BIE::EnsembleStat Class Reference

Caches an group of posterior states in a vector and provides member functions to characterize the posterior from this distribution and produce statistical diagnostics on the convergence of the simulation. More...

#include <EnsembleStat.h>

Inheritance diagram for BIE::EnsembleStat:

[legend]

Collaboration diagram for BIE::EnsembleStat:

[legend]

List of all members.

ComputeDistribution
Computes the covariance matrix and its eigenvalues/vectors Return = 1 (variance computed), Return = 0 (too few values)
class	boost::serialization::access
void	ComputeDistribution ()
	Compute distribution estimate as needed.
void	ComputeDistribution (int n)
void	ComputeDistributionMarginal (ostream &out)
int	trim_location (int m)
	Index of non-null space for subspace m.
int	trim_location_marginal (int m, int n)
	Index of non-null space for subspace.
EnsembleStat *	New ()
	Object factory (clone).
double	PDF (State &)
	Differential distribution function P(x).
double	logPDF (State &)
	Log of differential distribution function P(x).
double	logPDFMarginal (int m, int n, vector< double > &V)
	Log of differential distribution function for a component in subspace.
double	PDF (int, State &)
	Differential distribution function for a single component.
vector< double >	lower (void)
	Lower bound on distribution (in each dimension).
vector< double >	upper (void)
	Upper bound on distribution (in each dimension).
vector< double >	Mean (int m)
	Return mean of distribution (mulitvariate).
vector< double >	StdDev (int m)
	Return standard deviation of distribution (mulitvariate).
vector< double >	Moments (int m, int k)
	Return specifided moment of distribution (mulitvariate).
vector< double >	Width (int m)
	Width.
vector< double >	Sample (int m)
	Return random variate from distribution for the given subspace m.
double	SampleOne (int m, int j)
	Return random variate from fully marginalized distribution.
vector< double >	StdDevMarginal (int m, int n)
	Retrieve the std dev for the component marginal.
vector< double >	MeanMarginal (int m, int n)
	Retrieve the mean for the component marginal.
vector< double >	SampleMarginal (int m, int n)
	Sample the distribution for the component marginal.
vector< double >	WidthMarginal (int m, int n)
	Width for component marginal.
void	PrintDiag (ostream &out)
	Print current covariance matrix (debug and diag).
void	PrintDiag (string &outfile)
	To a given file.
Public Member Functions
	EnsembleStat ()
	Constructor.
	EnsembleStat (int nmix, int ndim)
	Mixture constructor.
	EnsembleStat (int nmix, int ndim, int level, int nburn, string filename, int keypos)
	General constructor.
	EnsembleStat (const EnsembleStat &)
	Copy constructor.
	~EnsembleStat ()
	Destructor.
void	Reset (int nmix, int ndim)
	Reinitialize the EnsembleStat.
void	Reset (int nmix, int ndim, int level, int nburn, string filename, int keypos)
	Reinitialize the EnsembleStat.
void	setDimensions (int nmix, int ndim)
	Set total size of arrays.
Public Attributes
vector< Matrix >	covar
	The covariance matrix for all parameters and weights in the deque.
vector< vector< Matrix > >	covarM
	The covariance matrix for all parameters and weights in the deque by component.
vector< Matrix >	Teigen
	The transpose of the eigenvector matrix.
vector< vector< Matrix > >	TeigeM
	The transpose of the eigenvector matrix for the components.
vector< Matrix >	eigen
	The eigenvector matrix.
vector< vector< Matrix > >	eigenM
	The eigenvectors for each component in the subspace.
vector< Vector >	mean
	The mean for each dimension.
vector< vector< Vector > >	meanM
	The mean for each dimension for each component.
vector< Vector >	evals
	The eigenvalues.
vector< vector< Vector > >	evalsM
	The eigenvalues by component.
vector< Vector >	Tmean
	The mean in the eigenspace.
vector< vector< Vector > >	TmeanM
	The mean in the component eigenspace.
Static Public Attributes
static int	minsub
	Minimum number of states per subspace to retain subspace (default 3).
Protected Attributes
vector< int >	icut
	Trimmed dimension for each subspace (that is, remove the null space).
vector< vector< int > >	mcut
	Trimmed dimension for each subspace for each component.
vector< int >	mcum
	Fraction of states in each subspace per component.
bool	var_computed
	True of mean and variance has been computed.
vector< vector< double > >	vmean
	Mean for each dimension.
vector< vector< vector< double > > >	mmean
vector< vector< double > >	vvar
	Variance for each dimension.
vector< vector< vector< double > > >	mvar
Normal *	normal
	Normal distribution generator.
double	factor
	Scaling factor for realizing variates from the multidimensional Gaussian defined by the mean and covariance of the ensemble.

---

Detailed Description

Caches an group of posterior states in a vector and provides member functions to characterize the posterior from this distribution and produce statistical diagnostics on the convergence of the simulation.

The posterior state is characterized from the ensemble variance by a principal component analysis.

---

Constructor & Destructor Documentation

BIE::EnsembleStat::EnsembleStat	(	int	nmix,
		int	ndim
	)

Mixture constructor.

Parameters:

	nmix	is the mixture number
	ndim	is the dimension of each mixture copmonent

BIE::EnsembleStat::EnsembleStat	(	int	nmix,
		int	ndim,
		int	level,
		int	nburn,
		string	filename,
		int	keypos
	)

General constructor.

Parameters:

	nmix	is the mixture number
	ndim	is the dimension of each mixture copmonent
	level	is the update level
	nburn	is the desired number of converged states
	filename	is the logfile for ensemble statistics
	keypos	is the parameter index for computing PDF and CDF evaluations

---

Member Function Documentation

int BIE::EnsembleStat::trim_location_marginal	(	int	m,
		int	n
	)			[inline]

Index of non-null space for subspace.

Parameters:

	m	for component
	n

double BIE::EnsembleStat::logPDFMarginal	(	int	m,
		int	n,
		vector< double > &	V
	)			[virtual]

Log of differential distribution function for a component in subspace.

Parameters:

	m,component
	n,and	parameter vector
	V

Reimplemented from BIE::Ensemble.

void BIE::EnsembleStat::PrintDiag

(

ostream &

out

)

[virtual]

Print current covariance matrix (debug and diag).

To a given ostream

Reimplemented from BIE::Ensemble.

---

Member Data Documentation

vector<Matrix> BIE::EnsembleStat::covar

The covariance matrix for all parameters and weights in the deque.

This and Teigen and eigen are only defined after EnsembleStat::ComputeDistribution() is called.

vector< vector<Matrix> > BIE::EnsembleStat::covarM

The covariance matrix for all parameters and weights in the deque by component.

This and TeigenM and eigenM are only defined after EnsembleStat::ComputeDistributionMarginal() is called.

vector<Matrix> BIE::EnsembleStat::eigen

The eigenvector matrix.

Eigenvectors are ranked in decreasing order by their eigenvalues. The eigenvectors for the subspace

---

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

• /home/weinberg/src/BIE/include/EnsembleStat.h