TFractionFitter and RooHist (TGraphAsymmErrors)


is there a version of the TFractionFitter that takes RooHist (or TGraphAsymmErrors)? I would like to take asymmetric poisson errors into account, instead of symmetric gauss errors.

Thanks Duc

I have contacted Frank Filthaut about your question.


I cannot imagine the use of TGraphAsymmErrors being appropriate. I haven’t studied that class in any detail, but its mere name suggests assumptions (uncertainties that are treated in a Gaussian approximation) that run counter to the TFractionFitter philosophy (please look at the original references if you need more detail on that. TfractionFitter does use proper Poisson errors).

Observing that RooHist inherits from TGraphAsymmErrors, it would seem that the same comment applies – unless you want to use the RooHist as a regular TH1? But in that case I fail to see the need for a RooHist to begin with.