Skip to content

Efficiency uncertainties to be computed with Copper-Pearson

Formula to be used for uncertainty computation in line: https://gitlab.cern.ch/cms-muonPOG/spark_tnp/-/blob/master/prepare.py#L27

should be something like:

def generateClopperPearsonInterval(num,den):
    confidenceLevel = 0.68
    alpha = 1 - confidenceLevel
    
    lowerLimit = round(ROOT.Math.beta_quantile(alpha/2,num,den-num + 1),4)
    if num==den:
        upperLimit=1
    else:
        upperLimit = round(ROOT.Math.beta_quantile(1-alpha/2,num + 1,den-num),4)
    return lowerLimit,upperLimit

the function computeEff will then provide 3 numbers (eff, up_err, low_err)

This is called in many parts of prepare.py:

Both propagated to getSF_cutAndCount and getSF

For the plotting part: we define histograms bin-by-bin https://gitlab.cern.ch/cms-muonPOG/spark_tnp/-/blob/master/prepare.py#L454

Should switch to TGraphAsymmErrors https://root.cern.ch/root/html534/TGraphAsymmErrors.html