burlap.behavior.singleagent.learnfromdemo.mlirl.differentiableplanners.diffvinit

Interface DifferentiableVInit

All Superinterfaces:

ParametricFunction, ValueFunction, ValueFunctionInitialization

All Known Implementing Classes:

LinearDiffRFVInit, LinearStateDiffVF, VanillaDiffVinit

public interface DifferentiableVInit
extends ValueFunctionInitialization, ParametricFunction

An interface for value function initialization that is differentiable with respect to some parameters. This interface is useful for DifferentiableSparseSampling which may be used to learn the value of leaf nodes in a finite horizon valueFunction.

Author:

James MacGlashan.

Nested Class Summary

Nested classes/interfaces inherited from interface burlap.behavior.valuefunction.ValueFunctionInitialization

ValueFunctionInitialization.ConstantValueFunctionInitialization

Nested classes/interfaces inherited from interface burlap.behavior.singleagent.vfa.ParametricFunction

ParametricFunction.ParametricStateActionFunction, ParametricFunction.ParametricStateFunction

Method Summary

Methods

Modifier and Type	Method and Description
FunctionGradient	getQGradient(State s, AbstractGroundedAction ga) Returns the Q-value function gradient.
FunctionGradient	getVGradient(State s) Returns the value function gradient.

Methods inherited from interface burlap.behavior.valuefunction.ValueFunctionInitialization

qValue

Methods inherited from interface burlap.behavior.valuefunction.ValueFunction

value

Methods inherited from interface burlap.behavior.singleagent.vfa.ParametricFunction

copy, getParameter, numParameters, resetParameters, setParameter

Method Detail

getVGradient

FunctionGradient getVGradient(State s)

Returns the value function gradient.

Parameters:

s - the state on which the value function is to be evaluated

Returns:

the value function gradient.

getQGradient

FunctionGradient getQGradient(State s,
                            AbstractGroundedAction ga)

Returns the Q-value function gradient.

Parameters:

s - the state on which the Q-value is to be evaluated.

ga - the action on which the Q-value is to be evaluated.

Returns:

the Q-value function gradient