Troubles with data in RooFit

Roofit and RooStats
1

Hello everyone,

I’m using RooFit in order to fit a distribution and find a parameter.
The data are from a TTree called “data” and it has a branch called “costheta”.
I wrote this code

using namespace std;

using namespace RooFit;

void fit()
{
 TFile *input = new TFile("cosine.root");
 TTree *cos = (TTree*)input->Get("data");

 RooRealVar x("x","x",-1.,1.);

 RooRealVar c("c","c",0.375); 
 RooRealVar na("na","na",1500,500,4000);
 RooGenericPdf std("std","c*(1+x**2)",RooArgList(c,x));
 
 RooRealVar a("a","a",0,-1.,1.);
 RooRealVar nb("nb","nb",1500,500,4000);
 RooGenericPdf asy("asy","a*x",RooArgList(a,x));

 RooAddPdf p("p","p",RooArgList(std,asy),RooArgList(na,nb));
 RooDataSet xdata("xdata","dataset in costheta", cos,x);

 RooFitResult *result = p.fitTo(xdata);

 RooPlot* xframe = x.frame();
 xdata.plotOn(xframe) ;
 p.plotOn(xframe);
 xframe->Draw();  
}

when i plot the RooDataSet xdata i get something completely wrong, i can’t understand why…obviously, the fit is not correct…i thought about a syntax error but it seems it is not the case…

Thanks in advance for help

2

Hi,

what do you mean by “completely wrong”? Could you share the input file in order to make the behaviour reproducible?

Danilo

3

Hi, thanks for the reply
The root file is this:
cosine.root (380 KB)
I hope it works, it is the first time i share i file here…

For “completely wrong” i mean this: the distribution is like a Dirac-delta function, all points are zero except one.

Thank you again

4

Hi,

thanks, that was the missing piece.
You are constructing your RooDataset in the wrong way. Here you have the correct standalone snippet:

using namespace RooFit;
TFile *input = new TFile("cosine.root");
TTree *cos1 = (TTree*)input->Get("data");

RooRealVar x("costheta","costheta",-1.,1.); // <- NOTICE THE NEW NAME HERE!!!!!
RooDataSet xdata("data for RooFit","dataset in costheta", cos1,x);

RooPlot* xframe = x.frame();
xdata.plotOn(xframe) ;
xframe->Draw(); 
c1->Draw()

Cheers,
D

5

Hi,

It worked perfectly, thank you very much for patience and time!