A factory for building PDFs and data for a number counting combination.

The factory produces a PDF for N channels with uncorrelated background uncertainty. Correlations can be added by extending this PDF with additional terms. The factory relates the signal in each channel to a master signal strength times the expected signal in each channel. Thus, the final test is performed on the master signal strength. This yields a more powerful test than letting signal in each channel be independent.

The problem has been studied in these references:

One can incorporate uncertainty on the expected signal by adding additional terms. For the future, perhaps this factory should be extended to include the efficiency terms automatically.

Definition at line 22 of file NumberCountingPdfFactory.h.

Public Member Functions
	NumberCountingPdfFactory ()
	constructor

virtual	~NumberCountingPdfFactory ()
	destructor

void	AddData (Double_t mainMeas, Double_t bkgMeas, Double_t db, Int_t nbins, RooWorkspace ws, const char *dsName="NumberCountingData")
	Arguments are an array of results from a main measurement, a measured background, and relative background uncertainty (eg.

void	AddDataWithSideband (Double_t mainMeas, Double_t sideband, Double_t tau, Int_t nbins, RooWorkspace ws, const char *dsName="ExpectedNumberCountingData")
	Arguments are an array of expected signal, expected background, and relative background uncertainty (eg.

void	AddExpData (Double_t sigExp, Double_t bkgExp, Double_t db, Int_t nbins, RooWorkspace ws, const char *dsName="ExpectedNumberCountingData")
	Arguments are an array of expected signal, expected background, and relative background uncertainty (eg.

void	AddExpDataWithSideband (Double_t sigExp, Double_t bkgExp, Double_t tau, Int_t nbins, RooWorkspace ws, const char *dsName="NumberCountingData")
	Arguments are an array of expected signal, expected background, and relative ratio of background expected in the sideband to that expected in signal region, and the number of channels.

void	AddModel (Double_t sigExp, Int_t nchan, RooWorkspace ws, const char pdfName="CombinedPdf", const char masterSignalName="masterSignal")
	This method produces a PDF for N channels with uncorrelated background uncertainty.

Private Member Functions
RooRealVar *	SafeObservableCreation (RooWorkspace ws, const char varName, Double_t value)
	need to be careful here that the range of observable in the dataset is consistent with the one in the workspace don't rescale unless necessary.

RooRealVar *	SafeObservableCreation (RooWorkspace ws, const char varName, Double_t value, Double_t maximum)
	need to be careful here that the range of observable in the dataset is consistent with the one in the workspace don't rescale unless necessary.

#include <RooStats/NumberCountingPdfFactory.h>

Constructor & Destructor Documentation

◆ NumberCountingPdfFactory()

NumberCountingPdfFactory::NumberCountingPdfFactory ( )

constructor

Definition at line 66 of file NumberCountingPdfFactory.cxx.

◆ ~NumberCountingPdfFactory()

NumberCountingPdfFactory::~NumberCountingPdfFactory ( )

virtual

destructor

Definition at line 72 of file NumberCountingPdfFactory.cxx.

Member Function Documentation

◆ AddData()

void NumberCountingPdfFactory::AddData	(	Double_t *	mainMeas,
		Double_t *	back,
		Double_t *	back_syst,
		Int_t	nbins,
		RooWorkspace *	ws,
		const char *	dsName = `"NumberCountingData"`
	)

Arguments are an array of results from a main measurement, a measured background, and relative background uncertainty (eg.

0.1 for 10% uncertainty), and the number of channels.

Definition at line 223 of file NumberCountingPdfFactory.cxx.

◆ AddDataWithSideband()

void NumberCountingPdfFactory::AddDataWithSideband	(	Double_t *	mainMeas,
		Double_t *	sideband,
		Double_t *	tauForTree,
		Int_t	nbins,
		RooWorkspace *	ws,
		const char *	dsName = `"ExpectedNumberCountingData"`
	)

Arguments are an array of expected signal, expected background, and relative background uncertainty (eg.

0.1 for 10% uncertainty), and the number of channels.

Definition at line 303 of file NumberCountingPdfFactory.cxx.

◆ AddExpData()

void NumberCountingPdfFactory::AddExpData	(	Double_t *	sig,
		Double_t *	back,
		Double_t *	back_syst,
		Int_t	nbins,
		RooWorkspace *	ws,
		const char *	dsName = `"ExpectedNumberCountingData"`
	)

Arguments are an array of expected signal, expected background, and relative background uncertainty (eg.

0.1 for 10% uncertainty), and the number of channels.

Definition at line 157 of file NumberCountingPdfFactory.cxx.

◆ AddExpDataWithSideband()

void NumberCountingPdfFactory::AddExpDataWithSideband	(	Double_t *	sigExp,
		Double_t *	bkgExp,
		Double_t *	tau,
		Int_t	nbins,
		RooWorkspace *	ws,
		const char *	dsName = `"NumberCountingData"`
	)

Arguments are an array of expected signal, expected background, and relative ratio of background expected in the sideband to that expected in signal region, and the number of channels.

Definition at line 177 of file NumberCountingPdfFactory.cxx.

◆ AddModel()

void NumberCountingPdfFactory::AddModel	(	Double_t *	sig,
		Int_t	nbins,
		RooWorkspace *	ws,
		const char *	pdfName = `"CombinedPdf"`,
		const char *	muName = `"masterSignal"`
	)

This method produces a PDF for N channels with uncorrelated background uncertainty.

It relates the signal in each channel to a master signal strength times the expected signal in each channel.

For the future, perhaps this method should be extended to include the efficiency terms automatically.

Definition at line 82 of file NumberCountingPdfFactory.cxx.

◆ SafeObservableCreation() [1/2]

RooRealVar * NumberCountingPdfFactory::SafeObservableCreation	(	RooWorkspace *	ws,
		const char *	varName,
		Double_t	value
	)

private

need to be careful here that the range of observable in the dataset is consistent with the one in the workspace don't rescale unless necessary.

If it is necessary, then rescale by x10 or a defined maximum.

Definition at line 197 of file NumberCountingPdfFactory.cxx.

◆ SafeObservableCreation() [2/2]

RooRealVar * NumberCountingPdfFactory::SafeObservableCreation	(	RooWorkspace *	ws,
		const char *	varName,
		Double_t	value,
		Double_t	maximum
	)

private

need to be careful here that the range of observable in the dataset is consistent with the one in the workspace don't rescale unless necessary.

If it is necessary, then rescale by x10 or a defined maximum.

Definition at line 206 of file NumberCountingPdfFactory.cxx.

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

roofit/roostats/inc/RooStats/NumberCountingPdfFactory.h
roofit/roostats/src/NumberCountingPdfFactory.cxx

Public Member Functions

Private Member Functions

Constructor & Destructor Documentation

◆ NumberCountingPdfFactory()

◆ ~NumberCountingPdfFactory()

Member Function Documentation

◆ AddData()

◆ AddDataWithSideband()

◆ AddExpData()

◆ AddExpDataWithSideband()

◆ AddModel()

◆ SafeObservableCreation() [1/2]

◆ SafeObservableCreation() [2/2]