IMSL, NAG Fortran Library vs. Open Source libraries

[Bart Vandewoestyne]

Hello,

I recently got asked the question again whether the NAG Fortran
Library was available on our server. In the past, I already got
the same question about the IMSL library...

There is one problem for me to answer that question: I like Open
Source Software...

So I would like to advise the persons asking me those questions
with an 'Open Source answer'.

To be honest... i always stick to the Fortran 95 standard when
programming, and up until now i could implement all the
'numerical code' that i needed myself, without the use of
external libraries but for example using Numerical Recipes or
other descriptions of algorithms in papers.

So... I have no experience with for example the GNU Scientific
Library in combination with Fortran 95 (or 2003). I know there
exists a Fortran interface [1], but i have not tested it.

So my questions to this group are:

* Next to GSL, Netlib, GAMS and Numerical Recipes, what other
Open Source libraries are there that can easily be used
from a Fortran 95 program and are a valuable alternative to the
NAG Fortran Library or the IMSL?

* Does anyone have experience with FGSL and would you advise its
use?

Feel free to share your ideas/comments. I want Open Source to
rule at my institution! ;-)

Kind regards,
Bart

[1]
http://www.lrz-muenchen.de/services/software/mathematik/gsl/fortran/index.html

--
"Share what you know. Learn what you don't."

[none]

Numerical Recipes is *not* open source. If you want to pursue open source
offerings then you must drop Numerical Recipes. You can regard Numerical
Recipes as being an alternative to NAg and IMSl, with the source code made
available, and where the code has little protection against simple
algorithmic problems (infinite loops, for example) compared to NAg & IMSL.

Netlib has, I believe, all that you would need from NAg - or perhaps, all
that I have needed from NAg. GAMS has helped a little with optimisation,
and Alan Miller's Fortran 90 translations have had the benefit of updating
some Netlib & Statlib material. Finally, Linda Petzold's website offers
updated versions of her DAE solver.

[Salvatore]

If you need to do sparse linear algebra, I can suggest my own
stuff :-)

www.mld2p4.it
www.ce.uniroma2.it/psblas

Regards
Salvatore

[anal...@hotmail.com]

On Sep 17, 12:16 pm, Bart Vandewoestyne

> [1]http://www.lrz-muenchen.de/services/software/mathematik/gsl/fortran/i...

>
> --
>         "Share what you know.  Learn what you don't."

There is COIN-OR for constrained optimization. It was sponsored by
IBM but now they have bought a commercial vendor (ILOG) - and so its
not clear what level of IBM support one can expect going forward.

http://www.coin-or.org/

[Gus Gassmann]

COIN-OR was spun out of IBM several years ago and now exists as an
independent entity. Development on key COIN-OR projects is continuing
daily. On the other hand, almost all the development is done in C++.

[anal...@hotmail.com]

On Sep 19, 9:56 am, Gus Gassmann <horand.gassm...@gmail.com> wrote:

> daily. On the other hand, almost all the development is done in C++.- Hide quoted text -
>
> - Show quoted text -

Thats good to know. Even if they don;t have fortran callable
libraries, one can always build problems in a standard format with
Fortran and use their routines as a black box to get the solution and
process the output with the language of one's choice.