随机算法

csuzhucong

已于 2024-01-16 19:37:54 修改

阅读量3.8k

点赞数 2

文章标签：算法

于 2021-03-27 18:17:37 首次发布

本文链接：https://blog.csdn.net/nameofcsdn/article/details/115269170

版权

一，蒙特卡洛算法

1，圆周率估算

2，弗里瓦德算法(Frievald's Algorithm)

51Nod 1140 矩阵相乘结果的判断

3，数组中出现次数超过一半的数字

力扣剑指 Offer 39. 数组中出现次数超过一半的数字

一，蒙特卡洛算法

蒙特卡洛算法是一类随机算法的总称，这些随机算法都是基于一个已知的分布规律进行计算的。

1，圆周率估算

取一个1*1的正方形，在其中随机取一个点，点在正方形内切圆中的概率就是PI/4

取的点越多，估算越精确。

2，弗里瓦德算法(Frievald's Algorithm)

弗里瓦德算法是专门用来解决下面这个问题的。

51Nod 1140 矩阵相乘结果的判断

给出三个N*N的矩阵A, B, C，问A * B是否等于C？

Input

第1行，1个数N。(0 <= N <= 500) 第2 - N + 1行：每行N个数，对应矩阵A的元素。（0 <= Mii <= 16) 第N + 2 - 2N + 1行：每行N个数，对应矩阵B的元素。（0 <= Mii <= 16) 第2N + 2 - 3N + 1行：每行N个数，对应矩阵C的元素。

Output

如果相等输出Yes，否则输出No。

Sample Input

Sample Output

Yes

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

int p = 1000000007;

template<typename T>
vector<vector<T>> multi(vector<vector<T>>&v1,vector<vector<T>>&v2)
{
    vector<vector<T>>ans(v1.size(),vector<T>(v2[0].size(),0));
    for(int i=0;i<v1.size();i++)for(int j=0;j<v2[0].size();j++)for(int k=0;k<v2.size();k++)ans[i][j]+=v1[i][k]*v2[k][j],ans[i][j]%=p;
    return ans;
}

#define CIN(x) while (!(cin >> x)) { \
        cin.clear();      \
        cin.ignore();     \
        }

//输入一维vector
template<typename T>
void fcin(vector<T>&v,int len)
{
    v.resize(len);
    for(int i=0;i<v.size();i++)CIN(v[i]);
}
//输入二维vector
template<typename T>
void fcin2(vector<vector<T>>&v,int row,int col)
{
    v.resize(row);
    for(int i=0;i<v.size();i++)fcin(v[i],col);
}


int main()
{
    int n;
    cin>>n;
    vector<vector<long long>>A,B,C;
    fcin2(A,n,n);
    fcin2(B,n,n);
    fcin2(C,n,n);
    vector<vector<long long>>x(n);
    for(int i=0;i<n;i++){
        x[i].resize(1);
        x[i][0]=rand();
    }
    vector<vector<long long>>ans1=multi(B,x),ans2=multi(A,ans1),ans3=multi(C,x);
    if(VecIsSame(ans2,ans3))cout<<"Yes";
    else cout<<"No";
    return 0;
}

看着代码很多，但是只有main函数是需要独立写的，其他都可以从我的ACM模板中找到。

3，数组中出现次数超过一半的数字

力扣OJ 剑指 Offer（31-68）_nameofcsdn的博客-CSDN 博客

力扣剑指 Offer 39. 数组中出现次数超过一半的数字

数组中有一个数字出现的次数超过数组长度的一半，请找出这个数字。

你可以假设数组是非空的，并且给定的数组总是存在多数元素。

示例 1:

