burlap.behavior.singleagent.planning.stochastic.dpoperator

Class SoftmaxOperator

java.lang.Object

burlap.behavior.singleagent.planning.stochastic.dpoperator.SoftmaxOperator

All Implemented Interfaces:

DPOperator

Direct Known Subclasses:

DifferentiableSoftmaxOperator

public class SoftmaxOperator
extends java.lang.Object
implements DPOperator

A softmax/Boltzmann operator. As its beta parameter goes to infinity, the operator goes to max. As it goes to zero, it takes an average. As it goes to -infinity, it goes to min. Beta is the inverse of the "temperature."

Author:

James MacGlashan.

Field Summary

Fields

Modifier and Type	Field and Description
protected double	beta
protected double	temp

Constructor Summary

Constructors

Constructor and Description
SoftmaxOperator() Initializes with beta = 1.0
SoftmaxOperator(double beta) Initializes.

Method Summary

Modifier and Type	Method and Description
double	apply(double[] qs) Applies the operator on the input q-values and returns the result.
double	getBeta()
void	setBeta(double beta)

Methods inherited from class java.lang.Object

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

Field Detail

beta

protected double beta

temp

protected double temp

Constructor Detail

SoftmaxOperator

public SoftmaxOperator()

Initializes with beta = 1.0

SoftmaxOperator

public SoftmaxOperator(double beta)

Initializes. As beta goes to infinity, the operator goes to max. As it goes to zero, it takes an average. As it goes to -infinity, it goes to min. Beta is the inverse of the "temperature."

Parameters:

beta - the softmax knob.

Method Detail

getBeta

public double getBeta()

setBeta

public void setBeta(double beta)

apply

public double apply(double[] qs)

Description copied from interface: DPOperator

Applies the operator on the input q-values and returns the result.

Specified by:

apply in interface DPOperator

Parameters:

qs - the q-values on which the operator is to be applied.

Returns:

the result of the operator