Re: [boost] multivariate polynomials

Subject: Re: [boost] multivariate polynomials
From: Paul A. Bristow (pbristow_at_[hidden])
Date: 2009-09-26 04:55:51

> -----Original Message-----
> From: boost-bounces_at_[hidden] [mailto:boost-bounces_at_[hidden]] On
Behalf Of
> Jeffrey Hellrung
> Sent: Friday, September 25, 2009 9:39 PM
> To: boost_at_[hidden]
> Subject: [boost] multivariate polynomials
> 
> I noticed Boost.Math has some univariate polynomial-related functions,
> but I couldn't find anything regarding multivariate polynomials in
> either boost or the sandbox.  It seems like something ripe for generic
> programming (hence I think of boost first).  Is anyone currently working
> on something this, or can anyone provide a recommendation for an outside
> library?
> 
> Our particular application deals with real (or complex) multivariate
> polynomials, and being able to do arithmetic, differentiation,
> integration, composition, and (of course) evaluation (among, perhaps,
> other operations).

I'm sure John Maddock will reply with more authority (when his builders have
left ;-)

But I know of no plans for work on this at present.

We didn't even come to a satisfactory conclusion of 

Pawel Kieliszczyk's Polynomial library

after a review that wanted more changes than the author has found time to
implement for a re-review.

http://lists.boost.org/Archives/boost/2009/03/149833.php

So one could argue that we need to walk before running with multivariates ;-)

Boost.Math does not do anything complex either yet - these things always seem to
turn out unexpectedly complex ;-) :-(

But if you are planning to tackle this, I am sure we are willing to offer help
in fitting into the Boost.Math scaffold.
Templating on real type (rather than just double)  is very useful (to some) but
causes extra brain hurt.
Policies to handle exceptional cases are also invaluable but cause yet more
complexity.

Paul

---
Paul A. Bristow
Prizet Farmhouse
Kendal, UK   LA8 8AB
[phone], mobile [phone]
pbristow_at_[hidden]