> > ok, so most of the difference is due to the extra signal events with > > mes<5.27. Can we apply the same cut (mes>5.27) in the RooFit > > implementation of the fit? Since the tail introduces large part of the > > sistematics it would be nice to cut it out as we did for the prl measurement. > > Hi, > I'm not sure that I've understood 100% what you say. > I'm fitting the number of events in the 5.27 -> 5.29 region as VubAnalysis > do. No signal events below 5.27 are considered here. > What you suggest is to fit the argus in the 5.21 -> 5.26 as VubAnalysis > and let the fit do the rest starting from those values or you're saying > something different? no, I just misinterpreted what you said: " The other small difference is the way in which signal events are computed: o VubAnalysis: the relative fraction of C.B. and Argus functions are fitted and then the C.B. function is integrated in the range 5.27-5.29 (with parameteres from the fit) and rescaled to the measured fraction. o VirVubFitter: the number of signal and background events are DIRECTLY fitted. The method is the "Extended Likelihood Fit" descibed by RooFitters here (Go to Documentation, Tutorials..) " you did not say that the range for the VirFit is 5.27-5.29. Daniele