burlap.behavior.singleagent.learnfromdemo.mlirl.differentiableplanners.dpoperator

Class SubDifferentiableMaxOperator

java.lang.Object

burlap.behavior.singleagent.planning.stochastic.dpoperator.BellmanOperator

burlap.behavior.singleagent.learnfromdemo.mlirl.differentiableplanners.dpoperator.SubDifferentiableMaxOperator

All Implemented Interfaces:

DifferentiableDPOperator, DPOperator

public class SubDifferentiableMaxOperator
extends BellmanOperator
implements DifferentiableDPOperator

Provides the sub gradient of the BellmanOperator max operator.

Author:

James MacGlashan.

Constructor Summary

Constructors

Constructor and Description
SubDifferentiableMaxOperator()

Method Summary

Modifier and Type	Method and Description
FunctionGradient	gradient(double[] qs, FunctionGradient[] qGradients) Returns the gradient of this DP operator, giving the Q-values on which it operates, their gradient.

Methods inherited from class burlap.behavior.singleagent.planning.stochastic.dpoperator.BellmanOperator

apply

Methods inherited from class java.lang.Object

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

Methods inherited from interface burlap.behavior.singleagent.planning.stochastic.dpoperator.DPOperator

apply

Constructor Detail

SubDifferentiableMaxOperator

public SubDifferentiableMaxOperator()

Method Detail

gradient

public FunctionGradient gradient(double[] qs,
                                 FunctionGradient[] qGradients)

Description copied from interface: DifferentiableDPOperator

Returns the gradient of this DP operator, giving the Q-values on which it operates, their gradient.

Specified by:

gradient in interface DifferentiableDPOperator

Parameters:

qs - the q-values

qGradients - the gradients of the Q-value

Returns:

the gradient of this operator