bdm::egiwmix Class Reference

Mixture of egiws. More...

#include <emix.h>

List of all members.

Public Member Functions

egiwmix ()

Default constructor.

void set_parameters (const vec &w, const Array< egiw * > &Coms, bool copy=false)

vec mean () const

return expected value

vec sample () const

return a sample from the density

vec variance () const

return the expected variance

void mean_mat (mat &M, mat &R) const

expected values of the linear coefficient and the covariance matrix are written to M and R , respectively

double evallog_nn (const vec &val) const

In this instance, val= [theta, r]. For multivariate instances, it is stored columnwise val = [theta_1 theta_2 ... r_1 r_2 ].

double lognc () const

logarithm of the normalizing constant, $\mathcal{I}$

shared_ptr< epdf > marginal (const RV &rv) const

Return marginal density on the given RV, the remainig rvs are intergrated out.

void marginal (const RV &rv, emix &target) const

vec & _w ()

returns a pointer to the internal mean value. Use with Care!

void ownComs ()

Auxiliary function for taking ownership of the Coms().

egiw * _Coms (int i)

access function

void set_rv (const RV &rv)

Name its rv.

egiw * approx ()

Approximation of a GiW mix by a single GiW pdf.

vec est_theta () const

LS estimate of $\theta$ .

ldmat est_theta_cov () const

Covariance of the LS estimate.

void pow (double p)

Power of the density, used e.g. to flatten the density.

virtual double evallog (const vec &val) const

Evaluate normalized log-probability.

virtual vec evallog_m (const mat &Val) const

Evaluate normalized log-probability for many samples.

virtual vec evallog_m (const Array< vec > &Val) const

Evaluate normalized log-probability for many samples.

virtual string to_string ()

This method returns a basic info about the current instance.

virtual void to_setting (Setting &set) const

This method save all the instance properties into the Setting structure.

virtual void validate ()

This method TODO.

Constructors

void set_parameters (int dimx0, ldmat V0, double nu0=-1.0)

Constructors

Construction of each epdf should support two types of constructors:

empty constructor,

copy constructor,

The following constructors should be supported for convenience:

constructor followed by calling set_parameters()

constructor accepting random variables calling set_rv()

All internal data structures are constructed as empty. Their values (including sizes) will be set by method set_parameters(). This way references can be initialized in constructors.

void set_parameters (int dim0)

Access attributes

ldmat & _V ()

const ldmat & _V () const

double & _nu ()

const double & _nu () const

void from_setting (const Setting &set)

Matematical Operations

virtual mat sample_m (int N) const

Returns N samples, $[x_1 , x_2 , \ldots \$ from density $ f_x(rv)$ .

virtual shared_ptr< mpdf > condition (const RV &rv) const

Return conditional density on the given RV, the remaining rvs will be in conditioning.

virtual void qbounds (vec &lb, vec &ub, double percentage=0.95) const

Lower and upper bounds of percentage % quantile, returns mean-2*sigma as default.

Connection to other classes

Description of the random quantity via attribute rv is optional. For operations such as sampling rv does not need to be set. However, for marginalization and conditioning rv has to be set. NB:

bool isnamed () const

True if rv is assigned.

const RV & _rv () const

Return name (fails when isnamed is false).

Access to attributes

int dimension () const

Size of the random variable.

Protected Attributes

vec w

weights of the components

Array< egiw * > Coms

Component (epdfs).

bool destroyComs

Flag if owning Coms.

ldmat V

Extended information matrix of sufficient statistics.

double nu

Number of data records (degrees of freedom) of sufficient statistics.

int dimx

Dimension of the output.

int nPsi

Dimension of the regressor.

int dim

dimension of the random variable

RV rv

Description of the random variable.

Detailed Description

Mixture of egiws.

Member Function Documentation

void bdm::egiw::from_setting ( const Setting & set ) [inline, virtual, inherited]

Create Gauss-inverse-Wishart density

$f(rv) = GiW(V,\nu)$

from structure

                class = 'egiw';
                V     = [];               // square matrix
                nu    = [];               // scalar \nu ((almost) degrees of freedom)
                                                                  // when missing, it will be computed to obtain proper pdf
                dimx  = [];               // dimension of the wishart part
                rv = RV({'name'})         // description of RV
                rvc = RV({'name'})        // description of RV in condition

Reimplemented from bdm::epdf.

