ConditionalEstimator Class Reference

estimates conditional mean, var, etc. of a stochastic process More...

Inheritance diagram for ConditionalEstimator:

Public Member Functions
	ConditionalEstimator (const Property &property, StochasticProcess *stochvar, StochasticEventGenerator *condition, Time *time, int pre, int post, int updateCorrection=0)
	Construct. More...
	ConditionalEstimator (const Property &property, StochasticProcess *stochvar, StochasticProcess *condition, Time *time, int pre, int post, int updateCorrection=0)
	Construct. More...
void	setRejection (int pre, int post)
	Add rejection interval. More...
void	removeRejection ()
	Remove rejection. More...
virtual void	setParameter (const Property &p, double value)
	set parameter value
virtual void	collect ()
	receive next data point
virtual Matrix	getEstimate (const Property &)
	get some property
void	init ()
	initialize
virtual	~ConditionalEstimator ()
	destruct
Public Member Functions inherited from Estimator
	Estimator (StochasticProcess *src, Time *time, const string &name="", const string &type="Estimator")
	Construct. More...
virtual	~Estimator ()
	Destroy. More...
Public Member Functions inherited from Parametric
string	getType () const
	Type of object.
string	getName () const
	Name of object.
void	setName (const string &name)
	Name of object.
string	getConfiguration () const
	Returns the configuration of the object. More...
string	getAllParameters () const
	String with all parameter settings.
void	addParameter (const string &name)
	Add a parameter. More...
void	rmParameter (const string &name)
	Remove a parameter. More...
virtual void	setParameter (const string &name, const string &value)
	Set parameter. More...
virtual string	getParameter (const string &name) const
	Get parameter. More...

Additional Inherited Members
Protected Member Functions inherited from Parametric
	Parametric (const string &name, const string &type)
	Create object of type and name.
virtual	~Parametric ()
	Destroy object.
Protected Attributes inherited from Estimator
Property	nEstimate
	what to record
StochasticProcess *	pSource
	the source of data
uint	nSamples
	number of samples recorded
Time *	estimatorTime
	Time process for registering and running.

Detailed Description

estimates conditional mean, var, etc. of a stochastic process

The conditional estimator records conditional probabilities. It links to a stochastic variable, and an event (both of which are given in the constructor). The variable and the state of the event are recorded each time prepareNextState() is called. If the event is true at a timestep, then the object will go on recording for a number of time steps, and then take all the properties. This enables you to take for instance the mean of a process at 20 time steps before to 10 time step after the occurance of a special event. In that case the 20 time steps would be given as 'pre', the 10 time steps as 'post' to the constructor. The main purpose of this class (at the time it was written) was the recording of spike-triggered averages in neurons.

Constructor & Destructor Documentation

ConditionalEstimator::ConditionalEstimator	(	const Property &	property,
		StochasticProcess *	stochvar,
		StochasticEventGenerator *	condition,
		Time *	time,
		int	pre,
		int	post,
		int	updateCorrection = 0
	)

Construct.

This version constructs an object which records (and adds) samples of a stochastic process, depending on whether an event has occured: result = { stochvar(t), t in [pre,post] | condition(0) }.

Parameters

property	a collection of things to record
stochvar	the address of the stochastic variable to record
condition	the address of the event on which the recording is conditioned
time	the global time object
pre	number of time steps before the event
post	number of time steps after the event
updateCorrection	optional; defaults to 0. number of time steps to shift the result; this is to correct different update cascades

ConditionalEstimator::ConditionalEstimator	(	const Property &	property,
		StochasticProcess *	stochvar,
		StochasticProcess *	condition,
		Time *	time,
		int	pre,
		int	post,
		int	updateCorrection = 0
	)

Construct.

This version constructs an object which records (and adds) samples of a stochastic process, multiplied with the value of another process: result = { stochvar(t)*condition(0), t in [pre,post] }.

Parameters

property	a collection of things to record
stochvar	the address of the stochastic variable to record
condition	the address of the process on which the recording is conditioned
time	the global time object
pre	number of time steps before the event
post	number of time steps after the event
updateCorrection	optional; defaults to 0. number of time steps to shift the result; this is to correct different update cascades

Member Function Documentation

void ConditionalEstimator::removeRejection

(

)

Remove rejection.

Sets the estimator to normal mode again. See setRejection().

void ConditionalEstimator::setRejection	(	int	pre,
		int	post
	)

Add rejection interval.

Special. You will hardly ever need this. If a rejection interval is set, traces of the process where additional events fall into the interval -pre - post are rejected. (this changes the meausured process

to

Parameters

pre	start of rejection interval, counted from the event (t=0)
post	end of rejection interval, counted from the event (t=0)