GFMath

class GFMath

    public:

                      GFMath()
                      GFMath(const GFMath&)
                      ~GFMath()
       static Float_t BinomialError(Int_t nRec, Int_t nGen, Bool_t upper)
       static TClass* Class()
       static TArrayD Efficiency(Double_t nCount, Double_t nDenom, Double_t nEffDenom, Double_t nFillDenom)
       static TArrayD EfficiencyBlobelError(Double_t nCount, Double_t nDenom, Double_t errCount, Double_t errDenom)
       static TArrayD EfficiencyGassnerError(Double_t nCount, Double_t nDenom, Double_t errCount, Double_t errDenom)
       static TArrayD EfficiencyTestError(Double_t nCount, Double_t nDenom, Double_t errCount, Double_t errDenom)
        static Bool_t EquidistBins(Double_t* bins, Int_t length, Double_t first, Double_t last)
        static Bool_t EquidistLogBins(Double_t* bins, Int_t length, Double_t first, Double_t last)
      virtual TClass* IsA() const
        static Bool_t IsEqualFloat(Double_t x1, Double_t x2)
       static TArrayD NormaliseTo1(const TArrayD& input, const TArrayD* inputErr = 0, TArrayD* outputErr = 0, TMatrixD* outputVariance = 0)
         virtual void ShowMembers(TMemberInspector& insp, char* parent)
         virtual void Streamer(TBuffer& b)
                 void StreamerNVirtual(TBuffer& b)
       static TArrayD X1ThroughX2BinomErr(Double_t x1, Double_t x2)
       static TArrayD X1ThroughX2CorrErr(Double_t x1, Double_t x2, Double_t sigmax1, Double_t sigmax2)

Data Members

Class Description

                                                                      
 GFMath                                                               
                                                                      
 Encapsulate Gero's math routines.                                    
                                                                      

 returns an array of length=2 with result[0] = x1/x2
                                   result[1] = sigma(x1/x2)
 assuming that x1 and x2 are 100% correlated

 returns an array of length=2 with result[0] = x1/x2
                                   result[1] = sigma(x1/x2)
 assuming that x1 is taken with binomial distribution from a sample of size x2

 return input normalised to its sum (no normalisation if sum == 0.)
 if inputErr is given, calculate outputErr and/or outputVariance (if given)
 assuming inputErr to be non-correlated
 no error calculation if inputErr has wrong length or sum == 0

 upper or lower 'error' of nRec/nGen from binomial according to V. Blobel

 returning 'efficiency = nCount/nDenom' and its error(s):
 nCount: surviving, nDenom: reference, nEffDenom: effective number entries for nDenom
 nFillDenom: hist fill actions for nDenom
 if(nEffDenom/nDenom >= 0.7): do asymm. binomial error GFMath::BinomialError
 then result[2/3] are the asymmetric errors
 result[1]: symmetric easy binomial error, but if asymmetric errors are possible,
 'scaled to nEffDenom'

 efficiency nCount/nDenom with 'Gassner' errors

 according to Blobel, 2004-09-08

 similar ansatz as Blobel, but rho = nCount/nDenom
 EXPERIMENTAL!!!

 Filling 'bins' with borders of 'nBins' bins between 'first' and 'last' that are
 equidistant when viewed in log scale, so 'bins' must have length nBins+1;
 If 'first', 'last' or 'nBins' are not positive, failure is reported.

 Filling 'bins' with borders of 'nBins' bins between 'first' and 'last' that are
 equidistant, so 'bins' must have length nBins+1;
 If 'first' >= 'last' or 'nBins' < 1, failure is reported.

 true if x1 , x2 are equal besides floating point surprises:
 difference < 1.e-6 and
 (if x1 != 0 && x2 != 0) relative deviation < 1.e-4
 (for the latter 0 means |x| < 1.e-6

Inline Functions

            TClass* Class()
            TClass* IsA() const
               void ShowMembers(TMemberInspector& insp, char* parent)
               void Streamer(TBuffer& b)
               void StreamerNVirtual(TBuffer& b)
             GFMath GFMath()
             GFMath GFMath(const GFMath&)
               void ~GFMath()