POJ 1487:Single-Player Games 浮点数高斯消元

探讨单人游戏树状结构下随机策略的期望得分计算方法。通过定义游戏树的结构,采用递归方式解析游戏状态,并运用高斯消元法解决线性方程组,最终得出随机游玩时的预期分数。

Single-Player Games
Time Limit: 1000MS Memory Limit: 10000K
Total Submissions: 1287 Accepted: 362

Description

Playing games is the most fun if other people take part. But other players are not always available if you need them, which led to the invention of single-player games. One of the most well-known examples is the infamous ``Solitaire'' packaged with Windows, probably responsible for more wasted hours in offices around the world than any other game. 
The goal of a single-player game is usually to make ``moves'' until one reaches a final state of the game, which results in a win or loss, or a score assigned to that final state. Most players try to optimize the result of the game by employing good strategies. In this problem we are interested in what happens if one plays randomly. After all, these games are mostly used to waste time, and playing randomly achieves this goal as well as any other strategy. 

Games can very compactly represented as (possibly infinite) trees. Every node of the tree repre- sents a possible game state. The root of the tree corresponds to the starting position of the game. For an inner node, its children are the game states to which one can move in a single move. The leaf nodes are the final states, and every one of them is assigned a number, which is the score one receives when ending up at that leaf. 

Trees are defined using the following grammar. 

Definition ::= Identifier "=" RealTree 
RealTree ::= "("Tree+")" 
Tree ::= Identifier | Integer | "("Tree+")" 
Identifier ::= a|b|...|z 
Integer ∈ {...,-3,-2,-1,0,1,2,3,...,} 

By using a Definition, the RealTree on the right-hand side of the equation is assigned to the Identifier on the left. A RealTree consists of a root node and one or more children, given as a sequence enclosed in brackets. And a Tree is either 

. the tree represented by a given Identifier, or 

. a leaf node, represented by a single Integer, or 

. an inner node, represented by a sequence of one or more Trees (its children), enclosed in brackets. 

Your goal is to compute the expected score, if one plays randomly, i.e. at each inner node selects one of the children uniformly at random. This expected score is well-defined even for the infinite trees definable in our framework as long as the probability that the game ends (playing randomly) is 1. 

Input

The input file contains several gametree descriptions. Each description starts with a line containing the number n of identifiers used in the description. The identifiers used will be the first n lowercase letters of the alphabet. The following n lines contain the definitions of these identifiers (in the order a, b, ...). Each definition may contain arbitrary whitespace (but of course there will be no spaces within a single integer). The right hand side of a definition will contain only identifiers from the first n lowercase letters. The inputs ends with a test case starting with n = 0. This test case should not be processed.

Output

For each gametree description in the input, first output the number of the game. Then, for all n identifiers in the order a, b, ..., output the following. If an identifier represents a gametree for which the probability of finishing the game is 1, print the expected score (when playing randomly). This value should be exact to three digits to the right of the decimal point. 

If the game described by the variable does not end with probability 1, print ``Expected score of id undefined'' instead. Output a blank line after each test case.

Sample Input

1
a = ((1 7) 6 ((8 3) 4))
2
a = (1 b)
b = (4 a)
1
a = (a a a)
0

Sample Output

Game 1
Expected score for a = 4.917

Game 2
Expected score for a = 2.000
Expected score for b = 3.000

Game 3
Expected score for a undefined

题意是给出一个树,然后树上的根节点有相应的值,所有根节点 对应的值*树到达该根节点的概率 相加就是1。用深搜的方法列方程,每一个字符列一个方程,最后用高斯消元解方程。

这题出得很严谨,注意精度。

今天出了各种问题,这道题调了差不多4个小时。。。

#pragma warning(disable:4996)  
#include <iostream>  
#include <algorithm>  
#include <cmath>  
#include <vector>  
#include <stdlib.h>
#include <string.h>  
#include <string>
#include <cstring>
using namespace std;

#define eps 1e-6

int n, p, len;
char s[350];
int free_x[500];
double x[500];
double val[500][500];

inline double zero(double x)
{
	if (fabs(x) > eps)return x;
	return 0;
}
int Gauss(int equ, int var)
{
	int i, j, k;
	int max_r;//当前这列绝对值最大的行
	int col;//当前处理的列
	double tb;
	double temp;

	for (int i = 0; i <= var; i++)
	{
		x[i] = 0;
		free_x[i] = 1;
	}
	//转换为阶梯阵
	col = 0;//当前处理的列
	for (k = 0; k < equ&&col < var; k++, col++)
	{
		//枚举当前处理的行
		//找到该col列元素绝对值最大的那行与第k行交换.(为了在除法时减少误差)
		max_r = k;
		for (i = k + 1; i < equ; i++)
		{
			if (fabs(val[i][col])>fabs(val[max_r][col]))
				max_r = i;
		}
		if (max_r != k)
		{//与第k行交换
			for (j = k; j < var + 1; j++)
				swap(val[k][j], val[max_r][j]);
		}
		if (fabs(val[k][col]) < eps)
		{
			k--;
			continue;
		}
		for (i = k + 1; i < equ; i++)
		{//枚举要删去的行
			if (fabs(val[i][col]) > eps)
			{
				tb = val[i][col] / val[k][col];
				for (j = col; j < var + 1; j++)
				{
					val[i][j] = (val[i][j] - val[k][j] * tb);
				}
			}
		}
	}

