算法_二路归并排序

思路

二路归并的基本思想是将元素表看成n个长度为1的有序表，然后对其进行两两归并，再对归并得到的有序表进行两两归并，直到得到一个长度为n的有序表。

实现

二路归并的过程可以描述成一棵归并二叉树，实现方法有两种

//1.自顶向下(递归)
void merge_sort(RecType R[], int n)
{
    mergeRC(R, 0, n-1);
}

void mergeRC(RecType R[], int low, int high)
{
    int mid;
    if (low < high)
    {
        mid = (low + high) / 2;
        mergeRC(R, low, mid);
        mergeRC(R, mid+1, high);
        merge(R, low, mid, high);
    }
}

//一次归并
void merge(RecType R[], int low, int mid, int high)
{
    int i = low, j = mid+1, k=0;
    RecType R1 = (RecType *)malloc(sizeof(RecType) * (high - low + 1));
    while (i <= low && j<= high)
    {
        if (R[i].key <= R[j].key)
        {
            R1[k] = R[i];
            k++;
            i++;
        }
        else
        {
            R1[k] = R[j];
            k++;
            j++;
        }
    }
    while (i <= low)
    {
        R1[k] = R[i];
        k++;
        i++;
    }
    while (j <= high)
    {
        R1[k] = R[j];
        k++;
        j++;
    }
    for (i=low, k=0; i<=high; i++,k++)
        R[i] = R1[k];
    free(R1);
}

//2.自底向上
void merge_sort(RecType R[], int n)
{
    int i;
    for (i=1; i<n; i *= 2)
        merge_pass(R, n, i);
}

//一趟归并
void merge_pass(RecType R[], int n, int length)
{
    int i;
    for (i=0; i+2*length-1<n; i += 2*length)
        merge(R, i, i+length-1, i+2*length-1);
    if (i+length-1<n-1)
        merge(R, i, i+length-1; n-1);
}

//一次归并函数同法1

算法分析

时间复杂度

算法的执行过程与元素有序性无关，所以最好、最坏和平均时间复杂度相同，为
$$O(nlo{g}_{2}n)$$