基于Numpy的MultiPoint交叉

Question

我试着用numpy对遗传算法进行交叉。我用父母1和父母2分割了人口。

population = np.random.randint(2, size=(4,8))
p1 = population[::2]
p2 = population[1::2]

但我无法找出任何蓝精灵或矮胖的命令来做一个多点交叉的父母。其概念是将ith row of p1和随机地与ith row of p2交换一些位。

Answer 1

我认为你想从p1和p2中随机选择，逐个细胞。

为了更容易理解，我将p1更改为10到15，p2更改为20至25。在这些范围内随机生成p1和p2。

p1
Out[66]: 
array([[15, 15, 13, 14, 12, 13, 12, 12],
       [14, 11, 11, 10, 12, 12, 10, 12],
       [12, 11, 14, 15, 14, 10, 13, 10],
       [11, 12, 10, 13, 14, 13, 12, 13]])

In [67]: p2
Out[67]: 
array([[23, 25, 24, 21, 24, 20, 24, 25],
       [21, 21, 20, 20, 25, 22, 24, 22],
       [24, 22, 25, 20, 21, 22, 21, 22],
       [22, 20, 21, 22, 25, 23, 22, 21]])

In [68]: sieve=np.random.randint(2, size=(4,8))
In [69]: sieve
Out[69]: 
array([[0, 1, 0, 1, 1, 0, 1, 0],
       [1, 1, 1, 0, 0, 1, 1, 1],
       [0, 1, 1, 0, 0, 1, 1, 0],
       [0, 0, 0, 1, 1, 1, 1, 1]])
In [70]: not_sieve=sieve^1  # Complement of sieve

In [71]: pn = p1*sieve + p2*not_sieve

In [72]: pn
Out[72]:
array([[23, 15, 24, 14, 12, 20, 12, 25],
       [14, 11, 11, 20, 25, 12, 10, 12],
       [24, 11, 14, 20, 21, 10, 13, 22],
       [22, 20, 21, 13, 14, 13, 12, 13]])

当筛子为1时，青少年的数字来自p1；当筛子为0时，20岁的数字来自p2

这也许可以使效率更高，但这是您期望的输出吗？