-
Notifications
You must be signed in to change notification settings - Fork 41
/
FA.py
166 lines (151 loc) · 5.99 KB
/
FA.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
# ！usr/bin/env python
# -*- coding: utf-8 -*-
# Time : 2021/12/15 15:02
# @Author : LucXiong
# @Project : Model
# @File : FA.py
"""
Ref:https://github.com/GoodLittleStar/Fireworks/blob/master/FireWork.py
Ref:Tan Y, Zhu Y. Fireworks Algorithm for Optimization[M].
Lecture Notes in Computer Science. City: Springer Berlin Heidelberg, 2010: 355-64[2021-12-08T08:42:21].
"""
import random # random Function
import numpy as np # numpy operations
import copy
import matplotlib.pyplot as plt
import math
import test_function
class FA():
def __init__(self, pop_size=50, n_dim=2, m=50, a=0.04, b=0.8, A=40, lb=-1e5, ub=1e5, max_iter=1000, func=None):
self.a = a
self.b = b
self.m = m
self.A = A
self.pop = pop_size
self.dim = n_dim
self.func = func
self.max_iter = max_iter # max iter
self.epsino = 1e-6
self.mutate = 0.1
self.lb, self.ub = np.array(lb) * np.ones(self.dim), np.array(ub) * np.ones(self.dim)
assert self.dim == len(self.lb) == len(self.ub), 'dim == len(lb) == len(ub) is not True'
assert np.all(self.ub > self.lb), 'upper-bound must be greater than lower-bound'
self.X = np.random.uniform(low=self.lb, high=self.ub, size=(self.pop, self.dim))
self.X = self.X.tolist()
self.Y = [self.func(self.X[i]) for i in range(len(self.X))] # y = f(x) for all particles
self.bestindex = self.Y.index(min(self.Y))
self.gbest_x = self.X[self.bestindex]
self.gbest_y = min(self.Y) # global best y for all particles
self.gbest_y_hist = [self.gbest_y] # gbest_y of every iteration
# self.update_gbest()
def CalculateSi(self):
self.MaxFitness = max(self.Y)
temp = 0.
self.Si = []
for i in range(0, self.pop):
temp = temp + self.MaxFitness - self.Y[i]
for i in range(0, self.pop):
self.Si.append(self.m * (self.MaxFitness - self.Y[i] + self.epsino) / (temp + self.epsino))
if self.Si[-1] < self.a * self.m:
self.Si[-1] = round(self.a * self.m)
elif self.Si[-1] > self.b * self.m:
self.Si[-1] = round(self.b * self.m)
else:
self.Si[-1] = round(self.Si[-1])
def CalculateExpo(self):
self.MinFitness = min(self.Y)
temp = 0.
self.Ai = []
for i in range(self.pop):
temp = temp + self.Y[i] - self.MinFitness
for i in range(self.pop):
self.Ai.append(self.A * (self.Y[i]- self.MinFitness + self.epsino) / (temp + self.epsino))
def Explosion(self):
for k in range(0, self.pop):
for i in range(self.Si[k]):
spark = copy.deepcopy(self.X[k])
z = round(self.dim * random.uniform(0, 1))
dim_list = range(self.dim)
rand_z = random.sample(dim_list, z)
h = self.Ai[k] * random.uniform(-1, 1)
for j in rand_z:
spark[j] += h
if spark[j] < self.lb[j] or spark[j] > self.ub[j]:
spark[j] = self.lb[j] + abs(spark[j]) % (self.ub[j] - self.lb[j])
self.X.append(spark)
self.Y.append(self.func(spark))
if(len(self.X) > 5 * self.pop):
break
def Mutation(self):
currentsize = len(self.X)
for k in range(round(self.mutate * currentsize)):
randindex = random.randint(0, currentsize - 1)
spark = copy.deepcopy(self.X[randindex])
# print(spark)
# print(randindex)
z = round(self.dim * random.uniform(0, 1))
dim_list = range(self.dim)
rand_z = random.sample(dim_list, z)
g = random.gauss(1, 1)
for j in rand_z:
spark[j] *= g
if spark[j] < self.lb[j] or spark[j] > self.ub[j]:
spark[j] = self.lb[j] + abs(spark[j]) % (self.ub[j] - self.lb[j])
self.X.append(spark)
self.Y.append(self.func(spark))
if (len(self.X) > 10 * self.pop):
break
def Selection(self):
newpop=[]
newpop.append(self.gbest_x)
self.Ri = []
for i in range(len(self.X)):
dis=0.
for j in range(len(self.X)):
for k in range(self.dim):
dis+= (self.X[i][k]-self.X[j][k])**2
self.Ri.append(math.sqrt(dis))
sr = sum(self.Ri)
px = [self.Ri[i]/sr for i in range(len(self.Ri))]
for i in range(self.pop-1):
rr=random.uniform(0,1)
index=0
for j in range(self.pop):
if j==0 and rr<px[j]:
index=j
elif rr>=px[j] and rr<px[j+1]:
index=j+1
newpop.append(self.X[index])
self.X = newpop
self.Y = [self.func(self.X[i]) for i in range(len(self.X))]
def run(self):
for i in range(self.max_iter):
# print(i)
# print(len(self.X))
self.CalculateSi()
self.CalculateExpo()
self.Explosion()
# print(len(self.X))
self.Mutation()
# print(len(self.X))
bestindex = self.Y.index(min(self.Y))
if self.gbest_y_hist[-1] > self.Y[bestindex]:
self.gbest_y_hist.append(self.Y[bestindex])
self.gbest_x = self.X[bestindex]
else:
self.gbest_y_hist.append(self.gbest_y_hist[-1])
self.Selection()
# print(self.gbest_y_hist[-1])
return self.gbest_x, self.gbest_y_hist[-1]
if __name__ == '__main__':
n_dim = 3
lb = [-100 for i in range(n_dim)]
ub = [100 for i in range(n_dim)]
demo_func = test_function.fm2
pop_size = 10
max_iter = 20
fa = FA(n_dim=n_dim, pop_size=pop_size, max_iter=max_iter, lb=lb, ub=ub, func=demo_func)
best_x, bext_y = fa.run()
print(f'{demo_func(fa.gbest_x)}\t{fa.gbest_x}')
plt.plot(fa.gbest_y_hist)
plt.show()