Hi Alberto,
For your first question, see the tutorials about fitting
$ROOTSYS/tutorials/multifit.C (sum of 3 gaussians)
$ROOTSYS/tutorials/FittingDemo.C (more general example).
I also recommend reading the chapter about fitting in the Users Guide.
For your last question, see:
http://root.cern.ch/root/htmldoc/TMath.html#TMath:Gaus
Using the CVS version, you can fot a normalized gaussian with
hist.Fit("gausn")
instead of
hist.Fit("gaus")
Rene Brun
Alberto Pulvirenti wrote:
>
> Dear all,
>
> I have made a study where I needed to manage a histogram which should
> have a shape which is the result of the superimposition of two distinct
> gaussians, with similar mean (around 0) and very different constant &
> sigma values.
>
> I managed to realize a macro which performs the fit of two gaussians,
> which is initialized first excluding the 'signal' events (the higher and
> narrower one), and fitting the background. Then, I excluded such
> background and fitted the signal.
> Finally, using the values obtained from these partial fits, I generated
> a function which i just the sum of two gaussians, and I refitted it.
>
> Of course, the result can be approximative (sometimes I get large chi
> square). A colleague of mine told me that in PAW there was an automatic
> procedure to perform a double-gaussian fit. I was wondering if such a
> procedure is present also in ROOT or not.
>
> Another question about fit is the following. We know that in a
> normalized gaussian there is a precise relationship between the constant
> and sigma, and then such a function has only two (not three) free
> parameters. Is there any "automatic" function which performs this?
>
> Cheers
>
> --
> Alberto Pulvirenti, Ph. D.
> University of Catania / INFN Catania
> Address: Via S. Sofia, 64 I-95123 Catania
> Phone: +39-095-378-5286
This archive was generated by hypermail 2b29 : Sun Jan 02 2005 - 05:50:09 MET