Home | News | Documentation | Download

How to fit a RooDecay (lifetime) in a range

Dear Roofit expert I need your help.

I want to fit a very simple RooDecay times an efficiency function.
I can fit this dataset with a Root TF1 very well, but when I try with roofit I get a consistent bias in the time constant. Reading old threads I think the problem lies in an uncorrect normalization since I have point only starting from ct = 0.007.

This is a snipplet of my code:

    auto ct = new RooRealVar("ct","ct [cm]",0.007,0.5);
    auto ctau = new RooRealVar("ctau", "ctau",  ctau_mc, 0.045, 0.055);
    auto ctTruth = new RooTruthModel("ctTruth", "ctTruth", *ct);

    auto ct_wReso = new RooDecay("ct_wReso","decay",*ct, *ctau, *ctTruth, RooDecay::SingleSided);
    auto ctEffFunc = new RooFormulaVar("ctEffFunc","eff Model", "exp(@0+@1*ct)*ROOT::Math::Chebyshev4(BsCt2DMC,@2,@3,@4,@5,@6)",   RooArgList(ctp0, ctp1, ctp2, ctp3, ctp4, ctp5, ctp6, *ct)); // Efficiency function fitted with Rooot

    auto ctSigPdf = new RooEffProd("ctSigPdf","model with efficiency", *ct_wReso, *ctEffFunc);
    // --- FIT

    RooFitResult* fitRes = ctSigPdf->fitTo(*data,Save(),NumCPU(8));

I tried evrything, from adding a "number of events as

    auto ctSigPdf = new RooEffProd("ctSigPdf","model with efficiency", *ct_wReso, *ctEffFunc);

    auto nSig = new RooRealVar("nSig", "Number of Signal Events", 0.6*data->sumEntries(), 0.,1.5*data->sumEntries());

    auto ctSigPdf_norm = new RooAddPdf("ctSigPdf_norm","ctSigPdf_norm", RooArgList(*ctSigPdf), RooArgList(*nSig));

To add a manual range as

RooFitResult* fitRes = ctSigPdf->fitTo(*data,Save(),NumCPU(8), Range(0.007,0.5));

The ctau fit value changes but nowere near the set value. The root fit:

TF1 *ctModel = new TF1("ctModel"," [8]*TMath::Exp(-x/[7])*(expo(0)*ROOT::Math::Chebyshev4(x,[2],[3],[4],[5],[6]))",0.007,0.5);

is spot on, so there is an issue with my RooFit model.

Can someone help me?

Thanks,

Alberto

May be @moneta or @StephanH can help you.

Hello Alberto,

I cannot see a problem in the snippet you posted, but it’s not enough to run anything myself.
First of all, do you get warnings / errors / is Minuit reporting a successful fit?

Hi Stephan,
the fit converges without any problem, simply at a wrong value.

Ok, hard to say what this is. I guess you need to attach the full macro.

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