目录
1. 归并排序动画
配合如下所示的每阶段排序后结果,更便于理解:
排序前 = [14, 2, 25, 17, 7, 38, 34, 5, 42, 15, 42]
nums = [2, 14, 25, 17, 7, 38, 34, 5, 42, 15, 42]
nums = [2, 14, 25, 17, 7, 38, 34, 5, 42, 15, 42]
nums = [2, 14, 25, 7, 17, 38, 34, 5, 42, 15, 42]
nums = [2, 14, 25, 7, 17, 38, 34, 5, 42, 15, 42]
nums = [2, 7, 14, 17, 25, 38, 34, 5, 42, 15, 42]
nums = [2, 7, 14, 17, 25, 38, 5, 34, 42, 15, 42]
nums = [2, 7, 14, 17, 25, 38, 5, 34, 42, 15, 42]
nums = [2, 7, 14, 17, 25, 38, 5, 34, 42, 15, 42]
nums = [2, 7, 14, 17, 25, 38, 5, 15, 34, 42, 42]
nums = [2, 5, 7, 14, 15, 17, 25, 34, 38, 42, 42]
排序后 = [2, 5, 7, 14, 15, 17, 25, 34, 38, 42, 42]
2. 归并排序 Java 实现
public static void main(String[] args) {
int[] numbers = {14, 2, 25, 17, 7, 38, 34, 5, 42, 15, 42};
MergeSort mergeSort = new MergeSort();
System.out.println("排序前 = " + Arrays.toString(numbers));
mergeSort.sort(numbers, 0, numbers.length - 1);
System.out.println("排序后 = " + Arrays.toString(numbers));
}
/**
* 对 [left,right] 范围进行排序
* @param nums 数组
* @param left 对 [left,right] 范围进行排序
* @param right 对 [left,right] 范围进行排序
*/
public void sort(int[] nums, int left, int right) {
if (left >= right) {
return;
}
// 找到数组中间值
int mid = left + (right - left) / 2;
// 对 [left,mid] 和 [mid+1,right] 分别排序
sort(nums, left, mid);
sort(nums, mid + 1, right);
// 对以排序的 [left,mid] 和 [mid+1,right] 进行合并
merge(nums, left, mid + 1, right);
System.out.println("nums = " + Arrays.toString(nums));
}
/**
* 对两段 [left,right-1][right,length] 排序过的数组范围进行合并
* @param nums 数组
* @param left 左起点 [left,right-1]
@param right 右起点 [right,length]
* @param length 排序端长度
*/
public void merge(int[] nums, int left, int right, int length) {
int mid = right - 1;
int i = left;
int j = right;
int index = 0;
int[] tmp = new int[length - left + 1];
// 对 [i,mid] [j,length] 两段区域进行合并
while (i <= mid && j <= length) {
tmp[index++] = nums[i] <= nums[j] ? nums[i++] : nums[j++];
}
// [i,mid] 有剩余则加入数组尾部
while (i <= mid) {
tmp[index++] = nums[i++];
}
// [j,length] 有剩余则加入数组尾部
while (j <= length) {
tmp[index++] = nums[j++];
}
// 排序过的区域赋值给原数组
for (int x = 0; x < tmp.length; x++) {
nums[left + x] = tmp[x];
}
}
3. 总结
时间复杂度
merge 函数时间复杂度 O(n)
复杂度
F(n) = 2 * F(n/2) + n
= 4 * F(n/4) + 2 * n
= 8 * F(n/8) + 3 * n
= 2^k * F(n/2^k) + k * n
当 F(n/2^k) = F(1) 时,也就是 n/2^k=1,我们得到 k=logn 。
F(n) = n * (常数) + nlogn = n * (常数 + logn) ≈ n * logn
所以归并排序时间复杂度为 O(nlogn)
空间复杂度
归并排序的空间复杂度为 O(n)
排序稳定性
归并排序的稳定性取决于左右两段数组合并方式是否稳定,主要在于如下所示代码的判断
tmp[index++] = nums[i] <= nums[j] ? nums[i++] : nums[j++];
i 为左半部分起点,j 为右半部分起点
当 nums[i] = nums[j] 时,使用的为左半部分的值,所以排序为稳定的排序
如将上述 nums[i] <= nums[j] 换成 nums[i] < nums[j],则排序为不稳定排序