
35
|
|
|
pass |
|
36
|
|
|
|
|
37
|
|
|
def __eq__(self, other): |
|
38
|
|
|
return self.Solution == other.Solution and self.Fitness == other.Fitness |
|
39
|
|
|
|
|
40
|
|
|
|
|
41
|
|
|
class CuckooSearchAlgorithm(object): |
|
42
|
|
|
"""Cuckoo Search algorithm. |
|
43
|
|
|
|
|
44
|
|
|
Date: 12. 2. 2018 |
|
45
|
|
|
|
|
46
|
|
|
Authors : Uros Mlakar |
|
47
|
|
|
|
|
48
|
|
|
License: MIT |
|
49
|
|
|
|
|
50
|
|
|
Reference paper: Yang, Xin-She, and Suash Deb. "Cuckoo search via Lévy flights." |
|
|
|
|
|
|
51
|
|
|
Nature & Biologically Inspired Computing, 2009. NaBIC 2009. |
|
|
|
|
|
|
52
|
|
|
|
|
53
|
|
|
TODO: Tests and validation! |
|
54
|
|
|
""" |
|
55
|
|
View Code Duplication |
|
|
|
|
|
|
|
56
|
|
|
def __init__(self, Np, D, nFES, Pa, Alpha, Lower, Upper, function): |
|
57
|
|
|
self.Np = Np |
|
58
|
|
|
self.D = D |
|
59
|
|
|
self.Pa = Pa |
|
60
|
|
|
self.Lower = Lower |
|
61
|
|
|
self.Upper = Upper |
|
62
|
|
|
self.Nests = [] |
|
63
|
|
|
self.nFES = nFES |
|
64
|
|
|
self.FEs = 0 |
|
65
|
|
|
self.Done = False |
|
66
|
|
|
self.Alpha = Alpha |
|
67
|
|
|
self.Beta = 1.5 |
|
68
|
|
|
Cuckoo.FuncEval = staticmethod(function) |
|
69
|
|
|
|
|
70
|
|
|
self.gBest = Cuckoo(self.D, self.Lower, self.Upper) |
|
71
|
|
|
|
|
72
|
|
|
def evalNests(self): |
|
73
|
|
|
for c in self.Nests: |
|
74
|
|
|
c.evaluate() |
|
75
|
|
|
if c.Fitness < self.gBest.Fitness: |
|
76
|
|
|
self.gBest = copy.deepcopy(c) |
|
77
|
|
|
|
|
78
|
|
|
def initNests(self): |
|
79
|
|
|
for _i in range(self.Np): |
|
80
|
|
|
self.Nests.append(Cuckoo(self.D, self.Lower, self.Upper)) |
|
81
|
|
|
|
|
82
|
|
|
def levyFlight(self, c): |
|
83
|
|
|
sigma = 0.6966 |
|
84
|
|
|
u = npx.random.randn(1, self.D) * sigma |
|
85
|
|
|
v = npx.random.randn(1, self.D) |
|
86
|
|
|
step = u / (abs(v)**(1 / self.Beta)) |
|
87
|
|
|
stepsize = self.Alpha * step * (npx.array(c.Solution) - npx.array(self.gBest.Solution)).flatten().tolist() |
|
88
|
|
|
|
|
89
|
|
|
c.Solution = (npx.array(c.Solution) + npx.array(stepsize) * npx.random.randn(1, self.D)).flatten().tolist() |
|
90
|
|
|
|
|
91
|
|
|
def tryEval(self, c): |
|
92
|
|
|
if self.FEs <= self.nFES: |
|
93
|
|
|
c.evaluate() |
|
94
|
|
|
self.FEs += 1 |
|
95
|
|
|
else: |
|
96
|
|
|
self.Done = True |
|
97
|
|
|
|
|
98
|
|
|
def moveNests(self, Nests): |
|
99
|
|
|
MovedNests = [] |
|
100
|
|
|
for c in Nests: |
|
101
|
|
|
self.levyFlight(c) |
|
102
|
|
|
c.simpleBound() |
|
103
|
|
|
self.tryEval(c) |
|
104
|
|
|
|
|
105
|
|
|
if c.Fitness < self.gBest.Fitness: |
|
106
|
|
|
self.gBest = copy.deepcopy(c) |
|
107
|
|
|
|
|
108
|
|
|
MovedNests.append(c) |
|
109
|
|
|
return MovedNests |
|
110
|
|
|
|
|
111
|
|
|
def resetNests(self, MovedNests): |
|
112
|
|
|
for _i in range(self.Np): |
|
113
|
|
|
if rnd.random() < self.Pa: |
|
114
|
|
|
m = rnd.randint(0, self.Np - 1) |
|
115
|
|
|
n = rnd.randint(0, self.Np - 1) |
|
116
|
|
|
|
|
117
|
|
|
newSolution = npx.array(MovedNests[_i].Solution) + ( |
|
118
|
|
|
rnd.random() * (npx.array(MovedNests[m].Solution) - npx.array(MovedNests[n].Solution))).flatten().tolist() |
|
119
|
|
|
MovedNests[_i].Solution = newSolution |
|
120
|
|
|
MovedNests[_i].simpleBound() |
|
121
|
|
|
self.tryEval(MovedNests[_i]) |
|
122
|
|
|
return MovedNests |
|
123
|
|
|
|
|
124
|
|
|
def run(self): |
|
125
|
|
|
self.initNests() |
|
126
|
|
|
self.evalNests() |
|
127
|
|
|
self.FEs += self.Np |
|
128
|
|
|
while not self.Done: |
|
129
|
|
|
MovedNests = self.moveNests(self.Nests) |
|
130
|
|
|
self.Nests = self.resetNests(MovedNests) |
|
131
|
|
|
|
|
132
|
|
|
return self.gBest.Fitness |
|
133
|
|
|
|
NiaOrg /
NiaPy
Loading history...
