Path: csiph.com!news.mixmin.net!newsfeed.fsmpi.rwth-aachen.de!newsfeed.straub-nv.de!news-1.dfn.de!news.dfn.de!news.rz.uni-kiel.de!not-for-mail
From: boettcher@physik.uni-kiel.de (Stephan =?utf-8?Q?B=C3=B6ttcher?=)
Newsgroups: comp.graphics.apps.gnuplot
Subject: Re: Gaussian fit - Singular matrix in Invert_RtR
Date: Thu, 28 Apr 2016 07:30:28 +0200
Organization: [posted via] Rechenzentrum der Universitaet Kiel, Germany
Lines: 17
Message-ID: <s6n7ffinvuz.fsf@falbala.ieap.uni-kiel.de>
References: <f2adc325-2828-4a1a-9bee-47e5afb8691f@googlegroups.com>
NNTP-Posting-Host: falbala.ieap.uni-kiel.de
Mime-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 8bit
X-Trace: inlux3.rz.uni-kiel.de 1461821428 16789 134.245.71.168 (28 Apr 2016 05:30:28 GMT)
X-Complaints-To: news@news.rz.uni-kiel.de
NNTP-Posting-Date: Thu, 28 Apr 2016 05:30:28 +0000 (UTC)
User-Agent: Gnus/5.13 (Gnus v5.13) Emacs/24.5 (gnu/linux)
Cancel-Lock: sha1:IqzOs3odR1mOvCQrmulLACSs1qQ=
Xref: csiph.com comp.graphics.apps.gnuplot:3316

Andrew Finlay <andrewlfinlay@gmail.com> writes:

> Contents of my data file (please note, the error values are the square
> root of the counts, but when this would be 0, I chose a value of
> 0.0001 to try to avoid issues):

0.0001 is too low, you weigh those bins to much. I usually do

 using 1:2:(sqrt($2+1))

Zero is quite probably for bins with expectation value 1

Stephan

-- 
Stephan Böttcher Tel: +49-431-880-2508
Extraterrestrische Physik, IEAP, Leibnizstr. 11, 24118 Kiel