Leetcode 4. 寻找两个正序数组的中位数

最新推荐文章于 2024-07-12 18:12:33 发布

dujiaosongmu

最新推荐文章于 2024-07-12 18:12:33 发布

阅读量73

点赞数

分类专栏：算法分析文章标签： leetcode 算法 c++

本文链接：https://blog.csdn.net/dujiaosongmu/article/details/129995431

版权

算法分析专栏收录该内容

4 篇文章 0 订阅

订阅专栏

Question

ref: https://leetcode.cn/problems/median-of-two-sorted-arrays/
给定两个大小分别为 m 和 n 的正序（从小到大）数组 nums1 和 nums2。请你找出并返回这两个正序数组的中位数。

算法的时间复杂度应该为 $O (l o g (m + n))$ 。

示例 1：

输入： $n u m s 1 = [1, 3], n u m s 2 = [2]$

输出： $2.00000$

解释：合并数组 $= [1, 2, 3]$ ，中位数 $2$

示例 $2$ ：

输入： $n u m s 1 = [1, 2], n u m s 2 = [3, 4]$

输出： $2.50000$

解释：合并数组 $= [1, 2, 3, 4]$ ，中位数 $(2 + 3) /2 = 2.5$

提示：

$\begin{align*} nums1.length == m \\ nums2.length == n \\ 0 <= m <= 1000 \\ 0 <= n <= 1000 \\ 1 <= m + n <= 2000 \\ -10^6 <= nums1[i], nums2[i] <= 10^6 \\ \end{align*}$

Answers1

The following code is my answer. but the space complexity is higher than other leet-coder and the rank of the time complexity is around of 25%. We need to further improve the answer quality and make it better.

#include<iostream>
#include<vector>
using namespace std;

double findMedianSortedArrays(vector<int>& nums1, vector<int>& nums2) {
    vector<int> all_num(0);
    int i1 = 0;
    int i2 = 0;
    int pos = 0;
    double result = 0;
    // merge two order list into single order list.
    // using nums1.size() to achieve the length of nums1.

    while (i1 < nums1.size() && i2 < nums2.size())
    {
        if (nums1[i1] <= nums2[i2])
        {
            // append a integer to the end of all_num array
            all_num.push_back(nums1[i1]);
            i1++;
        }
        else
        {
            all_num.push_back(nums2[i2]);
            i2++;
        }
    }

    while(i1 < nums1.size())
    {
        all_num.push_back(nums1[i1]);
        i1++;
    }

    while (i2 < nums2.size())
    {
        all_num.push_back(nums2[i2]);
        i2++;
    }
    
    if ((i1 + i2) % 2 == 0)
    {
        // warning: the result can rounded down(向下取整) with integer division
        pos = (i1 + i2) / 2;
        result = (all_num[pos] + all_num[pos - 1]) / 2.0;
    }
    else
    {
        pos = (i1 + i2) / 2;
        result = all_num[pos];
    }
    
    return result;
}

int main() 
{
    vector<int> nums1 = { 0,1,3 };
    vector<int> nums2 = { 2, 4};
    cout << findMedianSortedArrays(nums1, nums2) << endl;
    return 0;
}

Knowledge Corner

the int type is converted to float or double

int pos = 0;
float pos_f = pos;
\\or
cout << (float) pos << endl；

reference url

Cpp vector https://blog.csdn.net/qiancm/article/details/119611928

Answer 2

The following code don’t need to open more space to save the addition arrays, and just use pointer i and pointer j to solve this problem, the space complexity is converted from $O (m + n)$ to $O (1)$ .

class Solution {
public:
    double findMedianSortedArrays(vector<int>& nums1, vector<int>& nums2) {
        // 定义 i j 2个指针分别指针2个数组
        // 定义 l r 分别用来保存中位数；奇数个，只会用到 l 变量；偶数，会用到 l r 变量
        int m = nums1.size();
        int n = nums2.size();
        int i = 0, j = 0; 
        int l = 0, r = 0;
        for (int x = 0; x <= (m + n) / 2; x++) {
            l = r;
            r = (i < m && (j >= n || nums1[i] < nums2[j])) ?
                nums1[i++] : nums2[j++];
        }
        return (m + n) & 1 ? r : (l + r) / 2.0;
    }
};

Knowledge Corner

reference url:
bit operator: https://blog.csdn.net/weixin_43736974/article/details/84543970

Answer 3

We further improve the time complexity from $O (m + n)$ to $O (l o g (m + n))$ . The below is solution. The division method is used.

The main thought of the method is the problems is regarded as the question to find the $k$ -th biggest element in both arrays. And the pointer $i$ and pointer $j$ satisfy the following equation.

$i + j = k$
$nums1[i-1]<=nums2[j]\&\&nums2[j-1]<=nums1[i]$
In the end, the problem becomes $ma x (n u m s 1 [i - 1], n u m s 2 [j - 1])$ .

we just find the proper pointer i using division method, and the pointer j is acquired by equation $i + j = k$ .

class Solution {
public:
    int findKthElm(vector<int>& nums1,vector<int>& nums2,int k){
        assert(1<=k&&k<=nums1.size()+nums2.size());
        int le=max(0,int(k-nums2.size())),ri=min(k,int(nums1.size()));
        while(le<ri){
            int m=le+(ri-le)/2;
            if(nums2[k-m-1]>nums1[m]) le=m+1;//这为什么只写一个条件？！因为这是二分的排除条件，不是目标的符合条件，相当于排除条件，最后的结果才是符合条件的结果
            else ri=m;
        }//循环结束时的位置le即为所求位置，第k小即为max(nums1[le-1]),nums2[k-le-1])，但是由于le可以为0、k,所以
        //le-1或者k-le-1可能不存在所以下面单独判断下
        int nums1LeftMax=le==0?INT_MIN:nums1[le-1];
        int nums2LeftMax=le==k?INT_MIN:nums2[k-le-1];
        return max(nums1LeftMax,nums2LeftMax);
    }
    double findMedianSortedArrays(vector<int>& nums1, vector<int>& nums2) {
        int n=nums1.size()+nums2.size();
        if(n&1){//两个数组长度和为奇数
            return findKthElm(nums1,nums2,(n>>1)+1);
        }
        else{//为偶数
            return (findKthElm(nums1,nums2,n>>1)+findKthElm(nums1,nums2,(n>>1)+1))/2.0;
        }
    }
};