Problem to fit with RooNovosibirsk

Roofit and RooStats
1

Dear experts,

I got a problem to fit my background sideband dimuon distribution with
RooNovisibirsk. It gives me stupid result, just a vertical line on data plot
and initial parameters I put it: peak, width, tail.

I use ROOT 6.30/02 and my .C macro is on

/afs/cern.ch/work/a/anikiten/public/CMSSW_11_0_4/src/HiggsAnalysis/CombinedLimit/test/13TeV_2017UL_RooFitTest_SR3_NOVOSIBIRSK_FTEST.C

it uses root file with dumuon data at the same dir. So, you can try it by yourself.

Thanks for help !

Cheers, Sasha.

2

I guess @jonas can help

3

Hello Joans, All,

any progress to solve my problem. I have a CMS SUS General meeting next week where I would liketo show fit with RooNovosibirsk as a background for our dimuon analysis.

Thanks, Sasha.

4

Dear All,

is anyone working on problem with RooNovosibirsk ? Just to make sure that it is not forgotten. We asked by CMS SUSY PAG to do a bias test with two alternative functions. One is Chebychev which run w/o problems. For the second one I would like to try RooNovosibirsk.

Thank you, Sasha.

5

Hi @anikiten!

line 196 in your script seems wrong:

  RooProduct yieldBkgSideband("yieldBkgSideband","yieldBkgSideband",(norm,*intBkgSideband));

Why are you using round brackets there? In C++, doing this evaluates to the last operand of the comma operator. You can even see a warning, maybe you missed that:

13TeV_2017UL_RooFitTest_SR3_NOVOSIBIRSK_FTEST.C:196:70: warning: left operand of comma operator has no effect [-Wunused-value]
  RooProduct yieldBkgSideband("yieldBkgSideband","yieldBkgSideband",(norm,*intBkgSideband));

I think you meant to use {norm,*intBkgSideband} to implicitly create a RooFit collection like RooArgSet or RooArgList.

Maybe doing this correctly already fixes the problem?

If not, let me know!

Jonas

6

Hi Jonas,

thanks for check !

Replacing () to {} did not help unfortunately.

Cheers, Sasha.