	//无解的情况
	for (i = k; i < equ; i++)
	{
		if (val[i][col] > eps)
			return -1;
	}

	if (k < equ)
	{
		//注释处为求多解的自由变量  
		// 首先,自由变元有n - k个,即不确定的变元至少有n - k个.  
		int num = 0, freeidx;
		for (i = k - 1; i >= 0; --i)
		{
			num = 0;// 用于判断该行中的不确定的变元的个数,如果超过1个,则无法求解,它们仍然为不确定的变元.  
			double tmp = val[i][n];
			// 第i行一定不会是(0, 0, ..., 0)的情况,因为这样的行是在第k行到第m行.  
			// 同样,第i行一定不会是(0, 0, ..., a), a != 0的情况,这样的无解的.  
			for (j = 0; j < n; ++j)
			{
				if (fabs(val[i][j])>eps && free_x[j])
				{
					num++;
					freeidx = j;
				}
			}
			if (num > 1) continue; // 无法求解出确定的变元.  
								   // 说明就只有一个不确定的变元free_index,那么可以求解出该变元,且该变元是确定的.  
			tmp = val[i][n];
			for (j = 0; j < n; ++j)
			{
				if (fabs(val[i][j])>eps && j != freeidx) tmp -= val[i][j] * x[j];
			}
			x[freeidx] = tmp / val[i][freeidx];
			free_x[freeidx] = 0;
		}
		return n - k;
	}

	for (i = var - 1; i >= 0; i--)
	{
		temp = val[i][var];
		for (j = i + 1; j < var; j++)
		{
			if (val[i][j])
			{
				temp = temp - val[i][j] * x[j];
			}
		}
		x[i] = temp / val[i][i];
	}
	return 0;
}

void dfs(double *a)
{
	int cnt = 0;
	while (s[p] != ')')
	{
		if (s[p] == ' ')
		{
			p++;
		}
		else if (s[p] == '(')
		{
			double b[50];
			p++;

			memset(b, 0, sizeof(b));
			dfs(b);
			cnt++;
			for (int k = 0; k <= n; k++)
			{
				a[k] += b[k];
			}
		}
		else if (s[p] >= '0'&&s[p] <= '9')
		{
			cnt++;
			int num = 0;
			while (s[p] >= '0'&&s[p] <= '9')
			{
				num = num * 10;
				num = num + s[p] - '0';
				p++;
			}
			a[n] += num;
		}
		else if (s[p] == '-')
		{
			cnt++;
			int num = 0;
			p++;
			while (s[p] >= '0'&&s[p] <= '9')
			{
				num = num * 10;
				num = num + s[p] - '0';
				p++;
			}
			a[n] -= num;
		}
		else if (s[p] >= 'a'&&s[p] <= 'z')
		{
			cnt++;
			a[s[p] - 'a']++;
			p++;
		}
	}

	if (cnt)
	{
		for (int k = 0; k <= 27; k++)
		{
			a[k] = a[k] / cnt;
		}
	}
	p++;
	return;
}

int main()
{
	//freopen("i.txt", "r", stdin);
	//freopen("o.txt", "w", stdout);

	int i, k, ans;
	int itest = 0;
	while (scanf("%d", &n) != EOF)
	{
		if (n == 0)
			break;

		getchar();
		memset(val, 0, sizeof(val));

		printf("Game %d\n", ++itest);
		for (i = 0; i < n; i++)
		{
			gets_s(s);//VS2015用gets_s代替了get

			p = 0;
			while (s[p] != '(')
				p++;
			p++;
			dfs(val[i]);
			for (k = 0; k < n; k++)
				val[i][k] = -val[i][k];
			val[i][i]++;
		}
		ans = Gauss(n, n);
		for (k = 0; k < n; k++)
		{
			if (ans == -1)
			{
				printf("Expected score for %c undefined\n", k + 'a');
			}
			else if (ans == 0)
			{
				printf("Expected score for %c = %.3lf\n", k + 'a',x[k]);
			}
			else
			{
				if (free_x[k])
					printf("Expected score for %c undefined\n", k + 'a');
				else
					printf("Expected score for %c = %.3lf\n", k + 'a', x[k]);
			}

