Problems with RooProdPdf normalization and fit

Dear experts,

I really need your help! I must say that I’m rather new in using RooFit, but I’m experiencing the following problem: I would like to do a signal-over-background fit, where the background is defined as a RooProdPdf between a function (called “transfer function” TF) and a RooHistPdf (obtained from the so-called “control region”). The signal is instead a RooHistPdf obtained from a ROOT histo.

If I fix the parameters of TF the fit works and I obtain the number of expected events for background and signal, but when I try to leave the TF parameters free (which btw is my goal) I get absurd results!

I was thinking that maybe there is something that I’m mistaking when dealing with RooProdPdf, otherwise, I really don’t know what’s happening.

Thanks for your help, really much appreciated!


P.S.: I’m posting my code, let me know if there are some serious errors :wink:

NEW_ask.C (3.5 KB)

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.

Hi Davide,

did you try the inverse experiment, which is to multiply the TF with one? If the transfer function by itself is broken, the ProdPdf will just multiply broken things with the RooHistPdf.