力扣 528. 按权重随机选择 二分/堆

https://leetcode-cn.com/problems/random-pick-with-weight/

思路一：随机+二分。首先独立计算出每个数被选中的概率，然后计算该数组的前缀和。那么$a_i$被选中的概率就等于$su{m}_i-su{m}_{i-1}$，且$sum[0..n)=1$。因此每次$pick$可以先计算一个$[0,1]$内的随机数，然后看它落到哪个元素的区间范围内，利用二分可以加速这个过程。

class Solution {
public:
    Solution(vector<int>& w) {
        srand(time(NULL));
        int total=accumulate(w.begin(),w.end(),0);
        int n=w.size();
        sum.resize(n);
        sum[0]=w[0]*1.0/total;
        for(int i=1;i<n;i++)
            sum[i]=sum[i-1]+w[i]*1.0/total;
    }
    
    int pickIndex() {
        float rd=rand()*1.0/RAND_MAX;
        int l=0,r=sum.size()-1,mid;
        while(l<=r)
        {
            mid=(l+r)>>1;
            if(sum[mid]<rd)
                l=mid+1;
            else
                r=mid-1;
        }
        return l;
    }
private:
    vector<float> sum;
};

/**
 * Your Solution object will be instantiated and called as such:
 * Solution* obj = new Solution(w);
 * int param_1 = obj->pickIndex();
 */

思路二：给每个位置分配一个优先级，然后利用堆来调度。

class node{
public:
    float val,step;
    int idx;
    node(float v,float s,int i):val(v),step(s),idx(i){}
    bool operator <(const node& a)const
    {
        return val>a.val;
    }
};

class Solution {
public:

    Solution(vector<int>& w) {
        srand(time(NULL));
        int total=accumulate(w.begin(),w.end(),0);
        int n=w.size(),sum=0;
        for(int i=0;i<n;i++)
        {
            q.emplace(total*1.0/w[i],total*1.0/w[i],i);
        }
    }
    
    int pickIndex() {
        node t=q.top();
        q.pop();
        q.emplace(t.val+t.step,t.step,t.idx);
        return t.idx;
    }
private:
    priority_queue<node> q;
};

/**
 * Your Solution object will be instantiated and called as such:
 * Solution* obj = new Solution(w);
 * int param_1 = obj->pickIndex();
 */