Re: fitting many TGraphAsymmErrors

From: Rene BRUN <brun_at_mail.cern.ch>
Date: Wed, 9 Feb 2005 22:26:55 +0100 (MET)


Hi Costas,

You can do this directly with TMinuit, implementing your own objective function (see $ROOTSYS/test/minexam example). I agree that the best solution would be TMultiGraph::Fit.

Anna Kreshuk is currently working on Fitters. I will see with her if she can implement this function in the coming days.

Rene Brun

On Wed, 9 Feb
2005, Costas Andreopoulos wrote:

> Hello,
>
> I have a function (a vN cross section as a function of various params)
> that I want to fit to a quite large number of cross section
> measurements. Since these measurements come from different experiments I
> would like to allow them to 'float' with respect to each other as a way
> to handle unknown relative differences in flux normalizations.
>
> So, technically I have a TF1 and a list of TGraphAsymmErrors
> (each one includes all the data points from a single experiment).
> In total, I am trying to fit *Nm* parameters related with
> a physics model + 1 parameter for each of the *Nd* experiments
> included in the fit.
>
> I need to fit the function to all graphs in the list *simultaneously*.
>
> Is there a simpler way to do this other than implementing
> by myself the Fit() method in TMultiGraph?
>
>
> cheers,
> Costas
>
Received on Wed Feb 09 2005 - 22:33:41 MET

This archive was generated by hypermail 2.2.0 : Tue Jan 02 2007 - 14:45:04 MET