Hi Mat,
With the L option you use the Poisson distribution to calculate the
likelihhod function, default is the Gaussian distribution which
results in the minimization of a chi-square.
The Poisson Likelihood function has not a problem with a bin
contents of zero.
see for instance "data analysis" by Siegmund Brandt , page 196, example 7.4
The comment "L" Use Loglikelihood method (default is chisquare method)
in the code is confussing because we always use a Loglikelihood method
> X-Authentication-Warning: oxpc01.fnal.gov: msmartin owned process doing -bs
> Date: Mon, 21 May 2001 11:25:06 -0500 (CDT)
> From: Matthew Martin <msmartin@fnal.gov>
> To: roottalk@pcroot.cern.ch
> cc: Todd Huffman <t.huffman1@physics.ox.ac.uk>
> Subject: [ROOT] chisquare
> MIME-Version: 1.0
> X-Filter-Version: 1.3 (wren)
>
> Hi Rene,
> When I do something like:
>
> hist_bp_PtResidWeight->Fit("gaus","L");
>
> how is the resulting chisquare defined? Ie how does it deal with the
> possibility of some bins with zero events in them?
>
> cheers
>
> Mat
>
Eddy A.J.M. Offermann
Renaissance Technologies Corp.
Route 25A, East Setauket NY 11733
e-mail: eddy@rentec.com
http://www.rentec.com
This archive was generated by hypermail 2b29 : Tue Jan 01 2002 - 17:50:46 MET