[Bug-gsl] Bug in reproducing gsl_multilinear_wfit test case

[Jim Heasley]

Below is an email exchange I had with Rhys Ulerich regarding my 
inability to reproduce the
test case for the gsl_multiinear_wfit module. Rhys concluded thhat the 
problem was due to the
use of a different random number generator. As noted at the end of his 
email he suggested that I
ask that the example be documented as having used 
GSL_RNG_TYPE=mt19937_1999 or that a
new test be published for the new default generator.

Thank you.

Jim Heasley

-------- Original Message --------
Subject: 	Re: [Bug-gsl] Bug(?) in reproducing gsl_multilinear_wfit test 
case
Date:   Thu, 26 Apr 2012 09:12:24 -0500
From:   Rhys Ulerich <address@hidden>
To:     address@hidden
CC:     address@hidden



Hi Jim,

>  I have been unable to get the sample code that executes the
>  routine gsl_multilinear_wfit.c (which I literally cut and
>  pasted from the on-line HTML documentation) to give an answer
>  that is the same or even close to what is given in the
>  documentation.
>
>  ....
>
>  My compiled  code gives the result
>
>          # best fit: Y = 1.18246 + 0.184573 X + 1.3031 X^2

I also see the result you describe when copy'n'pasting from
http://www.gnu.org/software/gsl/manual/html_node/Fitting-Examples.html.
FWIW, I can reproduce the 'demo' program output shown earlier in that
section.

I think there has been a change to the default gsl_rng_env_setup
(http://www.gnu.org/software/gsl/manual/html_node/Random-number-environment-variables.html)
since when that example was written?  Going through rng/ChangeLog I
noticed

Thu Jul  9 13:56:20 1998  Brian Gough<address@hidden>
        
        * default.c (gsl_rng_env_setup): made gsl_rng_mt19937 the default
        generator

but it looks like the fitting example in question was generated in
2001 (http://bzr.savannah.gnu.org/lh/gsl/trunk/revision/2797).

A quick sweep through the available random number generators (try
setting an incorrect GSL_RNG_TYPE value before running ./generate to
get a list) shows that the example in the documentation was produced
using the 'mt19937_1999' generator and not the current default.

Try running 'GSL_RNG_TYPE=mt19937_1999 ./generate | ./fit 19' to see
if it reproduces what the manual documents.  It did for me.

If it does, please file a bug on savannah asking that either the
example be documented with this GSL_RNG_TYPE=mt19937_1999 detail or
the results be updated for the new default generator.

- Rhys