This is the mail archive of the
gsl-discuss@sources.redhat.com
mailing list for the GSL project.
Re: Landau distribution
- To: GSL mailing list <gsl-discuss at sourceware dot cygnus dot com>
- Subject: Re: Landau distribution
- From: Gerard Jungman <jungman at lanl dot gov>
- Date: Thu, 22 Mar 2001 13:38:28 -0700
- Organization: LANL T-3
- References: <3AB0E229.6527EBE0@bnl.gov> <15025.9105.62915.159472@debian>
Brian Gough wrote:
>
> If the approximation is the only game in town then it's better to use
> that rather than not have a useful function. The limitations should
> just be noted in the documentation.
I don't think I agree with that. Ideally there should be
some parameter which controls the accuracy. Then users can
decide for themselves what tradeoff to make.
> For special functions an error estimate is part of the computed answer
> so the function can accommodate a region where the approximation is
> less good.
Up to a point. People still expect some uniformity
of behaviour. You don't want the thing to just drop out,
especially if it is going to drop below standard
single-precision accuracy. People are probably not
expecting that when they are writing application
code, so mistakes could be made.
Anyway, I would be curious to see the range of choices
in the literature for this Landau thing. I assume Schorr's
method is not the only game in town. Can somebody tell
me where to look for this stuff?
Thanks.
--
G. Jungman