Classes
class	AbsSparseUnfoldingFilterND

class	SparseUnfoldingFilterND

class	SparseUnfoldingFilterNDReader

class	AbsSparseUnfoldND

class	EigenParameters

class	SmoothedEMSparseUnfoldND

class	SparseUnfoldingBandwidthScannerND

class	TrlanDiagnostics

Typedefs
typedef gs::StaticReader< SparseUnfoldingFilterNDReader >	StaticSparseUnfoldingFilterNDReader

Functions
template<class Matrix >
CPP11_auto_ptr< std::vector< Eigen::Triplet< double, typename Matrix::Index > > >	pruneCovariance (const Matrix &m, double tol)

template<class Matrix >
int	trlanEigensystem (const Matrix &covmat, const EigenParameters &params, std::vector< double > eigenvalues, std::vector< double > eigenvectors, TrlanDiagnostics *diagnostics)

Detailed Description

Namespace "emsunfold" (EMS stands for Expectation-Maximization with Smoothing) is used for classes and functions in the NPStat package which implement EMS unfolding and rely on the Eigen package for sparse linear algebra. See http://eigen.tuxfamily.org/

Typedef Documentation

◆ StaticSparseUnfoldingFilterNDReader

typedef gs::StaticReader<SparseUnfoldingFilterNDReader> emsunfold::StaticSparseUnfoldingFilterNDReader

Factory for deserializing filters representable by sparse matrices

Function Documentation

◆ pruneCovariance()

template<class Matrix >

CPP11_auto_ptr<std::vector<Eigen::Triplet<double,typename Matrix::Index> > > emsunfold::pruneCovariance	(	const Matrix &	m,
		double	tol
	)

This function will prune off-diagonal elements of a sparse symmetric positive-semidefinite matrix which correspond to correlation coefficients smaller in magnitude than "tol". This function will also make sure that the correlation coefficients do not exceed 1 in magnitude.

◆ trlanEigensystem()

template<class Matrix >

int emsunfold::trlanEigensystem	(	const Matrix &	covmat,
		const EigenParameters &	params,
		std::vector< double > *	eigenvalues,
		std::vector< double > *	eigenvectors,
		TrlanDiagnostics *	diagnostics
	)

Determine eigenvalues and eigenvectors of the argument covariance matrix using TRLAN, steered by the parameters given. This function returns the TRLAN error code (status). 0 means everything is OK. Consult the TRLAN user guide for the meaning of other error codes. Note that, even when 0 is returned, the number of eigenpairs actually found could be less than the number of eigenpairs requested and/or tail fraction stopping condition might not be satisfied.

The eigenvalues will be returned in the increasing order. Check diagnostics->nConverged() to see how many of them are returned. Note that the vector of eigenvalues can actually have more elements, but only "diagnostics->nConverged()" of them are valid. Eigenvectors will be placed into "eigenvectors" vector one after another. If the matrix "covmat" has "nrows" rows then &eigenvectors[0] will point to the first eigenvector, &eigenvectors[0] + nrows to the second, etc. The order of eigenvectors corresponds to the order of eigenvalues.

Classes