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
这是初始排序后生成排列的渐近最优方式O(n*n!)。
有n个!最多进行一次置换,且具有下一次置换(..),以O(n)时间复杂度运行
在3个步骤中,
找到最大的j,使a[j]可以增加以最小可行量增加a[j]找到扩展新a[0..j]的字典最少方法
'''
Lexicographic permutation generation
consider example array state of [1,5,6,4,3,2] for sorted [1,2,3,4,5,6]
after 56432(treat as number) ->nothing larger than 6432(using 6,4,3,2) beginning with 5
so 6 is next larger and 2345(least using numbers other than 6)
so [1, 6,2,3,4,5]
'''
def hasNextPermutation(array, len):
' Base Condition '
if(len ==1):
return False
'''
Set j = last-2 and find first j such that a[j] < a[j+1]
If no such j(j==-1) then we have visited all permutations
after this step a[j+1]>=..>=a[len-1] and a[j]<a[j+1]
a[j]=5 or j=1, 6>5>4>3>2
'''
j = len -2
while (j >= 0 and array[j] >= array[j + 1]):
j= j-1
if(j==-1):
return False
# print(f"After step 2 for j {j} {array}")
'''
decrease l (from n-1 to j) repeatedly until a[j]<a[l]
Then swap a[j], a[l]
a[l] is the smallest element > a[j] that can follow a[l]...a[j-1] in permutation
before swap we have a[j+1]>=..>=a[l-1]>=a[l]>a[j]>=a[l+1]>=..>=a[len-1]
after swap -> a[j+1]>=..>=a[l-1]>=a[j]>a[l]>=a[l+1]>=..>=a[len-1]
a[l]=6 or l=2, j=1 just before swap [1, 5, 6, 4, 3, 2]
after swap [1, 6, 5, 4, 3, 2] a[l]=5, a[j]=6
'''
l = len -1
while(array[j] >= array[l]):
l = l-1
# print(f"After step 3 for l={l}, j={j} before swap {array}")
array[j], array[l] = array[l], array[j]
# print(f"After step 3 for l={l} j={j} after swap {array}")
'''
Reverse a[j+1...len-1](both inclusive)
after reversing [1, 6, 2, 3, 4, 5]
'''
array[j+1:len] = reversed(array[j+1:len])
# print(f"After step 4 reversing {array}")
return True
array = [1,2,4,4,5]
array.sort()
len = len(array)
count =1
print(array)
'''
The algorithm visits every permutation in lexicographic order
generating one by one
'''
while(hasNextPermutation(array, len)):
print(array)
count = count +1
# The number of permutations will be n! if no duplicates are present, else less than that
# [1,4,3,3,2] -> 5!/2!=60
print(f"Number of permutations: {count}")
其他回答
为了节省您可能的搜索和实验时间,下面是Python中的非递归置换解决方案,它也适用于Numba(从0.41版开始):
@numba.njit()
def permutations(A, k):
r = [[i for i in range(0)]]
for i in range(k):
r = [[a] + b for a in A for b in r if (a in b)==False]
return r
permutations([1,2,3],3)
[[1, 2, 3], [1, 3, 2], [2, 1, 3], [2, 3, 1], [3, 1, 2], [3, 2, 1]]
要给人留下绩效印象:
%timeit permutations(np.arange(5),5)
243 µs ± 11.1 µs per loop (mean ± std. dev. of 7 runs, 1 loop each)
time: 406 ms
%timeit list(itertools.permutations(np.arange(5),5))
15.9 µs ± 8.61 ns per loop (mean ± std. dev. of 7 runs, 100000 loops each)
time: 12.9 s
因此,只有在必须从njit函数调用它时才使用此版本,否则更倾向于itertools实现。
此解决方案实现了一个生成器,以避免在内存中保留所有排列:
def permutations (orig_list):
if not isinstance(orig_list, list):
orig_list = list(orig_list)
yield orig_list
if len(orig_list) == 1:
return
for n in sorted(orig_list):
new_list = orig_list[:]
pos = new_list.index(n)
del(new_list[pos])
new_list.insert(0, n)
for resto in permutations(new_list[1:]):
if new_list[:1] + resto <> orig_list:
yield new_list[:1] + resto
我使用了一种基于阶乘数系统的算法——对于长度为n的列表,您可以逐项组装每个排列,从每个阶段留下的项目中进行选择。第一项有n个选项,第二项有n-1个选项,最后一项只有一个选项,因此可以使用阶乘数系统中数字的数字作为索引。这是数字0到n-1对应于词典顺序中的所有可能的排列。
from math import factorial
def permutations(l):
permutations=[]
length=len(l)
for x in xrange(factorial(length)):
available=list(l)
newPermutation=[]
for radix in xrange(length, 0, -1):
placeValue=factorial(radix-1)
index=x/placeValue
newPermutation.append(available.pop(index))
x-=index*placeValue
permutations.append(newPermutation)
return permutations
permutations(range(3))
输出:
[[0, 1, 2], [0, 2, 1], [1, 0, 2], [1, 2, 0], [2, 0, 1], [2, 1, 0]]
此方法是非递归的,但在我的计算机上速度稍慢,xrange在n!太大,无法转换为C长整数(我的n=13)。当我需要它的时候,它已经足够了,但它远没有itertools.permutations。
该算法是最有效的算法,它避免了递归调用中的数组传递和操作,适用于Python 2、3:
def permute(items):
length = len(items)
def inner(ix=[]):
do_yield = len(ix) == length - 1
for i in range(0, length):
if i in ix: #avoid duplicates
continue
if do_yield:
yield tuple([items[y] for y in ix + [i]])
else:
for p in inner(ix + [i]):
yield p
return inner()
用法:
for p in permute((1,2,3)):
print(p)
(1, 2, 3)
(1, 3, 2)
(2, 1, 3)
(2, 3, 1)
(3, 1, 2)
(3, 2, 1)
使用计数器
from collections import Counter
def permutations(nums):
ans = [[]]
cache = Counter(nums)
for idx, x in enumerate(nums):
result = []
for items in ans:
cache1 = Counter(items)
for id, n in enumerate(nums):
if cache[n] != cache1[n] and items + [n] not in result:
result.append(items + [n])
ans = result
return ans
permutations([1, 2, 2])
> [[1, 2, 2], [2, 1, 2], [2, 2, 1]]