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
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。
其他回答
这是受Haskell实现使用列表理解的启发:
def permutation(list):
if len(list) == 0:
return [[]]
else:
return [[x] + ys for x in list for ys in permutation(delete(list, x))]
def delete(list, item):
lc = list[:]
lc.remove(item)
return lc
from typing import List
import time, random
def measure_time(func):
def wrapper_time(*args, **kwargs):
start_time = time.perf_counter()
res = func(*args, **kwargs)
end_time = time.perf_counter()
return res, end_time - start_time
return wrapper_time
class Solution:
def permute(self, nums: List[int], method: int = 1) -> List[List[int]]:
perms = []
perm = []
if method == 1:
_, time_perm = self._permute_recur(nums, 0, len(nums) - 1, perms)
elif method == 2:
_, time_perm = self._permute_recur_agian(nums, perm, perms)
print(perm)
return perms, time_perm
@measure_time
def _permute_recur(self, nums: List[int], l: int, r: int, perms: List[List[int]]):
# base case
if l == r:
perms.append(nums.copy())
for i in range(l, r + 1):
nums[l], nums[i] = nums[i], nums[l]
self._permute_recur(nums, l + 1, r , perms)
nums[l], nums[i] = nums[i], nums[l]
@measure_time
def _permute_recur_agian(self, nums: List[int], perm: List[int], perms_list: List[List[int]]):
"""
The idea is similar to nestedForLoops visualized as a recursion tree.
"""
if nums:
for i in range(len(nums)):
# perm.append(nums[i]) mistake, perm will be filled with all nums's elements.
# Method1 perm_copy = copy.deepcopy(perm)
# Method2 add in the parameter list using + (not in place)
# caveat: list.append is in-place , which is useful for operating on global element perms_list
# Note that:
# perms_list pass by reference. shallow copy
# perm + [nums[i]] pass by value instead of reference.
self._permute_recur_agian(nums[:i] + nums[i+1:], perm + [nums[i]], perms_list)
else:
# Arrive at the last loop, i.e. leaf of the recursion tree.
perms_list.append(perm)
if __name__ == "__main__":
array = [random.randint(-10, 10) for _ in range(3)]
sol = Solution()
# perms, time_perm = sol.permute(array, 1)
perms2, time_perm2 = sol.permute(array, 2)
print(perms2)
# print(perms, perms2)
# print(time_perm, time_perm2)
```
对于Python 2.6及以上版本:
import itertools
itertools.permutations([1, 2, 3])
这将作为生成器返回。使用列表(排列(xs))作为列表返回。
生成所有可能的排列
我正在使用python3.4:
def calcperm(arr, size):
result = set([()])
for dummy_idx in range(size):
temp = set()
for dummy_lst in result:
for dummy_outcome in arr:
if dummy_outcome not in dummy_lst:
new_seq = list(dummy_lst)
new_seq.append(dummy_outcome)
temp.add(tuple(new_seq))
result = temp
return result
测试用例:
lst = [1, 2, 3, 4]
#lst = ["yellow", "magenta", "white", "blue"]
seq = 2
final = calcperm(lst, seq)
print(len(final))
print(final)
人们确实可以对每个排列的第一个元素进行迭代,正如tzwen的答案。然而,这样编写此解决方案更有效:
def all_perms(elements):
if len(elements) <= 1:
yield elements # Only permutation possible = no permutation
else:
# Iteration over the first element in the result permutation:
for (index, first_elmt) in enumerate(elements):
other_elmts = elements[:index]+elements[index+1:]
for permutation in all_perms(other_elmts):
yield [first_elmt] + permutation
这个解决方案大约快了30%,显然是因为递归以len(元素)<=1而不是0结尾。它的内存效率也高得多,因为它使用了一个生成器函数(通过yield),就像Riccardo Reyes的解决方案一样。