[英]Cross-over is failing binary encoded string
我正在實施 G.netic 算法 (GA)。 有43
個號碼[救護車位置]可供選擇(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)
, 我選擇了3
地方,因為我有3
輛救護車.
我只能將我的救護車停在1-39 locations
個地點中的3
地點(限制) 。
染色體樣本: [000010000000000000100000000000100000000]
。 這代表我要把我的Ambulance放在5th, 19th, and 31 positions.
第5th
位、第19th
位和第31
位的位是1
,而 rest 位是0
。 換句話說,我正在打開5-bit, 19-bit, and 31-bit
。
比方說
Parent1 (111000000000000000000000000000000000000)
和
Parent2 (000000000000000000000000000000000000111)
cross-over
,我得到這個:
('111000000000000000000000000000000000111', '000000000000000000000000000000000000000')
在第一個后代中,我有six 1's
,在第二個后代中,我Zero 1's
。 這些generated off-springs
對我來說是illegal
的,因為我只需要three
1's
后代字符串。
我正在使用單點交叉。 這是我的代碼:
from typing import Union
import random
Parent 1 ="111000000000000000000000000000000000000"
Parent 2 ="000000000000000000000000000000000000111"
def crossover(cs1: str, cs2: str) -> Union[str, str]:
index: int = random.randint(0, len(cs1))
return cs1[:index] + cs2[index:], cs2[:index] + cs1[index:]
crossover(Cs1,Cs2)
在保持1-39 bits
中的3
位的同時執行交叉的好方法是什么?
IIUC,你想隨機混合兩個父母,正好保持 3'1's?
您可以隨機獲取每個父項中的 1 和 select 的索引:
import random
Parent1 ="111000000000000000000000000000000000000"
Parent2 ="000000000000000000000000000000000000111"
def indices(s):
return {i for i,c in enumerate(s) if c=='1'}
def crossover(p1, p2, k=3):
idx = set(random.sample(list(indices(p1) | indices(p2)), k=k))
return ''.join('1' if i in idx else '0' for i in range(len(p1)))
out = crossover(Parent1, Parent2, k=Parent1.count('1'))
# '110000000000000000000000000000000000100'
如果您想為兩個字符串中均為 1 的 position 賦予更多權重,您可以修改上面的內容以使用Counter
代替集合:
import random
from collections import Counter
Parent1 ="111000000000000000000000000000000000000"
Parent2 ="000000000000000000000000000000000000111"
def indices(s):
return Counter(i for i,c in enumerate(s) if c=='1')
def crossover(p1, p2, k=3):
# count the number of 1 per position
pool = indices(p1) | indices(p2)
# randomly select indices
# using the counts as weights
idx = set(random.sample(list(pool),
counts=pool.values(),
k=k))
return ''.join('1' if i in idx else '0' for i in range(len(p1)))
out = crossover(Parent1, Parent2, k=Parent1.count('1'))
# '010000000000000000000000000000000000101'
使用集合操作
import random
def indices(s):
return {i for i,c in enumerate(s) if c=='1'}
def crossover(p1, p2):
# get positions of 1s for each string
idx1 = indices(p1)
idx2 = indices(p2)
# positions that are different in both strings
differ = idx1.symmetric_difference(idx2)
# identical positions
common = idx1&idx2
# pick half of the different positions randomly
select = set(random.sample(list(differ), k=len(differ)//2))
# offspring 1 get those positions + the common ones
select1 = select | common
# offstring 2 gets the other positions + the common ones
select2 = (differ-select) | common
# make strings from the selected positions for each offspring
out1 = ''.join('1' if i in select1 else '0' for i in range(len(p1)))
out2 = ''.join('1' if i in select2 else '0' for i in range(len(p1)))
return out1, out2
crossover(Parent1, Parent2)
例如 output:
('101000000000000000000000000000000000001',
'010000000000000000000000000000000000110')
聲明:本站的技術帖子網頁,遵循CC BY-SA 4.0協議,如果您需要轉載,請注明本站網址或者原文地址。任何問題請咨詢:yoyou2525@163.com.