Event System, Logging & Statistics

This page contains the reference documentation for the event and logging system.

Event System

These are interfaces, classes and utility functions of the event system:

Subscriber	Event aggregation object.
Pubsub	Implementation of a message broker (Pubsub stands for publish and subscribe).
event_topic_factory	Constructs a proxy instance of the event interface, as required by EventService and LogStatsAggregator.
EventScope	Base class for all services that integrate with the event system and therefore use EventService as their backend.
EventService	Manages the recording of event invocations and provides simple event routing functionality.
EventCollection	A collection of EventRecord instances that can be queried by event specification.
EventRecord	This auxiliary class is used to record calls to the event interface

Event Logging

These are the components of the event system:

StepEventLog	Logs all events dispatched by the environment during one step.
EpisodeEventLog	Keeps logs of all events dispatched by an environment during one episode.
KpiCalculator	Interface for calculating KPI metrics.
LogEventsWriterRegistry	Handles registration of event log writers.
LogEventsWriter	Interface for modules writing out the event log data.
LogEventsWriterTSV	Writes event logs into TSV files.
EventRow	Represents one row into the output file for the LogEventsWriterTSV.
SimpleEventLoggingSetup	Simple setup for logging of environment events with all their attributes.
ObservationEvents	Event topic class with logging statistics based only on observations, therefore applicable to any valid reinforcement learning environment.
ActionEvents	Event topic class with logging statistics based only on Gym space actions, therefore applicable to any valid reinforcement learning environment.
RewardEvents	Event topic class with logging statistics based only on rewards, therefore applicable to any valid reinforcement learning environment.
create_categorical_plot	Checks the type of value and calls the correct plotting function accordingly.
create_histogram	Creates simple matplotlib histogram of value.
create_relative_bar_plot	Counts the categories in value and prepares a relative bar plot of these.
create_violin_distribution	Creates simple matplotlib violin plot of value.

Statistics Logging

These are the components of the statistics logging system:

LogStatsEnv	Interface to access logging statistics generated by the environment.
LogStatsWriterConsole	Log statistics writer implementation for the console, mainly for debugging purposes.
LogStatsWriterTensorboard	Log statistics writer implementation for Tensorboard.
LogStatsLevel	Log statistics aggregation levels.
LogStatsConsumer	An interface to receive log statistics.
LogStatsAggregator	Complements the event system by providing aggregation functionality.
LogStatsWriter	A minimal interface concrete log statistics writers must implement.
GlobalLogState	Internal class that encapsulates the global state of the logging system.
LogStatsLogger	Auxiliary class returned by get_stats_logger.
register_log_stats_writer	Set the concrete writer implementation that will receive all successive statistics logging.
log_stats	Helper function.
increment_log_step	Notifies the logging system that the current step is finished.
get_stats_logger	Creates an object that can be used to pipe LogStatAggregator instances with the logging writers.
define_step_stats	Event method decorator, defines a new step statistics calculation for this event.
define_episode_stats	Event method decorator, defines a new episode statistics calculation for this event.
define_epoch_stats	Event method decorator, defines a new epoch statistics calculation for this event.
define_stats_grouping	Event method decorator, defines a grouping of all calculated statistics by an attribute.
define_plot	Event method decorator, defines a plot.
histogram	the histogram reducer function
LogStatsValue	Basic data structure for log statistics
LogStatsGroup	Basic data structure for log statistics
LogStatsKey	Basic data structure for log statistics
LogStats	Basic data structure for log statistics