OpenANN::Environment Class Reference

A reinforcement learning environment. More...

#include <Environment.h>

Inheritance diagram for OpenANN::Environment:

Public Member Functions
virtual	~Environment ()
virtual void	restart ()=0
	Restart environment. More...
virtual const State &	getState () const =0
	Get current state. More...
virtual const Action &	getAction () const =0
	Get current action. More...
virtual void	stateTransition (const Action &action)=0
	Perform an action. More...
virtual double	reward () const =0
	Get reward. More...
virtual bool	terminalState () const =0
	Check for terminal state. More...
virtual bool	successful () const =0
	Check if the agent was successful. More...
virtual int	stepsInEpisode () const =0
	Number of steps during the episode. More...
virtual double	deltaT () const
	Time between two simulation steps. More...
Public Member Functions inherited from OpenANN::StateSpace
virtual	~StateSpace ()
virtual int	stateSpaceDimension () const =0
virtual bool	stateSpaceContinuous () const =0
virtual int	stateSpaceElements () const =0
virtual const State &	stateSpaceLowerBound () const =0
virtual const State &	stateSpaceUpperBound () const =0
virtual const S &	getDiscreteStateSpace () const =0
Public Member Functions inherited from OpenANN::ActionSpace
virtual	~ActionSpace ()
virtual int	actionSpaceDimension () const =0
virtual bool	actionSpaceContinuous () const =0
virtual int	actionSpaceElements () const =0
virtual const Action &	actionSpaceLowerBound () const =0
virtual const Action &	actionSpaceUpperBound () const =0
virtual const A &	getDiscreteActionSpace () const =0

Additional Inherited Members
Public Types inherited from OpenANN::StateSpace
typedef Eigen::VectorXd	State
typedef std::vector< State >	S
Public Types inherited from OpenANN::ActionSpace
typedef Eigen::VectorXd	Action
typedef std::vector< Action >	A

Detailed Description

A reinforcement learning environment.

Constructor & Destructor Documentation

virtual OpenANN::Environment::~Environment ( )

inlinevirtual

Member Function Documentation

virtual double OpenANN::Environment::deltaT ( ) const

inlinevirtual

Time between two simulation steps.

Reimplemented in DoublePoleBalancing, and SinglePoleBalancing.

virtual const Action& OpenANN::Environment::getAction ( ) const

pure virtual

Get current action.

Returns

action

Implemented in DoublePoleBalancing, and SinglePoleBalancing.

virtual const State& OpenANN::Environment::getState ( ) const

pure virtual

Get current state.

Returns

state

Implemented in DoublePoleBalancing, and SinglePoleBalancing.

virtual void OpenANN::Environment::restart ( )

pure virtual

Restart environment.

Implemented in DoublePoleBalancing, and SinglePoleBalancing.

virtual double OpenANN::Environment::reward ( ) const

pure virtual

Get reward.

Returns

reward for the last state transition

Implemented in DoublePoleBalancing, and SinglePoleBalancing.

virtual void OpenANN::Environment::stateTransition ( const Action &  action)

pure virtual

Perform an action.

Parameters

action

next action

Implemented in DoublePoleBalancing, and SinglePoleBalancing.

virtual int OpenANN::Environment::stepsInEpisode ( ) const

pure virtual

Number of steps during the episode.

Implemented in DoublePoleBalancing, and SinglePoleBalancing.

virtual bool OpenANN::Environment::successful ( ) const

pure virtual

Check if the agent was successful.

Returns

was the agent successful?

Implemented in DoublePoleBalancing, and SinglePoleBalancing.

virtual bool OpenANN::Environment::terminalState ( ) const

pure virtual

Check for terminal state.

Returns

is the environment in a terminal state?

Implemented in DoublePoleBalancing, and SinglePoleBalancing.

---

The documentation for this class was generated from the following file:

• OpenANN/rl/Environment.h