-
Notifications
You must be signed in to change notification settings - Fork 0
/
genetic.py
185 lines (150 loc) · 5.85 KB
/
genetic.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
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
from numpy import cumsum
from numpy.random import choice, rand
from random import randint, random, uniform, choices
class Agent():
def __init__(self, length):
self.length = length
self.value = choice([0, 1], size=(self.length,), p=[8/9, 1/9]).tolist()
self.fitness = -1
def __str__(self):
return 'Value: ' + str(self.value) + ' Fitness: ' + str(self.fitness) + '\n'
class Population():
def __init__(self, pop_size, bits, pc, pm):
self.pop_size = pop_size
self.bits = bits
self.pc = pc
self.pm = pm
self.population = [Agent(self.bits) for _ in range(self.pop_size)]
def __str__(self):
string = ''
for agent in self.population:
string += 'Value: ' + str(agent.value) + ' Fitness: ' + str(agent.fitness) + '\n'
return string
def cost_roullete(self):
# total fitness
total_fitness = sum(a.fitness for a in self.population)
# probs of each agent
probs = [a.fitness/total_fitness for a in self.population]
# accumulative probs
q_probs = cumsum(probs).tolist()
new_gen = []
# spin roullete pop_size times
for _ in range(self.pop_size):
# generate prob
p = uniform(0, 1)
# find correct slot
for j in range(self.pop_size):
if p <= q_probs[j]:
new_gen.append(self.population[j])
break
self.population = []
self.population = new_gen[:]
return self
def rank_roullete(self, q_probs=[]):
# sort by best
agents = sorted(self.population, key=lambda a: a.fitness, reverse=True)
denom = sum(range(1, self.pop_size))
q_probs = cumsum([(self.pop_size-i+1)/denom for i in range(1, self.pop_size+1)]).tolist()
new_gen = []
# spin roullete pop_size times
for _ in range(self.pop_size):
# generate prob
p = uniform(0, 1)
# find correct slot
for j in range(self.pop_size):
if p <= q_probs[j]:
new_gen.append(agents[j])
break
self.population = []
self.population = new_gen[:]
return self
def tournament(self):
new_gen = []
for _ in range(self.pop_size):
fight = choices(self.population, k=randint(2, self.pop_size//2))
best_agent = sorted(fight, key=lambda a: a.fitness, reverse=True)[0]
new_gen.append(best_agent)
self.population = new_gen[:]
def selection(self, select='cost_roullete'):
if select == 'cost_roullete':
self.cost_roullete()
elif select == 'rank_roullete':
self.rank_roullete()
elif select == 'tournament':
self.tournament()
else:
raise('Not valid option!')
def single_point_cross(self):
offspring = []
for i in range(0, self.pop_size, 2):
if random() < self.pc:
# pick pairs
bs1, bs2 = self.population[i], self.population[i+1]
# random index to slice
pt = randint(1, self.bits-2)
# create ancestors
child1 = Agent(self.bits)
child2 = Agent(self.bits)
# give them the values
child1.value = bs1.value[:pt] + bs2.value[pt:]
child2.value = bs2.value[:pt] + bs1.value[pt:]
# append offsprings
offspring.append(child1)
offspring.append(child2)
else:
offspring.append(self.population[i])
offspring.append(self.population[i+1])
self.population = offspring[:]
def multi_point_cross(self, N):
for _ in range(N):
self.single_point_cross()
def uniform_cross(self):
offspring = []
for i in range(0, self.pop_size, 2):
# pick pairs
bs1, bs2 = self.population[i], self.population[i+1]
if random() < self.pc:
# create ancestors
child1 = Agent(self.bits)
child2 = Agent(self.bits)
probs = rand(self.pop_size)
for i in range(len(probs)):
if probs[i] < 0.5:
temp = bs1.value[i]
bs1.value[i] = bs2.value[i]
bs2.value[i] = temp
# give them the values
child1.value = bs1.value
child2.value = bs2.value
# append offsprings
offspring.append(child1)
offspring.append(child2)
else:
offspring.append(self.population[i])
offspring.append(self.population[i+1])
self.population = offspring[:]
def crossover(self, select='single', N=50):
if select == 'single':
self.single_point_cross()
elif select == 'multi':
for _ in range(N):
self.single_point_cross()
elif select == 'uniform':
self.uniform_cross()
else:
raise('Not valid selection option!')
return self
def mutation(self):
best = self.get_fittest()
for agent in self.population:
if agent != best:
for i in range(self.bits):
if random() <= self.pm:
agent.value[i] = 1 - agent.value[i]
return self
def get_fittest(self):
index = 0
for i in range(1, len(self.population)):
if self.population[i].fitness > self.population[index].fitness:
index = i
return self.population[index]