orbital.algorithm.template
Class MarkovDecisionProblem.DefaultTransition

java.lang.Object
  orbital.algorithm.template.MarkovDecisionProblem.DefaultTransition

All Implemented Interfaces:

java.io.Serializable, java.lang.Comparable, MarkovDecisionProblem.Transition, TransitionModel.Transition

Enclosing interface:

MarkovDecisionProblem

public static class MarkovDecisionProblem.DefaultTransition
extends java.lang.Object
implements MarkovDecisionProblem.Transition, java.io.Serializable

Default implementation of transition options for Markov decision processes.

Author:

André Platzer

See Also:

Serialized Form

Stereotype:

Structure, Implementation

Constructor Summary

MarkovDecisionProblem.DefaultTransition(double probability, double cost)
Deprecated. convenience constructor, prefer to use ValueFactory.valueOf(double)..

MarkovDecisionProblem.DefaultTransition(Scalar probability, Real cost)
Create a new option <p,c>.

Method Summary

int	compareTo(java.lang.Object o) Compares transition options.
Real	getCost() Get the cost of taking the action leading to this transition.
Scalar	getProbability() Get the transition probability.
java.lang.String	toString()

Methods inherited from class java.lang.Object

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

Methods inherited from interface orbital.algorithm.template.TransitionModel.Transition

equals, hashCode

Constructor Detail

MarkovDecisionProblem.DefaultTransition

public MarkovDecisionProblem.DefaultTransition(Scalar probability,
                                               Real cost)

Create a new option <p,c>.

Parameters:

probability - the probability of reaching a state s'.

cost - the immediate cost of taking the action which took us to that state s'.

MarkovDecisionProblem.DefaultTransition

public MarkovDecisionProblem.DefaultTransition(double probability,
                                               double cost)

Deprecated. convenience constructor, prefer to use ValueFactory.valueOf(double)..

Method Detail

compareTo

public int compareTo(java.lang.Object o)

Description copied from interface: TransitionModel.Transition

Compares transition options. In nondeterministic cases, implementations will usually compare transition options according to their probabilities. Deterministic cases, however, may prefer comparisons involving cost or accumulated cost. Those comparisons can also be combined. In any case, implementations are not required to use any specific order.

Specified by:

compareTo in interface java.lang.Comparable

Specified by:

compareTo in interface TransitionModel.Transition

toString

public java.lang.String toString()

Overrides:

toString in class java.lang.Object

getProbability

public Scalar getProbability()

Description copied from interface: TransitionModel.Transition

Get the transition probability.

Specified by:

getProbability in interface TransitionModel.Transition

Returns:

the transition probability p in [0,1] of taking this transition.

getCost

public Real getCost()

Description copied from interface: MarkovDecisionProblem.Transition

Get the cost of taking the action leading to this transition.

Note: Since c:S×A->R is a function, the cost should only depend on the action a taken, and the state s in which it was taken, not the actual outcome (which is unknown a priori). Otherwise, subset construction would not be applicable, etc.

Specified by:

getCost in interface MarkovDecisionProblem.Transition

Returns:

c(s,a) the cost of taking the action a in A(s) that took us here from state s in S.