burlap.behavior.singleagent.shaping.potential

Interface PotentialFunction

All Known Implementing Classes:

PotentialShapedRMax.RMaxPotential

public interface PotentialFunction

Defines an interface for reward potential functions. This interface will be used by potential-based reward shaping. Note: potential functions should always be defined to return 0 for terminal states.

Author:

James MacGlashan

Method Summary

Methods

Modifier and Type	Method and Description
double	potentialValue(State s) Returns the reward potential from the given state.

Method Detail

potentialValue

double potentialValue(State s)

Returns the reward potential from the given state. Note: the potential function should always return 0 for terminal states.

Parameters:

s - the input state for which to get the reward potential.

Returns:

the reward potential from the given state.