希尔排序

希尔排序(Shell Sort) 是插入排序的一种。也称缩小增量排序，是直接插入排序算法的一种更高效的改进版本。希尔排序是非稳定排序算法。该方法因DL．Shell于1959年提出而得名。 希尔排序是把记录按下标的一定增量分组，对每组使用直接插入排序算法排序；随着增量逐渐减少，每组包含的关键词越来越多，当增量减至1时，整个文件恰被分成一组，算法便终止。

希尔排序过程

希尔排序的基本思想是：将数组列在一个表中并对列分别进行插入排序，重复这过程，不过每次用更长的列（步长更长了，列数更少了）来进行。最后整个表就只有一列了。将数组转换至表是为了更好地理解这算法，算法本身还是使用数组进行排序。

例如，假设有这样一组数[ 13 14 94 33 82 25 59 94 65 23 45 27 73 25 39 10 ]，如果我们以步长为5开始进行排序，我们可以通过将这列表放在有5列的表中来更好地描述算法，这样他们就应该看起来是这样(竖着的元素是步长组成)：

<span style="color:#000000"><code>13 14 94 33 82
25 59 94 65 23
45 27 73 25 39
10
</code></span>

然后我们对每列进行排序：

<span style="color:#000000"><code>10 14 73 25 23
13 27 94 33 39
25 59 94 65 82
45
</code></span>

将上述四行数字，依序接在一起时我们得到：[ 10 14 73 25 23 13 27 94 33 39 25 59 94 65 82 45 ]。这时10已经移至正确位置了，然后再以3为步长进行排序：

<span style="color:#000000"><code>10 14 73
25 23 13
27 94 33
39 25 59
94 65 82
45
</code></span>

排序之后变为：

<span style="color:#000000"><code>10 14 13
25 23 33
27 25 59
39 65 73
45 94 82
94
</code></span>

最后以1步长进行排序（此时就是简单的插入排序了）

希尔排序的分析

<span style="color:#000000"><code class="language-py"><span style="color:#a71d5d">def</span> <span style="color:#795da3">shell_sort</span><span style="color:#969896">(</span>alist<span style="color:#969896">)</span><span style="color:#969896">:</span>
    n <span style="color:#df5000">=</span> <span style="color:#df5000">len</span><span style="color:#969896">(</span>alist<span style="color:#969896">)</span>
    <span style="color:#969896"># 初始步长</span>
    gap <span style="color:#df5000">=</span> n <span style="color:#df5000">/</span> <span style="color:#c76b29">2</span>
    <span style="color:#a71d5d">while</span> gap <span style="color:#df5000">></span> <span style="color:#c76b29">0</span><span style="color:#969896">:</span>
        <span style="color:#969896"># 按步长进行插入排序</span>
        <span style="color:#a71d5d">for</span> i <span style="color:#a71d5d">in</span> <span style="color:#df5000">range</span><span style="color:#969896">(</span>gap<span style="color:#969896">,</span> n<span style="color:#969896">)</span><span style="color:#969896">:</span>
            j <span style="color:#df5000">=</span> i
            <span style="color:#969896"># 插入排序</span>
            <span style="color:#a71d5d">while</span> j<span style="color:#df5000">>=</span>gap <span style="color:#df5000">and</span> alist<span style="color:#969896">[</span>j<span style="color:#df5000">-</span>gap<span style="color:#969896">]</span> <span style="color:#df5000">></span> alist<span style="color:#969896">[</span>j<span style="color:#969896">]</span><span style="color:#969896">:</span>
                alist<span style="color:#969896">[</span>j<span style="color:#df5000">-</span>gap<span style="color:#969896">]</span><span style="color:#969896">,</span> alist<span style="color:#969896">[</span>j<span style="color:#969896">]</span> <span style="color:#df5000">=</span> alist<span style="color:#969896">[</span>j<span style="color:#969896">]</span><span style="color:#969896">,</span> alist<span style="color:#969896">[</span>j<span style="color:#df5000">-</span>gap<span style="color:#969896">]</span>
                j <span style="color:#df5000">-=</span> gap
        <span style="color:#969896"># 得到新的步长</span>
        gap <span style="color:#df5000">=</span> gap <span style="color:#df5000">/</span> <span style="color:#c76b29">2</span>

alist <span style="color:#df5000">=</span> <span style="color:#969896">[</span><span style="color:#c76b29">54</span><span style="color:#969896">,</span><span style="color:#c76b29">26</span><span style="color:#969896">,</span><span style="color:#c76b29">93</span><span style="color:#969896">,</span><span style="color:#c76b29">17</span><span style="color:#969896">,</span><span style="color:#c76b29">77</span><span style="color:#969896">,</span><span style="color:#c76b29">31</span><span style="color:#969896">,</span><span style="color:#c76b29">44</span><span style="color:#969896">,</span><span style="color:#c76b29">55</span><span style="color:#969896">,</span><span style="color:#c76b29">20</span><span style="color:#969896">]</span>
shell_sort<span style="color:#969896">(</span>alist<span style="color:#969896">)</span>
<span style="color:#a71d5d">print</span><span style="color:#969896">(</span>alist<span style="color:#969896">)</span>
</code></span>

时间复杂度

• 最优时间复杂度：根据步长序列的不同而不同
• 最坏时间复杂度：O(n2)
• 稳定想：不稳定

希尔排序演示