输入: [1, 2, 3, 2, 2, 2, 5, 4, 2]
输出: 2

限制：

1 <= 数组长度 <= 50000

思路：

不校验的随机算法

一层逻辑：随机取2个数，如果不同就继续取，直到2个数相同，此时这个数大概率就是正确答案。

二层逻辑：取2个数，如果不同就继续取，直到2个数相同，此时这个数大概率就是正确答案，但是，不是随机取，而是直接取一层逻辑得到的值。

同理，我们可以不停的这样嵌套，想要几层就有几层。

代码：

class Solution {
public:
    int err=123456789;
    int get0(vector<int>& nums){
        return nums[rand()%nums.size()];
    }
    int get1(vector<int>& nums){
        int a=0,b=1;
        while(a!=b){
            a=err,b=err;
            while(a==err)a=get0(nums);
            while(b==err)b=get0(nums);
        }
        return a;
    }
    int get2(vector<int>& nums){
        int a=0,b=1;
        while(a!=b){
            a=err,b=err;
            while(a==err)a=get1(nums);
            while(b==err)b=get1(nums);
        }
        return a;
    }
    int get3(vector<int>& nums){
        int a=0,b=1;
        while(a!=b){
            a=err,b=err;
            while(a==err)a=get2(nums);
            while(b==err)b=get2(nums);
        }
        return a;
    }
    int get4(vector<int>& nums){
        int a=0,b=1;
        while(a!=b){
            a=err,b=err;
            while(a==err)a=get3(nums);
            while(b==err)b=get3(nums);
        }
        return a;
    }
    int majorityElement(vector<int>& nums) {
        return get4(nums);
    }
};

然而最坏情况下算法是错的，失败用例：

[233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,233,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333,2333]

这是501个233和499个2333

改进思路：

带校验的随机算法

不需要嵌套，只要任何时候有怀疑的值，直接确定性检测即可

class Solution {
public:
    int err=123456789;
    int get0(vector<int>& nums){
        return nums[rand()%nums.size()];
    }
    int get(vector<int>& nums){
        int a=0,b=1;
        while(a!=b){
            a=err,b=err;
            while(a==err)a=get0(nums);
            while(b==err)b=get0(nums);
        }
        return a;
    }
    int majorityElement(vector<int>& nums) {
        while(true){
            int a=get(nums),s=0;
            for(int i=0;i<nums.size();i++)if(nums[i]==a)s++;
            if(s>nums.size()/2)return a;
        }
    }
};

4，其他蒙特卡洛方法

蒙特卡洛方法_nameofcsdn的博客-CSDN博客

二，蒙特卡洛算法的正确性

随机算法根据是否校验计算结果，可以分为两种。

1，带校验的随机算法

带校验的随机算法，可以确保计算结果正确，算法没有时间上限。

例如：

剑指 Offer 39. 数组中出现次数超过一半的数字其中的带校验的随机算法

2，不校验的随机算法

不校验的随机算法，无法保证计算结果正确，但相对来说不需要大量重复，有些算法有时间上限，有的没有。

例如：

圆周率估算，这个只能说误差不超过某个值的概率至少为多少，不能保证误差一定不超过它。

弗里瓦德算法，这个可以重复几次，增大正确的概率，但是还是无法保证正确性。

剑指 Offer 39. 数组中出现次数超过一半的数字其中的不校验的随机算法

三，快速排序

快速排序选取划分节点的这一步，可以随机选一个，也可以固定选待排序段的第一个元素，本质上是一样的，所以也可以说快速排序也是一种随机算法。
这种排序算法不属于蒙特卡洛算法，运行结果一定是正确的而且运行时间有上界。

四，数论中的随机算法

Pollard‘s rho大数分解算法

Pollard‘s rho大数分解算法_nameofcsdn的博客-CSDN博客

通过用递推数列模拟随机数列，用周期找到因子

csuzhucong

关注

2
点赞
踩
15

收藏

觉得还不错? 一键收藏
3
评论
随机算法

目录一，蒙特卡洛算法1，圆周率2，弗里瓦德算法(Frievald's Algorithm)51Nod - 1140 矩阵相乘结果的判断一，蒙特卡洛算法蒙特卡洛算法是一类随机算法的总称，这些随机算法都是基于一个已知的分布规律进行计算的。1，圆周率取一个1*1的正方形，在其中随机取一个点，点在正方形内切圆中的概率就是PI/4取的点越多，估算越精确。2，弗里瓦德算法(Frievald's Algorithm)弗里瓦德算法是专门用来解决下面这个问题的。51Nod -
复制链接

扫一扫