		}
		printf("\n");
	}
	//system("pause");
	return 0;
}


标题SpringBoot智能在线预约挂号系统研究AI更换标题第1章引言介绍智能在线预约挂号系统的研究背景、意义、国内外研究现状及论文创新点。1.1研究背景与意义阐述智能在线预约挂号系统对提升医疗服务效率的重要性。1.2国内外研究现状分析国内外智能在线预约挂号系统的研究与应用情况。1.3研究方法及创新点概述本文采用的技术路线、研究方法及主要创新点。第2章相关理论总结智能在线预约挂号系统相关理论,包括系统架构、开发技术等。2.1系统架构设计理论介绍系统架构设计的基本原则和常用方法。2.2SpringBoot开发框架理论阐述SpringBoot框架的特点、优势及其在系统开发中的应用。2.3数据库设计与管理理论介绍数据库设计原则、数据模型及数据库管理系统。2.4网络安全与数据保护理论讨论网络安全威胁、数据保护技术及其在系统中的应用。第3章SpringBoot智能在线预约挂号系统设计详细介绍系统的设计方案,包括功能模块划分、数据库设计等。3.1系统功能模块设计划分系统功能模块,如用户管理、挂号管理、医生排班等。3.2数据库设计与实现设计数据库表结构,确定字段类型、主键及外键关系。3.3用户界面设计设计用户友好的界面,提升用户体验。3.4系统安全设计阐述系统安全策略,包括用户认证、数据加密等。第4章系统实现与测试介绍系统的实现过程,包括编码、测试及优化等。4.1系统编码实现采用SpringBoot框架进行系统编码实现。4.2系统测试方法介绍系统测试的方法、步骤及测试用例设计。4.3系统性能测试与分析对系统进行性能测试,分析测试结果并提出优化建议。4.4系统优化与改进根据测试结果对系统进行优化和改进,提升系统性能。第5章研究结果呈现系统实现后的效果,包括功能实现、性能提升等。5.1系统功能实现效果展示系统各功能模块的实现效果,如挂号成功界面等。5.2系统性能提升效果对比优化前后的系统性能
在金融行业中,对信用风险的判断是核心环节之一,其结果对机构的信贷政策和风险控制策略有直接影响。本文将围绕如何借助机器学习方法,尤其是Sklearn工具包,建立用于判断信用状况的预测系统。文中将涵盖逻辑回归、支持向量机等常见方法,并通过实际操作流程进行说明。 一、机器学习基本概念 机器学习属于人工智能的子领域,其基本理念是通过数据自动学习规律,而非依赖人工设定规则。在信贷分析中,该技术可用于挖掘历史数据中的潜在规律,进而对未来的信用表现进行预测。 二、Sklearn工具包概述 Sklearn(Scikit-learn)是Python语言中广泛使用的机器学习模块,提供多种数据处理和建模功能。它简化了数据清洗、特征提取、模型构建、验证与优化等流程,是数据科学项目中的常用工具。 三、逻辑回归模型 逻辑回归是一种常用于分类任务的线性模型,特别适用于二类问题。在信用评估中,该模型可用于判断借款人是否可能违约。其通过逻辑函数将输出映射为0到1之间的概率值,从而表示违约的可能性。 四、支持向量机模型 支持向量机是一种用于监督学习的算法,适用于数据维度高、样本量小的情况。在信用分析中,该方法能够通过寻找最佳分割面,区分违约与非违约客户。通过选用不同核函数,可应对复杂的非线性关系,提升预测精度。 五、数据预处理步骤 在建模前,需对原始数据进行清理与转换,包括处理缺失值、识别异常点、标准化数值、筛选有效特征等。对于信用评分,常见的输入变量包括收入水平、负债比例、信用历史记录、职业稳定性等。预处理有助于减少噪声干扰,增强模型的适应性。 六、模型构建与验证 借助Sklearn,可以将数据集划分为训练集和测试集,并通过交叉验证调整参数以提升模型性能。常用评估指标包括准确率、召回率、F1值以及AUC-ROC曲线。在处理不平衡数据时,更应关注模型的召回率与特异性。 七、集成学习方法 为提升模型预测能力,可采用集成策略,如结合多个模型的预测结果。这有助于降低单一模型的偏差与方差,增强整体预测的稳定性与准确性。 综上,基于机器学习的信用评估系统可通过Sklearn中的多种算法,结合合理的数据处理与模型优化,实现对借款人信用状况的精准判断。在实际应用中,需持续调整模型以适应市场变化,保障预测结果的长期有效性。 资源来源于网络分享,仅用于学习交流使用,请勿用于商业,如有侵权请联系我删除!
评论
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值