zoukankan      html  css  js  c++  java
  • HDU 3076:ssworld VS DDD(概率DP)

    http://acm.split.hdu.edu.cn/showproblem.php?pid=3076

    ssworld VS DDD



    Problem Description
     
    One day, sssworld and DDD play games together, but there are some special rules in this games.
    They both have their own HP. Each round they dice respectively and get the points P1 and P2 (1 <= P1, P2 <= 6). Small number who, whose HP to reduce 1, the same points will remain unchanged. If one of them becomes 0 HP, he loses. 
    As a result of technical differences between the two, each person has different probability of throwing 1, 2, 3, 4, 5, 6. So we couldn’t predict who the final winner. 

     
    Input
    There are multiple test cases.
    For each case, the first line are two integer HP1, HP2 (1 <= HP1, HP2 <= 2000), said the first player sssworld’s HP and the second player DDD’s HP. 
    The next two lines each have six floating-point numbers per line. The jth number on the ith line means the the probability of the ith player gets point j. The input data ensures that the game always has an end. 
     
    Output
     
    One float with six digits after point, indicate the probability sssworld won the game.
     
    Sample Input
     
    5 5
    1.000 0.000 0.000 0.000 0.000 0.000
    0.000 0.000 0.000 0.000 0.000 1.000
    5 5
    0.000 0.000 0.000 0.000 0.000 1.000
    1.000 0.000 0.000 0.000 0.000 0.000
     
    Sample Output
     
    0.000000
    1.000000

    题意:两个人玩游戏,给出各自掷骰子得到不同点数的概率,点数小的扣一血,相同不扣血。

    思路:概率DP,处理到第二个人血量剩余1的时候,然后再处理他降到0的时候的情况,不能直接循环第二个人血量为0的情况,即ans += dp[i][0] (1 <= i <= h1),因为这样的话dp[i][0]继承了dp[i+1][0]的状态,但是为0的时候游戏已经停止了,这样是不合理的。应该是ans += dp[i][1] * win / (1 - ave),这样的话才不会错,要注意考虑(1-ave)的情况,因为当前回合如果平局的话,下一个回合还是有机会赢的。(PS:很感谢ZLW师兄帮我debug了这题好久233)

     1 #include <cstdio>
     2 #include <cstring>
     3 #include <algorithm>
     4 #include <iostream>
     5 #include <cmath>
     6 using namespace std;
     7 const double eps = 0.00000001;
     8 double d[2][6];
     9 double dp[2005][2005];
    10 
    11 int main()
    12 {
    13     int h1, h2;
    14     while(~scanf("%d%d", &h2, &h1)) { //题目输入反了
    15         double win = 0, lose = 0, ave = 0;
    16         for(int i = 0; i < 2; i++)
    17             for(int j = 0; j < 6; j++)
    18                 scanf("%lf", &d[i][j]);
    19         for(int i = 0; i < 6; i++) {
    20             for(int j = 0; j < 6; j++) {
    21                 if(i == j) ave += d[0][i] * d[1][j];
    22                 else if(i < j) lose += d[0][i] * d[1][j];
    23                 else win += d[0][i] * d[1][j];
    24             }
    25         }
    26 //        printf("SIZE : %d
    ", sizeof(dp));
    27         double ans = 0;
    28         if(eps >= fabs(1 - ave)) {
    29             printf("%.6f
    ", ans);
    30             continue;
    31         }
    32         for(int i = 0; i <= h1+1; i++)
    33             for(int j = 0; j <= h2+1; j++)
    34                 dp[i][j] = 0;
    35         // dp[h1+1][h2-1] = dp[h1-1][h2+1] = dp[h1+1][h2] = dp[h1][h2+1] = 0;
    36         dp[h1][h2] = 1;
    37         for(int i = h1; i >= 0; i--) {
    38             for(int j = h2; j >= 0; j--) {
    39                 if(i == h1 && j == h2) continue;
    40                 dp[i][j] = (dp[i+1][j] * lose + dp[i][j+1] * win) / ((double)1 - ave);
    41             }
    42         }
    43         for(int i = 1; i <= h1; i++)
    44             ans += dp[i][1] * win / ((double)1 - ave);
    45         printf("%.6f
    ", ans);
    46     }
    47     return 0;
    48 }
  • 相关阅读:
    Leetcode-Daily: Jewels and Stones
    Spark专题(二):Hadoop Shuffle VS Spark Shuffle
    Spark专题(一):Spark工作原理图
    我司建立图书借阅的一点思考
    机器学习的分类方法——逻辑回归
    机器学习的分类算法——集成学习
    浅谈对离散型随机变量期望的理解
    sklearn学习:make_multilabel_classification——多标签数据集方法
    外星人在阴间修仙
    RabbitMQ
  • 原文地址:https://www.cnblogs.com/fightfordream/p/5792163.html
Copyright © 2011-2022 走看看