burlap.behavior.singleagent.auxiliary.performance

Class PerformancePlotter.Trial

java.lang.Object

burlap.behavior.singleagent.auxiliary.performance.PerformancePlotter.Trial

Enclosing class:

PerformancePlotter

protected class PerformancePlotter.Trial
extends java.lang.Object

A datastructure for maintaining all the metric stats for a single trial.

Author:

James MacGlashan

Field Summary

Fields

Modifier and Type	Field and Description
java.util.List<java.lang.Double>	averageEpisodeReward Stores the average reward by episode
java.util.List<java.lang.Double>	cumulativeEpisodeReward Stores the cumulative reward by episode
java.util.List<java.lang.Double>	cumulativeStepEpisode Stores the cumulative steps by episode
java.util.List<java.lang.Double>	cumulativeStepReward Stores the cumulative reward by step
double	curEpisodeReward The cumulative reward of the episode so far
protected java.util.List<java.lang.Double>	curEpisodeRewards A list of the reward sequence in the current episode
int	curEpisodeSteps The number of steps in the episode so far
java.util.List<java.lang.Double>	medianEpisodeReward Stores the median reward by episode
java.util.List<java.lang.Double>	stepEpisode Stores the steps by episode
int	totalEpisodes The total number of episodes in the trial
int	totalSteps the total number of steps in the trial

Constructor Summary

Constructors

Modifier	Constructor and Description
protected	Trial()

Method Summary

Modifier and Type	Method and Description
void	setupForNewEpisode() Completes the last episode and sets up the datastructures for the next episode
void	stepIncrement(double r) Updates all datastructures with the reward received from the last step

Methods inherited from class java.lang.Object

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

Field Detail

cumulativeStepReward

public java.util.List<java.lang.Double> cumulativeStepReward

Stores the cumulative reward by step

cumulativeEpisodeReward

public java.util.List<java.lang.Double> cumulativeEpisodeReward

Stores the cumulative reward by episode

averageEpisodeReward

public java.util.List<java.lang.Double> averageEpisodeReward

Stores the average reward by episode

medianEpisodeReward

public java.util.List<java.lang.Double> medianEpisodeReward

Stores the median reward by episode

cumulativeStepEpisode

public java.util.List<java.lang.Double> cumulativeStepEpisode

Stores the cumulative steps by episode

stepEpisode

public java.util.List<java.lang.Double> stepEpisode

Stores the steps by episode

curEpisodeReward

public double curEpisodeReward

The cumulative reward of the episode so far

curEpisodeSteps

public int curEpisodeSteps

The number of steps in the episode so far

totalSteps

public int totalSteps

the total number of steps in the trial

totalEpisodes

public int totalEpisodes

The total number of episodes in the trial

curEpisodeRewards

protected java.util.List<java.lang.Double> curEpisodeRewards

A list of the reward sequence in the current episode

Constructor Detail

Trial

protected Trial()

Method Detail

stepIncrement

public void stepIncrement(double r)

Updates all datastructures with the reward received from the last step

Parameters:

r - the last reward received

setupForNewEpisode

public void setupForNewEpisode()

Completes the last episode and sets up the datastructures for the next episode