[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Axiom-math] Numeric ODE and DAE solvers

From: root
Subject: Re: [Axiom-math] Numeric ODE and DAE solvers
Date: Tue, 13 Dec 2005 11:02:20 -0500


Axiom still has the hooks for the NAG software but, as you pointed
out, you'd have to buy the NAG software to use them.

There is an effort underway to create a replacement library using
open source versions but it is still in the early stages (BLAS).
Virtually all of the numeric code is in FORTRAN. Replacing the
libraries will not be easy and will take time. NAG has a high
quality library with a lot of invested effort and it is unlikely
to be equaled anytime soon. 

> So here are my questions:
> - Is the best solution to interface Axiom directly with these Fortran 
> libraries?
> - Would modifying the NAG library link be a good way to proceed?

Well that's the current approach. There is code in Axiom to enable the
lisp-fortran interface. The process is to start with the BLAS routines
and work upward in functionality. This is probably a year's worth of work.
There was a minor discussion of this effort just last month. Check the
mailing list archive.

Of course, given Axiom's literate slant we need to do more than just
bolt the code to the system. There needs to be an explanation of the
theory behind the code. People have written good, clear explanations
of the theory and I've been asking permission to quote their work as
part of the documentation.

> - If so, are there any comments as to which library is "the best" as far 
> as functionality, efficiency, and accuracy?

That's too many paramaters to optimize. NAG, IBM's ESSL, and a few others
are probably best along various scales but are not free.

As an aside, robotics is dear to my heart as I'm a robotics person from
way back (I used to work for Unimation and then IBM Research doing
robotics research) so it would be nice to see Axiom used there.


reply via email to

[Prev in Thread] Current Thread [Next in Thread]