Calculation of fit errors

ROOT
1

Consider this script:

void gausserrors()
{
  gStyle -> SetOptFit(kTRUE);
  TH1F * h = new TH1F("h1", "h1", 11, -5.0, 5.0);
  h -> FillRandom("gaus", 500);
  //for	(unsigned char bind = 1; bind <= h -> GetNbinsX(); bind ++)
  //  {
  //	h -> SetBinError(bind, 0.0);
  //  }
  h -> Fit("gaus");
  TCanvas *c = new TCanvas();
  h -> Draw();
}

How are the errors on fit parameters computed? And why does the fit fail if the bin errors are set to 0.0 (instead of producing a fit with error 0.0)?

Please read tips for efficient and successful posting and posting code

ROOT Version: Not Provided
Platform: Not Provided
Compiler: Not Provided

2

Bins with zero errors are called “empty bins”. See: TH1::Fit

3

There could be an option that produces the best fitting curve and forget about chi square in case of bins of nonzero content and zero errors.

4

Hi,
if you use a Likelihood fit, should work way better for a gaussian fit of an histogram , especially with only 500 entry. The likelihood treats better the bins with low number of entries

The method to use the Likelihood fit is

h -> Fit("gaus","L");

Cheers,
Stefano

1 Like
5

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