bdm::migamma Class Reference

Inverse-Gamma random walk. More...

#include <libEF.h>

Inheritance diagram for bdm::migamma:

Collaboration diagram for bdm::migamma:

List of all members.

Public Member Functions

migamma (const RV &rv, const RV &rvc)

Constructor.

void set_parameters (double k0)

Set value of k.

void condition (const vec &val)

Update ep so that it represents this mpdf conditioned on rvc = cond.

virtual vec samplecond (const vec &cond, double &ll)

Returns a sample from the density conditioned on cond, $x \sim epdf(rv|cond)$ .

virtual mat samplecond_m (const vec &cond, vec &ll, int N)

Returns.

virtual double evallogcond (const vec &dt, const vec &cond)

Shortcut for conditioning and evaluation of the internal epdf. In some cases, this operation can be implemented efficiently.

virtual vec evallogcond_m (const mat &Dt, const vec &cond)

Matrix version of evallogcond.

RV _rvc () const

access function

RV _rv () const

access function

epdf & _epdf ()

access function

epdf * _e ()

access function

Protected Attributes

eigamma epdf

Internal epdf that arise by conditioning on rvc.

double k

Constant $k$ .

vec * _beta

cache of epdf.beta

vec * _alpha

chaceh of epdf.alpha

RV rv

modeled random variable

RV rvc

random variable in condition

epdf * ep

pointer to internal epdf

Detailed Description

Inverse-Gamma random walk.

Mean value, $\mu$ , of this density is given by rvc . Standard deviation of the random walk is proportional to one $k$ -th the mean. This is achieved by setting $\alpha=\mu/k+2$ and $\beta=\mu(\alpha-1)$ .

The standard deviation of the walk is then: $\mu/\sqrt(k)$ .

Member Function Documentation

virtual vec bdm::mpdf::samplecond	(	const vec &	cond,
		double &	ll
	)			`[inline, virtual, inherited]`

Returns a sample from the density conditioned on cond, $x \sim epdf(rv|cond)$ .

Parameters:

	cond	is numeric value of `rv`
	ll	is a return value of log-likelihood of the sample.

Reimplemented in bdm::mprod.

References bdm::mpdf::condition(), bdm::mpdf::ep, bdm::epdf::evallog(), and bdm::epdf::sample().

Referenced by bdm::MPF< BM_T >::bayes(), and bdm::PF::bayes().

virtual mat bdm::mpdf::samplecond_m	(	const vec &	cond,
		vec &	ll,
		int	N
	)			`[inline, virtual, inherited]`

Returns.

Parameters:

	N	samples from the density conditioned on `cond`, $x \sim epdf(rv\|cond)$ .
	cond	is numeric value of `rv`
	ll	is a return value of log-likelihood of the sample.

References bdm::mpdf::condition(), bdm::RV::count(), bdm::mpdf::ep, bdm::epdf::evallog(), bdm::mpdf::rv, and bdm::epdf::sample().

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

work/git/mixpp/bdm/stat/libEF.h


Public Member Functions
	migamma (const RV &rv, const RV &rvc)
	Constructor.
void	set_parameters (double k0)
	Set value of `k`.
void	condition (const vec &val)
	Update `ep` so that it represents this mpdf conditioned on `rvc` = cond.
virtual vec	samplecond (const vec &cond, double &ll)
	Returns a sample from the density conditioned on `cond`, $x \sim epdf(rv\|cond)$ .
virtual mat	samplecond_m (const vec &cond, vec &ll, int N)
	Returns.
virtual double	evallogcond (const vec &dt, const vec &cond)
	Shortcut for conditioning and evaluation of the internal epdf. In some cases, this operation can be implemented efficiently.
virtual vec	evallogcond_m (const mat &Dt, const vec &cond)
	Matrix version of evallogcond.
RV	_rvc () const
	access function
RV	_rv () const
	access function
epdf &	_epdf ()
	access function
epdf *	_e ()
	access function
Protected Attributes
eigamma	epdf
	Internal epdf that arise by conditioning on `rvc`.
double	k
	Constant .
vec *	_beta
	cache of epdf.beta
vec *	_alpha
	chaceh of epdf.alpha
RV	rv
	modeled random variable
RV	rvc
	random variable in condition
epdf *	ep
	pointer to internal epdf