1
0

Delete article

Deleted articles cannot be recovered.

Draft of this article would be also deleted.

Are you sure you want to delete this article?

More than 5 years have passed since last update.

スターリンソートをまともに使えるようにしてみた(オーダわかんね)

Last updated at Posted at 2019-07-31

計算量O(n)の画期的なソートアルゴリズムであるスターリンソートをHaskell で実装してみた #Haskell - Qiita

が面白そうだったのでまともなソートに使えないか考えてみました

考え方

  1. 粛清する
  2. 粛清された人を生き返らせてまた粛清する
  3. 1の生き残りと2の生き残りをマージする
  4. 2,3を粛清された人が0になるまで繰り返す

オーダー

うーんわからない(笑)...

マージ部分で$O(N^2)$はかかってる気がするので$O(N^2logN)$とかいう恐ろしい予想が立ちます

誰か考えて()

Pythonで実装

from functools import reduce

def teppop(comp_method=int.__le__):
    def _teppop(acc, elm):
        # acc = (sorted, purged)
        if len(acc[0]) == 0 or comp_method(acc[0][-1], elm):
            acc[0].append(elm) 
        else: # 粛清
            acc[1].append(elm)
        return acc
    return _teppop

def marge_lis(a, b, comp_method=int.__le__):
    res = []
    a_len, b_len = len(a), len(b)
    while a_len > 0 and b_len > 0:
        if comp_method(a[0], b[0]):
            res.append(a.pop(0))
            a_len -= 1
        else:
            res.append(b.pop(0))
            b_len -= 1
    res += b if a_len == 0 else a
    return res

def pure_stalin_sort(lis, reversed=False):
    cm = type(lis[0]).__ge__ if reversed else type(lis[0]).__le__
    return reduce(teppop(comp_method=cm), lis, [[], []])[0]

def stalin_sort(lis, reversed=False):
    cm = type(lis[0]).__ge__ if reversed else type(lis[0]).__le__
    res, purged = reduce(teppop(comp_method=cm), lis, [[], []])
    while len(purged) > 0:
        tmp, purged = reduce(teppop(comp_method=cm), purged, [[], []])
        res = marge_lis(res, tmp, comp_method=cm)
    return res

if __name__=="__main__":
    test_lis = [1, 2, 1, 1, 4, 3, 9]
    print(test_lis, "=>", stalin_sort(test_lis))
    from random import randint
    for _ in range(5):
        test_lis = [randint(1,20) for _ in range(randint(10, 20))]
        res = stalin_sort(test_lis)
        assert sorted(test_lis) == res
        print(test_lis, "=>", res)

実行結果例

[1, 2, 1, 1, 4, 3, 9] => [1, 1, 1, 2, 3, 4, 9]
[6, 9, 12, 13, 10, 17, 3, 20, 15, 5, 18, 2] => [2, 3, 5, 6, 9, 10, 12, 13, 15, 17, 18, 20]
[4, 6, 18, 16, 2, 10, 16, 10, 2, 17, 11, 9, 4, 15, 4, 19, 7] => [2, 2, 4, 4, 4, 6, 7, 9, 10, 10, 11, 15, 16, 16, 17, 18, 19]
[9, 19, 15, 12, 4, 17, 1, 3, 12, 18, 15, 12, 17, 10, 6] => [1, 3, 4, 6, 9, 10, 12, 12, 12, 15, 15, 17, 17, 18, 19]
[8, 19, 3, 9, 17, 20, 3, 16, 18, 10, 19] => [3, 3, 8, 9, 10, 16, 17, 18, 19, 19, 20]                                        4, 15, 20]
[8, 3, 6, 14, 12, 14, 12, 15, 20, 4, 5, 13, 3, 7, 1, 12, 1, 8, 2] => [1, 1, 2, 3, 3, 4, 5, 6, 7, 8, 8, 12, 12, 12, 13, 14, 14, 15, 20]

粛清とreduceをかけてみました(何も面白くない)

余談

オリジナルスターリンソート、Rubyだと簡潔に書けます。

def stalin_sort(lis)
    $top = lis[0]
    lis.select {|elm|
        if $top <= elm then $top = elm end
    }
end

p stalin_sort([1, 2, 1, 1, 4, 3, 9]) # => [1, 2, 4, 9]

闇を使うとPythonでも比較的短くなる...?

def stalin_sort(lis):
    b = {"top": lis[0]}
    return list(filter(lambda elm: not(b["top"] > elm or b.__setitem__("top",elm)), lis))

print(stalin_sort([1, 2, 1, 1, 4, 3, 9])) #=> [1, 2, 4, 9]

最後に

ゴミ記事を量産しまして誠に申し訳ございませんでした。m(_ _)m

1
0
0

Register as a new user and use Qiita more conveniently

  1. You get articles that match your needs
  2. You can efficiently read back useful information
  3. You can use dark theme
What you can do with signing up
1
0

Delete article

Deleted articles cannot be recovered.

Draft of this article would be also deleted.

Are you sure you want to delete this article?