如何生成列表的所有排列?例如:
permutations([])
[]
permutations([1])
[1]
permutations([1, 2])
[1, 2]
[2, 1]
permutations([1, 2, 3])
[1, 2, 3]
[1, 3, 2]
[2, 1, 3]
[2, 3, 1]
[3, 1, 2]
[3, 2, 1]
如何生成列表的所有排列?例如:
permutations([])
[]
permutations([1])
[1]
permutations([1, 2])
[1, 2]
[2, 1]
permutations([1, 2, 3])
[1, 2, 3]
[1, 3, 2]
[2, 1, 3]
[2, 3, 1]
[3, 1, 2]
[3, 2, 1]
当前回答
def permutation(word, first_char=None):
if word == None or len(word) == 0: return []
if len(word) == 1: return [word]
result = []
first_char = word[0]
for sub_word in permutation(word[1:], first_char):
result += insert(first_char, sub_word)
return sorted(result)
def insert(ch, sub_word):
arr = [ch + sub_word]
for i in range(len(sub_word)):
arr.append(sub_word[i:] + ch + sub_word[:i])
return arr
assert permutation(None) == []
assert permutation('') == []
assert permutation('1') == ['1']
assert permutation('12') == ['12', '21']
print permutation('abc')
输出:['abc','acb','bac','bca','cab','cba']
其他回答
注意,该算法具有n个阶乘时间复杂度,其中n是输入列表的长度
打印跑步结果:
global result
result = []
def permutation(li):
if li == [] or li == None:
return
if len(li) == 1:
result.append(li[0])
print result
result.pop()
return
for i in range(0,len(li)):
result.append(li[i])
permutation(li[:i] + li[i+1:])
result.pop()
例子:
permutation([1,2,3])
输出:
[1, 2, 3]
[1, 3, 2]
[2, 1, 3]
[2, 3, 1]
[3, 1, 2]
[3, 2, 1]
我看到在这些递归函数中进行了很多迭代,而不是纯粹的递归。。。
所以对于那些连一个循环都不能遵守的人来说,这里有一个粗略的、完全不必要的完全递归的解决方案
def all_insert(x, e, i=0):
return [x[0:i]+[e]+x[i:]] + all_insert(x,e,i+1) if i<len(x)+1 else []
def for_each(X, e):
return all_insert(X[0], e) + for_each(X[1:],e) if X else []
def permute(x):
return [x] if len(x) < 2 else for_each( permute(x[1:]) , x[0])
perms = permute([1,2,3])
另一种解决方案:
def permutation(flag, k =1 ):
N = len(flag)
for i in xrange(0, N):
if flag[i] != 0:
continue
flag[i] = k
if k == N:
print flag
permutation(flag, k+1)
flag[i] = 0
permutation([0, 0, 0])
常规实现(无收益-将在内存中完成所有操作):
def getPermutations(array):
if len(array) == 1:
return [array]
permutations = []
for i in range(len(array)):
# get all perm's of subarray w/o current item
perms = getPermutations(array[:i] + array[i+1:])
for p in perms:
permutations.append([array[i], *p])
return permutations
产量实施:
def getPermutations(array):
if len(array) == 1:
yield array
else:
for i in range(len(array)):
perms = getPermutations(array[:i] + array[i+1:])
for p in perms:
yield [array[i], *p]
基本思想是在第一个位置遍历数组中的所有元素,然后在第二个位置遍历所有其他元素,而没有为第一个位置选择的元素,等等。您可以使用递归来实现这一点,其中停止条件是到达一个由1个元素组成的数组,在这种情况下,您返回该数组。
对于性能,一个由Knuth启发的numpy解决方案(第22页):
from numpy import empty, uint8
from math import factorial
def perms(n):
f = 1
p = empty((2*n-1, factorial(n)), uint8)
for i in range(n):
p[i, :f] = i
p[i+1:2*i+1, :f] = p[:i, :f] # constitution de blocs
for j in range(i):
p[:i+1, f*(j+1):f*(j+2)] = p[j+1:j+i+2, :f] # copie de blocs
f = f*(i+1)
return p[:n, :]
复制大量内存可节省时间-它比列表(itertools.permutations(range(n))快20倍:
In [1]: %timeit -n10 list(permutations(range(10)))
10 loops, best of 3: 815 ms per loop
In [2]: %timeit -n100 perms(10)
100 loops, best of 3: 40 ms per loop