Profile likelihood fit using multiple histograms, each corresponding to different value of parameter of interest

Roofit and RooStats
1

Dear Roofit community,

First of all I must apologize for creating another topic similar to this one: Template Profile likelihood fit using morphed histograms but I haven’t received reply to my latest question and I am still struggling to make this work.

I will briefly repeat what my question is, because I may not have been clear enough and I am sorry for that.

I have different set of 1D histograms (~50 histograms) for one observable and each histograms represents simulation with different values of top quark decay width (from 0.05 GeV to ~5 GeV), these are NOT systematically varied histograms, these are just histograms representing different values of parameter of interest. I would like to extract the width from data using profile likelihood fit to reduce systematic uncertainties. Of course I also have systematically varied histograms for each width value, but this is not a problem. My question is how can I technically do this using RooHis/Roostats? To my understanding the problem is that we don’t have continuous parameter for our point of interest (top quark decay width) we have only set of histograms that are NOT ±1 sigma variations. My idea was to use template morphing manually for and somehow plug this into RooHist framework but I am not sure how to do this or if this is even possible.

Thanks and cheers,
Tomas

2

I am posting to keep this thread alive. I will give this one last try, anybody has any idea how to implement this?

Thanks and cheers,
Tomas

3

@moneta did reply to your first post in that thread. He is currently away and will surely have a look at this new one when he will be back.

4

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