Class PigSolver

java.lang.Object

PigSolver

public class PigSolver
extends java.lang.Object

PigSolver.java - Solver for Pig with JUnit testing interface. See http://cs.gettysburg.edu/~tneller/nsf/pig/ for details.

Constructor Summary

Constructors

Constructor and Description
PigSolver(int goal, double epsilon) Solve Pig for a given goal score using value iteration with a given convergence parameter epsilon.

Method Summary

Modifier and Type	Method and Description
double	pWin(int i, int j, int k) Return the probability of winning with optimal play given the current score, opponent score, and current turn total.
boolean	shouldRoll(int i, int j, int k) Return whether or not an optimal player should roll given the current player score, opponent score, and current turn total.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

PigSolver

public PigSolver(int goal,
                 double epsilon)

Solve Pig for a given goal score using value iteration with a given convergence parameter epsilon. A solution maximizes the probability of winning.

Parameters:

goal - player goal score to win

epsilon - minimum magnitude of pWin change in one value iteration for continuation

Method Detail

pWin

public double pWin(int i,
                   int j,
                   int k)

Return the probability of winning with optimal play given the current score, opponent score, and current turn total. If i + k ≥ goal, the player has won. If j ≥ goal, the player has lost. Otherwise, return the non-terminal state's probability for winning with optimal play.

Parameters:

i - current player score

j - opponent score

k - current turn total

Returns:

probability of winning in the current state with optimal play

shouldRoll

public boolean shouldRoll(int i,
                          int j,
                          int k)

Return whether or not an optimal player should roll given the current player score, opponent score, and current turn total. Assume j < goal and i + k < goal.

Parameters:

i - current player score

j - opponent score

k - current turn total

Returns:

whether or not an optimal player should roll in the current state