TFeldmanCousins

protected:
Int_t FindLimitsFromTable(Double_t mu) Double_t Prob(Int_t N, Double_t mu, Double_t B) public:
TFeldmanCousins(Double_t newCL = 0.9, TString options = ) TFeldmanCousins(const TFeldmanCousins&) virtual ~TFeldmanCousins() Double_t CalculateLowerLimit(Double_t Nobserved, Double_t Nbackground) Double_t CalculateUpperLimit(Double_t Nobserved, Double_t Nbackground) static TClass* Class() Double_t GetCL() const Double_t GetLowerLimit() const Double_t GetMuMax() const Double_t GetMuMin() const Double_t GetMuStep() const Double_t GetNbackground() const Double_t GetNMax() const Double_t GetNobserved() const Double_t GetUpperLimit() const virtual TClass* IsA() const TFeldmanCousins& operator=(const TFeldmanCousins&) void SetCL(Double_t newCL) void SetMuMax(Double_t newMax = 50.0) void SetMuMin(Double_t newMin = 0.0) void SetMuStep(Double_t newMuStep = 0.005) void SetNbackground(Double_t Nbg) void SetNobserved(Double_t NObs) virtual void ShowMembers(TMemberInspector& insp, char* parent) virtual void Streamer(TBuffer& b) void StreamerNVirtual(TBuffer& b)

protected:
Double_t fCL confidence level as a fraction [e.g. 90% = 0.9] Double_t fUpperLimit the calculated upper limit Double_t fLowerLimit the calculated lower limit Double_t fNobserved input number of observed events Double_t fNbackground input number of background events Double_t fMuMin minimum value of signal to use in calculating the tables Double_t fMuMax maximum value of signal to use in calculating the tables Double_t fMuStep the step in signal to use when generating tables Int_t fNMuStep = (int)(fMuStep) Int_t fNMax = (int)(fMuMax) Int_t fQUICK take a short cut to speed up the process of generating a

Class Description

 TFeldmanCousins

 class to calculate the CL upper limit using
 the Feldman-Cousins method as described in PRD V57 #7, p3873-3889

 The default confidence interval calvculated using this method is 90%
 This is set either by having a default the constructor, or using the
 appropriate fraction when instantiating an object of this class (e.g. 0.9)

 The simple extension to a gaussian resolution function bounded at zero
 has not been addressed as yet -> `time is of the essence' as they write
 on the wall of the maze in that classic game ...

    VARIABLES THAT CAN BE ALTERED
    -----------------------------
 => depending on your desired precision: The intial values of fMuMin,
 fMuMax, fMuStep and fNMax are those used in the PRD:
   fMuMin = 0.0
   fMuMax = 50.0
   fMuStep= 0.005
 but there is total flexibility in changing this should you desire.


 see example of use in $ROOTSYS/tutorials/FeldmanCousins.C

 see note about: "Should I use TRolke, TFeldmanCousins, TLimit?"
  in the TRolke class description.

 Author: Adrian Bevan, Liverpool University

 Copyright Liverpool University 2001       bevan@slac.stanford.edu

Double_t CalculateLowerLimit(Double_t Nobserved, Double_t Nbackground)

 given Nobserved and Nbackground, try different values of mu that give lower limits that
 are consistent with Nobserved.  The closed interval (plus any stragglers) corresponds  
 to the F&C interval

Double_t CalculateUpperLimit(Double_t Nobserved, Double_t Nbackground)

 given Nobserved and Nbackground, try different values of mu that give upper limits that
 are consistent with Nobserved.  The closed interval (plus any stragglers) corresponds  
 to the F&C interval

Int_t FindLimitsFromTable( Double_t mu )

 calculate the probability table for a given mu for n = 0, NMAX
 and return 1 if the number of observed events is consistent   
 with the CL bad

TFeldmanCousins

class TFeldmanCousins : public TObject

Data Members

Class Description