I’m noticing an issue similar to the one reported here: Difference in error between TEfficiency and Divide(TH1's). However I’m using a much more current version of ROOT. TEfficiency is giving me errors that are identical to TH2.Divide assuming completely uncorrelated numerator and denominator, while it should match TH2.Divide using the binomial option. Here is a notebook with a minimal reproduction of the issue. I’m sure it’s correct to use the TH2 divide option in this case, but figured this should be reported (or maybe I’m somehow using TEfficiency incorrectly).
Here is a notebook with a minimal reproduction of the issue: CERNBox
I cannot reproduce your problem. If I add this cell in your notebook, I get the same values for TH2::DIvide with option ‘B’ and TEFiiciency with the normal option:
for i in range (1,RealEfficiency2D_el_pt_eta.GetNbinsX()+1):
for j in range (1,RealEfficiency2D_el_pt_eta.GetNbinsY()+1):
ibin = RealEfficiency2D_el_pt_eta.GetBin(i,j)
print("\n")
print("h2",RealEfficiency2D_el_pt_eta.GetBinContent(ibin),"+/-",RealEfficiency2D_el_pt_eta.GetBinError(ibin))
print("TE",real_efficiency_el.GetEfficiency(ibin),"+/-",real_efficiency_el.GetEfficiencyErrorLow(ibin))