Re: fit parameter errors

From: Edmond Offermann <edmondoffermann_at_yahoo.com>
Date: Sat, 1 Jul 2006 20:29:20 -0700 (PDT)


Hi Guy,

No, those who claim that parameter correlations are not taken into account
when taking the diagonal elements of the error matrix are wrong !!

You can read up on it by checking the "Minuit" documentation, in particular
section 7.3.1, page 41 "Error Matrix" amd further on .

Google for "Minuit documentation" and you will find for instance :

http://hep.fi.infn.it/minuit.pdf

Eddy

> Dear Rooters,
> I am frequently asked about the fit parameters errors given in
> the statistics/fit info box. People claim that the correlations
> in between parameters are not included in the parametors errors.
> Is it right?
> From the screen output of the fit parameters, it is not clear to
> me whether the diogonal elements of the error matrix are the only
> ones taken into account (thus meaning no correlation taken into
> account).
> Well, I was just looking into the TMinuit class, source file
> section, seems to be the case.
> Thanks for your help,
> Guy.
>
>
>
Received on Sun Jul 02 2006 - 05:29:30 MEST

This archive was generated by hypermail 2.2.0 : Mon Jan 01 2007 - 16:31:59 MET