aima.core.probability.mdp.impl

Class MDP<S,A extends Action>

java.lang.Object

aima.core.probability.mdp.impl.MDP<S,A>

Type Parameters:

S - the state type.

A - the action type.

All Implemented Interfaces:

MarkovDecisionProcess<S,A>

public class MDP<S,A extends Action>
extends java.lang.Object
implements MarkovDecisionProcess<S,A>

Default implementation of the MarkovDecisionProcess interface.

Author:

Ciaran O'Reilly, Ravi Mohan

Constructor Summary

Constructors

Constructor and Description
MDP(java.util.Set<S> states, S initialState, ActionsFunction<S,A> actionsFunction, TransitionProbabilityFunction<S,A> transitionProbabilityFunction, RewardFunction<S> rewardFunction)

Method Summary

Modifier and Type	Method and Description
java.util.Set<A>	actions(S s) Get the set of actions for state s.
S	getInitialState() Get the initial state s0 for this instance of a Markov decision process.
double	reward(S s) Get the reward associated with being in state s.
java.util.Set<S>	states() Get the set of states associated with the Markov decision process.
double	transitionProbability(S sDelta, S s, A a) Return the probability of going from state s using action a to s' based on the underlying transition model P(s' | s, a).

Methods inherited from class java.lang.Object

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

Constructor Detail

MDP

public MDP(java.util.Set<S> states,
           S initialState,
           ActionsFunction<S,A> actionsFunction,
           TransitionProbabilityFunction<S,A> transitionProbabilityFunction,
           RewardFunction<S> rewardFunction)

Method Detail

states

public java.util.Set<S> states()

Description copied from interface: MarkovDecisionProcess

Get the set of states associated with the Markov decision process.

Specified by:

states in interface MarkovDecisionProcess<S,A extends Action>

Returns:

the set of states associated with the Markov decision process.

getInitialState

public S getInitialState()

Description copied from interface: MarkovDecisionProcess

Get the initial state s₀ for this instance of a Markov decision process.

Specified by:

getInitialState in interface MarkovDecisionProcess<S,A extends Action>

Returns:

the initial state s₀.

actions

public java.util.Set<A> actions(S s)

Description copied from interface: MarkovDecisionProcess

Get the set of actions for state s.

Specified by:

actions in interface MarkovDecisionProcess<S,A extends Action>

Parameters:

s - the state.

Returns:

the set of actions for state s.

transitionProbability

public double transitionProbability(S sDelta,
                                    S s,
                                    A a)

Description copied from interface: MarkovDecisionProcess

Return the probability of going from state s using action a to s' based on the underlying transition model P(s' | s, a).

Specified by:

transitionProbability in interface MarkovDecisionProcess<S,A extends Action>

Parameters:

sDelta - the state s' being transitioned to.

s - the state s being transitions from.

a - the action used to move from state s to s'.

Returns:

the probability of going from state s using action a to s'.

reward

public double reward(S s)

Description copied from interface: MarkovDecisionProcess

Get the reward associated with being in state s.

Specified by:

reward in interface MarkovDecisionProcess<S,A extends Action>

Parameters:

s - the state whose award is sought.

Returns:

the reward associated with being in state s.