Re: Landau mean value

As Rene says the mean is not that meaningful for Landau distribution. Without a cut it's not even well defined. What's often used for de/dx measurements is the truncated mean -- the mean of the smallest X% of samples where X~50.

An alterative data intrepretation is to to a likelihood fit for peak, etc. This uses all the data even out on the tails and his in princple better than truncated mean .. but usually not by much.

Anyway if you want the mean within some cut I think you just have to do the integrals (mean is not a sensible parameter for landau distribution).

ebar=Int(e*dedx(e),e=low..high)/Int(dedx,e=low.high)

This would be easy in maple and is prob. not hard in root. TF1 comes with integral functions so if you define a new TF1 that e times your landau function, you should be able to do both integrals.

Something like:

ebar=ededx.Integral(low,high)/dedx(low,high);

A.E. Snyder, Group EC                        \!c*p?/
SLAC Mail Stop #95                          ((.   .))
Box 4349                                        |
Stanford, Ca, USA, 94309                      '\|/`
e-mail:snyder_at_slac.stanford.edu                 o
phone:650-926-2701                              _
http://www.slac.stanford.edu/~snyder          BaBar
FAX:650-926-2657                          Collaboration

On Wed, 7 Feb 2007, Y. Roed wrote:

> Hi,
> I'm working on my diploma thesis evaluating nuclear physics data with root.
>
> When I fit a Landau function to my histogram with fixed upper and lower limits
> the fit parameters show the most probable value, the sigma, the chisquare, and
> the constant. Is there a possibility to get the mean value either printed in
> the parameter list or shown in the root command line?
>
> With respect to the question from 2002 (see below) is there a function that
> gives a relation between these two values when having fixed limits?
> Unfortunately, the question hasn't been discussed further on.
>
> Yours, Yvonne
>
> Hi jacek,
> Does it make sense? The mean value depends where you cut the tails
> Rene Brun
> On Thu, 28 Nov 2002, Jacek M. Holeczek wrote:
> > Hi,
> > Just a stupid question ...
> > Is there any easy formula that gives the relation between the Landau's
> > Most Probable Value and its Mean Value ?
> > Thanks in advance,
> > Jacek.
>
> *******************************************************
> Yvonne Roed
> Institut fuer Experimentelle und Angewandte Physik,
> University of Kiel
> Leibnizstrasse 11,
> D-24118 Kiel
> Germany
> phone: (+) 49 431 880 2543; Fax: (+) 49 431 880 3968
> e-mail: roed_at_physik.uni-kiel.de
> www: www.ieap.uni-kiel.de/et
>
>
Received on Wed Feb 07 2007 - 18:55:52 CET