Inheritance diagram for Mdp::DelayedQLearning:

Public Member Functions
	DelayedQLearning (std::shared_ptr< Context > c, TabularActionValues *av)

virtual void	updateActionValues (state_t previousState, state_t nextState, action_t previousAction, double reward)

Public Member Functions inherited from Mdp::EligibilityTraceAlgo
	EligibilityTraceAlgo (std::shared_ptr< Context > c, TabularActionValues *av)

virtual void	init ()

virtual void	end ()

Public Member Functions inherited from Mdp::RlBackupAlgorithm
	RlBackupAlgorithm (std::shared_ptr< Context > c, ActionValuesFunction *av)

virtual	~RlBackupAlgorithm ()

virtual double	getMaxQ (state_t state)

virtual std::pair< action_t, double >	getBestActionAndQ (state_t state)

virtual action_t	getBestAction (state_t state)

virtual void	updateBestActionAndQ (state_t state)

virtual void	notifyUpdateNeeded ()

Static Public Attributes
static constexpr const char *	configKey = "delayedQLearning"

Additional Inherited Members
Protected Member Functions inherited from Mdp::EligibilityTraceAlgo
void	updateState (state_t previousState, action_t previousAction, double reward)

Protected Member Functions inherited from Mdp::RlBackupAlgorithm
virtual void	initAlpha ()

virtual void	updateAlpha ()

virtual void	updateIfNeeded (state_t state)

Protected Attributes inherited from Mdp::EligibilityTraceAlgo
TabularActionValues *	tabularAv {nullptr}

state_t	previousPreviousState {0}

action_t	previousPreviousAction {0}

double	previousReward {0.0}

std::vector< std::vector< double > >	e

double	lambda {0.5}

double	discountFactor {0.5}

size_t	stateSize {0}

size_t	actionSize {0}

Protected Attributes inherited from Mdp::RlBackupAlgorithm
std::shared_ptr< Context >	context {nullptr}

ActionValuesFunction *	actionValues {nullptr}

double	alpha {-1.0}

double	alpha0 {0.1}

double	alphaCounter {1.0}

double	alphaDecaySpeed {1.0}

bool	hyperbolic {false}

bool	stepwise {false}

unsigned long long	stepwiseCounter {0}

unsigned long long int	stepLength {0}

std::vector< double >	bestQ

std::vector< action_t >	bestAction

std::vector< bool >	needsUpdate

Detailed Description

Note: this is not an eligibility trace algorithm, but it is here for testing purposes

Definition at line 21 of file delayedQLearning.h.

Constructor & Destructor Documentation

Mdp::DelayedQLearning::DelayedQLearning	(	std::shared_ptr< Context >	c,
		TabularActionValues *	av
	)

inline

Definition at line 26 of file delayedQLearning.h.

void DelayedQLearning::updateActionValues	(	state_t	previousState,
		state_t	nextState,
		action_t	previousAction,
		double	reward
	)

virtual

Definition at line 14 of file delayedQLearning.cpp.

constexpr const char* Mdp::DelayedQLearning::configKey = "delayedQLearning"

static

Definition at line 24 of file delayedQLearning.h.

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