遗传算法求二元函数极值怎么编码_用遗传算法求复杂函数的极值点

确定遗传

在元素个体，遗传得到的个体和变异个体中选取最好的30个个体(对应的函数值最大的30个个体)作为下一次迭代的父样本。

from random import randint

from numpy import sin

def decode(g):

return [((g&0xfff) - 2048) * 0.001, ((g>>12) - 2048) * 0.001]

def function_g(g):

x = decode(g)

return function(x[0], x[1])

def function(x, y):

return 100 * (sin(x) ** 2 - sin(y)) ** 2 + (1 - sin(x)) ** 2

def cmp(g1, g2):

key = function_g(g1) - function_g(g2)

if key > 0: return 1

elif key < 0: return -1

else: return 0

def GA(num = 30, round = 10):

gene = [randint(0, (1<<24) - 1) for i in range(num)]

rnd = 0

while rnd < round:

rnd += 1

gene_c = [g ^ (1<

gene_h = []

for g1 in gene:

for g2 in gene:

mask = (1<

gene_h.append(g2 & ~mask | g1 & mask)

gene_h.append(g1 & ~mask | g2 & mask)

gene_tot = gene + gene_h + gene_c

gene_tot.sort(cmp = cmp, reverse = True)

gene = gene_tot[:num]

print "round", rnd, ":", decode(gene[0]), function_g(gene[0])

return decode(gene[0]) + [function_g(gene[0])]

if __name__ == '__main__':

print GA(30, 10),

结果

经过10得到如下结果， 每一行的三个数字分别对应x, y, f(x, y)

round 1 : [1.571, -1.464] 397.724305554

round 2 : [1.571, -1.539] 399.797824716

round 3 : [-1.605, -1.548] 403.426161017

round 4 : [-1.605, -1.556] 403.486264841

round 5 : [-1.541, -1.591] 403.561685598

round 6 : [-1.541, -1.575] 403.639747518

round 7 : [-1.573, -1.579] 403.984587994

round 8 : [-1.573, -1.571] 403.998039526

round 9 : [-1.569, -1.571] 403.998694536

round 10 : [-1.569, -1.571] 403.998694536