Hi,
you cannot change with a simple option the errors in the chi2
calculations. You would need to write yourself the chi2 function and
pass to the fit,
using for example the option "U".
You can however use the option "L" to perform a binned likelihood fit.
In this case you will have the correct treatment of the bins
statistics in particular if you have empty bins
Best Regards
Lorenzo
On Nov 5, 2009, at 5:20 PM, Jordi Nadal Serrano wrote:
> Hello guys,
>
> I am trying to fit a function to a histogram. Like this:
>
> %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
>
> Double_t myfunction (Double_t *x, Double_t *par){
> int bin = data_8175->GetXaxis()->FindBin(x[0]);
> Double_t f = 0;
> f= par[0] * ttbar->GetBinContent(bin) + par[1] * h_w_jets-
> >GetBinContent(bin) + par[2] * h_others->GetBinContent(bin);
> return f;
>
> }
>
>
> void myfunc(){
>
> TF1 *f1 = new TF1("myfunc",myfunction,minBin_fit,maxBin_fit,3);
>
> f1->SetParLimits(0,0,2);
> f1->SetParLimits(1,0,2);
> f1->FixParameter(2,1);
> f1->Draw();
> }
> %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
>
> Once my histograms are filled. I do the fit this way:
>
> %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
> myfunc();
> TF1 *f1= (TF1*)gROOT->GetFunction("myfunc");
> data_8175->Fit("myfunc","ER0");
> %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
>
>
> I wonder what are the errors that are entered in the chi square. I
> believe that the errors are the sqrt( number of entries in a bin )
>
> Can I change these errors to be the theoretical ones, i.e, the sqrt
> of the value of myfunction per bin .
>
>
> Many thanks in advance,
>
> Jordi
>
Received on Fri Nov 06 2009 - 12:12:34 CET
This archive was generated by hypermail 2.2.0 : Fri Nov 06 2009 - 17:50:03 CET