References bdm::egiw::dimx, bdm::UI::get(), bdm::egiw::nu, and bdm::egiw::V.

void bdm::egiwmix::set_parameters	(	const vec &	w,
		const Array< egiw * > &	Coms,
		bool	copy = `false`
	)

Set weights w and components Coms By default Coms are copied inside. Parameter copy can be set to false if Coms live externally. Use method ownComs() if Coms should be destroyed by the destructor.

References bdm_assert_debug, bdm_error, Coms, destroyComs, bdm::epdf::dim, bdm::epdf::dimension(), and w.

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

emix.h
emix.cpp


Public Member Functions
	egiwmix ()
	Default constructor.
void	set_parameters (const vec &w, const Array< egiw * > &Coms, bool copy=false)
vec	mean () const
	return expected value
vec	sample () const
	return a sample from the density
vec	variance () const
	return the expected variance
void	mean_mat (mat &M, mat &R) const
	expected values of the linear coefficient and the covariance matrix are written to `M` and `R` , respectively
double	evallog_nn (const vec &val) const
	In this instance, val= [theta, r]. For multivariate instances, it is stored columnwise val = [theta_1 theta_2 ... r_1 r_2 ].
double	lognc () const
	logarithm of the normalizing constant, $\mathcal{I}$
shared_ptr< epdf >	marginal (const RV &rv) const
	Return marginal density on the given RV, the remainig rvs are intergrated out.
void	marginal (const RV &rv, emix &target) const
vec &	_w ()
	returns a pointer to the internal mean value. Use with Care!
void	ownComs ()
	Auxiliary function for taking ownership of the Coms().
egiw *	_Coms (int i)
	access function
void	set_rv (const RV &rv)
	Name its rv.
egiw *	approx ()
	Approximation of a GiW mix by a single GiW pdf.
vec	est_theta () const
	LS estimate of $\theta$ .
ldmat	est_theta_cov () const
	Covariance of the LS estimate.
void	pow (double p)
	Power of the density, used e.g. to flatten the density.
virtual double	evallog (const vec &val) const
	Evaluate normalized log-probability.
virtual vec	evallog_m (const mat &Val) const
	Evaluate normalized log-probability for many samples.
virtual vec	evallog_m (const Array< vec > &Val) const
	Evaluate normalized log-probability for many samples.
virtual string	to_string ()
	This method returns a basic info about the current instance.
virtual void	to_setting (Setting &set) const
	This method save all the instance properties into the Setting structure.
virtual void	validate ()
	This method TODO.
Constructors
void	set_parameters (int dimx0, ldmat V0, double nu0=-1.0)
Constructors
Construction of each epdf should support two types of constructors: empty constructor, copy constructor, The following constructors should be supported for convenience: constructor followed by calling `set_parameters()` constructor accepting random variables calling `set_rv()` All internal data structures are constructed as empty. Their values (including sizes) will be set by method `set_parameters()`. This way references can be initialized in constructors.
void	set_parameters (int dim0)
Access attributes
ldmat &	_V ()
const ldmat &	_V () const
double &	_nu ()
const double &	_nu () const
void	from_setting (const Setting &set)
Matematical Operations
virtual mat	sample_m (int N) const
	Returns N samples, $[x_1 , x_2 , \ldots \$ from density .
virtual shared_ptr< mpdf >	condition (const RV &rv) const
	Return conditional density on the given RV, the remaining rvs will be in conditioning.
virtual void	qbounds (vec &lb, vec &ub, double percentage=0.95) const
	Lower and upper bounds of `percentage` % quantile, returns mean-2*sigma as default.
Connection to other classes
Description of the random quantity via attribute `rv` is optional. For operations such as sampling `rv` does not need to be set. However, for `marginalization` and `conditioning` `rv` has to be set. NB:
bool	isnamed () const
	True if rv is assigned.
const RV &	_rv () const
	Return name (fails when isnamed is false).
Access to attributes
int	dimension () const
	Size of the random variable.
Protected Attributes
vec	w
	weights of the components
Array< egiw * >	Coms
	Component (epdfs).
bool	destroyComs
	Flag if owning Coms.
ldmat	V
	Extended information matrix of sufficient statistics.
double	nu
	Number of data records (degrees of freedom) of sufficient statistics.
int	dimx
	Dimension of the output.
int	nPsi
	Dimension of the regressor.
int	dim
	dimension of the random variable
RV	rv
	Description of the random variable.