This is the mail archive of the
gsl-discuss@sourceware.org
mailing list for the GSL project.
Re: Orthogonal polynomials in gsl/poly
- From: Brian Gough <bjg at network-theory dot co dot uk>
- To: Richard Mathar <mathar at strw dot leidenuniv dot nl>
- Cc: gsl-discuss at sourceware dot org
- Date: Mon, 11 Sep 2006 22:53:11 +0100
- Subject: Re: Orthogonal polynomials in gsl/poly
- References: <20060909160258.GA20563@strw.leidenuniv.nl>
- Reply-to: ":gsl-discuss"@sources.redhat.com
At Sat, 9 Sep 2006 18:02:58 +0200,
Richard Mathar wrote:
> It would be useful, given the general importance of orthogonal polynomials
> in applications, to have some shortcuts to evaluate the most important
> ones in "by name" in the subdirectory poly, instead of defining them by
> a list of values that are interpolated.
> A proposal/implementation for the Hermite, Laguerre and Chebyshev polynomials
> is attached. I can implement more of these if this finds support.
Thanks for the email. I agree it would be useful to access the
different polynomials by name, there are a few in specfunc (based on
explicit representations for small n and recurrences for general n I
think) but more would be useful. A first question would be regarding
the relative merits of different ways of computing them, e.g stability
& speed, as I haven't looked at this problem so I have no idea what
the best way is.
--
Brian Gough