aima.probability.decision
Class MDP<STATE_TYPE,ACTION_TYPE>

java.lang.Object
  aima.probability.decision.MDP<STATE_TYPE,ACTION_TYPE>

public class MDP<STATE_TYPE,ACTION_TYPE>
extends java.lang.Object

Constructor Summary

MDP(MDPSource<STATE_TYPE,ACTION_TYPE> source)

Method Summary

MDP<STATE_TYPE,ACTION_TYPE>	emptyMdp()
MDPPerception<STATE_TYPE>	execute(STATE_TYPE state, ACTION_TYPE action, Randomizer r)
java.util.List<ACTION_TYPE>	getAllActions()
STATE_TYPE	getInitialState()
double	getRewardFor(STATE_TYPE state)
double	getTransitionProbability(MDPTransition<STATE_TYPE,ACTION_TYPE> transition)
java.util.List<MDPTransition<STATE_TYPE,ACTION_TYPE>>	getTransitionsWith(STATE_TYPE initialState, ACTION_TYPE action)
MDPUtilityFunction<STATE_TYPE>	initialUtilityFunction()
boolean	isTerminalState(STATE_TYPE state)
MDPUtilityFunction<STATE_TYPE>	policyEvaluation(MDPPolicy<STATE_TYPE,ACTION_TYPE> pi, MDPUtilityFunction<STATE_TYPE> U, double gamma, int iterations)
MDPPolicy<STATE_TYPE,ACTION_TYPE>	policyIteration(double gamma)
MDPPolicy<STATE_TYPE,ACTION_TYPE>	randomPolicy()
void	setReward(STATE_TYPE state, double reward)
void	setTransitionProbability(MDPTransition<STATE_TYPE,ACTION_TYPE> transition, double probability)
java.lang.String	toString()
Pair<MDPUtilityFunction<STATE_TYPE>,java.lang.Double>	valueIterateOnce(double gamma, MDPUtilityFunction<STATE_TYPE> presentUtilityFunction)
MDPUtilityFunction<STATE_TYPE>	valueIteration(double gamma, double error, double delta)
MDPUtilityFunction<STATE_TYPE>	valueIterationForFixedIterations(int numberOfIterations, double gamma)
MDPUtilityFunction<STATE_TYPE>	valueIterationTillMAximumUtilityGrowthFallsBelowErrorMargin(double gamma, double errorMargin)

Methods inherited from class java.lang.Object

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

Constructor Detail

MDP

public MDP(MDPSource<STATE_TYPE,ACTION_TYPE> source)

Method Detail

emptyMdp

public MDP<STATE_TYPE,ACTION_TYPE> emptyMdp()

valueIteration

public MDPUtilityFunction<STATE_TYPE> valueIteration(double gamma,
                                                     double error,
                                                     double delta)

valueIterationForFixedIterations

public MDPUtilityFunction<STATE_TYPE> valueIterationForFixedIterations(int numberOfIterations,
                                                                       double gamma)

valueIterationTillMAximumUtilityGrowthFallsBelowErrorMargin

public MDPUtilityFunction<STATE_TYPE> valueIterationTillMAximumUtilityGrowthFallsBelowErrorMargin(double gamma,
                                                                                                  double errorMargin)

valueIterateOnce

public Pair<MDPUtilityFunction<STATE_TYPE>,java.lang.Double> valueIterateOnce(double gamma,
                                                                              MDPUtilityFunction<STATE_TYPE> presentUtilityFunction)

policyIteration

public MDPPolicy<STATE_TYPE,ACTION_TYPE> policyIteration(double gamma)

policyEvaluation

public MDPUtilityFunction<STATE_TYPE> policyEvaluation(MDPPolicy<STATE_TYPE,ACTION_TYPE> pi,
                                                       MDPUtilityFunction<STATE_TYPE> U,
                                                       double gamma,
                                                       int iterations)

randomPolicy

public MDPPolicy<STATE_TYPE,ACTION_TYPE> randomPolicy()

initialUtilityFunction

public MDPUtilityFunction<STATE_TYPE> initialUtilityFunction()

getInitialState

public STATE_TYPE getInitialState()

getRewardFor

public double getRewardFor(STATE_TYPE state)

setReward

public void setReward(STATE_TYPE state,
                      double reward)

setTransitionProbability

public void setTransitionProbability(MDPTransition<STATE_TYPE,ACTION_TYPE> transition,
                                     double probability)

getTransitionProbability

public double getTransitionProbability(MDPTransition<STATE_TYPE,ACTION_TYPE> transition)

execute

public MDPPerception<STATE_TYPE> execute(STATE_TYPE state,
                                         ACTION_TYPE action,
                                         Randomizer r)

isTerminalState

public boolean isTerminalState(STATE_TYPE state)

getTransitionsWith

public java.util.List<MDPTransition<STATE_TYPE,ACTION_TYPE>> getTransitionsWith(STATE_TYPE initialState,
                                                                                ACTION_TYPE action)

getAllActions

public java.util.List<ACTION_TYPE> getAllActions()

toString

public java.lang.String toString()

Overrides:

toString in class java.lang.Object