npstat is hosted by Hepforge, IPPP Durham
NPStat  5.10.0
npstat::NonparametricCompositeBuilder< Point > Class Template Reference

#include <NonparametricCompositeBuilder.hh>

Inheritance diagram for npstat::NonparametricCompositeBuilder< Point >:
npstat::AbsCompositeDistroBuilder< Point >

Public Types

typedef AbsCompositeDistroBuilder< Point > B
 
typedef B::WeightedPointPtr WeightedPointPtr
 
typedef B::WeightedPtrVec WeightedPtrVec
 
typedef B::WeightedValue WeightedValue
 
typedef B::WeightedValueVec WeightedValueVec
 
- Public Types inherited from npstat::AbsCompositeDistroBuilder< Point >
typedef CompositeDistributionND result_type
 
typedef std::pair< const Point *, double > WeightedPointPtr
 
typedef std::vector< WeightedPointPtr > WeightedPtrVec
 
typedef std::pair< typename Point::value_type, double > WeightedValue
 
typedef std::vector< WeightedValue > WeightedValueVec
 

Public Member Functions

 NonparametricCompositeBuilder (AbsCopulaSmootherBase *smoothND, const std::vector< AbsMarginalSmootherBase * > &smooth1D, AbsResponseBoxBuilder< Point > *boxBuilder, bool takePointerOwnership, unsigned interpolationDegreeCopula=1U, unsigned interpolationDegreeMarginals=1U)
 
double lastCopulaBandwidth () const
 
double lastMarginBandwidth (const unsigned i) const
 
unsigned dim () const
 
void setArchive (gs::AbsArchive *ar, const char *category=0)
 
- Public Member Functions inherited from npstat::AbsCompositeDistroBuilder< Point >
virtual CompositeDistributionNDbuild (unsigned long uniqueId, const double *predictorCoords, unsigned nPredictors, const BoxND< double > &predictorBox, std::vector< OrderedPointND< Point > > &data) const
 
virtual CompositeDistributionNDbuildWeighted (unsigned long uniqueId, const double *predictorCoords, unsigned nPredictors, const BoxND< double > &predictorBox, const WeightedPtrVec &data, const unsigned *dimsToUse, unsigned nDimsToUse) const
 

Detailed Description

template<class Point>
class npstat::NonparametricCompositeBuilder< Point >

A class for building nonparametric estimates of multivariate densities by performing separate density estimation for each marginal and the copula. This allows us to introduce a larger number of bandwidth parameters into the system which can often lead to a better estimate.

Constructor & Destructor Documentation

◆ NonparametricCompositeBuilder()

template<class Point >
npstat::NonparametricCompositeBuilder< Point >::NonparametricCompositeBuilder ( AbsCopulaSmootherBase smoothND,
const std::vector< AbsMarginalSmootherBase * > &  smooth1D,
AbsResponseBoxBuilder< Point > *  boxBuilder,
bool  takePointerOwnership,
unsigned  interpolationDegreeCopula = 1U,
unsigned  interpolationDegreeMarginals = 1U 
)

Constructor arguments are as follows:

smoothND – copula smoother pointer

smooth1D – array of pointers for margin smoothers

boxBuilder – pointer for the response box builder

takePointerOwnership – if true, delete smoothND, boxBuilder, and pointers inside smooth1D in the destructor

interpolationDegreeCopula – interpolation degree for the copula density array. Currently must be 0 or 1, limited by the abilities of BinnedDensityND class.

interpolationDegreeMarginals – interpolation degree for the marginal density array. Currently must be 0 or 1, limited by the abilities of BinnedDensity1D class.

Member Function Documentation

◆ dim()

template<class Point >
unsigned npstat::NonparametricCompositeBuilder< Point >::dim ( ) const
inline

Dimensionality of the reconstructed density

◆ lastCopulaBandwidth()

template<class Point >
double npstat::NonparametricCompositeBuilder< Point >::lastCopulaBandwidth ( ) const
inline

Bandwidth factor used in the most recent copula smoothing

◆ lastMarginBandwidth()

template<class Point >
double npstat::NonparametricCompositeBuilder< Point >::lastMarginBandwidth ( const unsigned  i) const
inline

Bandwidth used in the most recent smoothing of the marginals

◆ setArchive()

template<class Point >
void npstat::NonparametricCompositeBuilder< Point >::setArchive ( gs::AbsArchive *  ar,
const char *  category = 0 
)

Set the archive for storing the histograms


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