shouldn’t there be no difference between calling the fitTo method
and doing an interactive minimization like
RooAbsReal* nll = model.createNLL(data);
since fitTo performs a maximum likelihood fit “executed through the MINUIT commands MIGRAD, HESSE and MINOS in succession”. When I do both procedures I get different results.
The first one should be equivalent to the second one without hesse and minos. If you want to run those in fitTo, I think you need to pass the option explicitly in the fitTo command.
ok, actually I thought so. But in the documentation of RooAbsPdf it is said that fitTo calls migrad hesse and minos by default. (root.cern.ch/root/html/RooAbsPdf … sPdf:fitTo)
Anyway, I checked it and like you said fitTo yields the same results as running migrad only.