Guile bindings for GSL and LAPACK

Guile bindings for GSL and LAPACK

[Arno Peters]

This is to announce the Guile bindings for GSL and (to a much lesser
extent) LAPACK.  The bindings includes support for the following:

* complex numbers
* vector and matrix functions
* GSL BLAS
* linear algebra
* random number generators
* statistics
* fast fourier transforms
* random distributions
* special functions
* monte carlo integration
* quasi random sequences
* permutations, combinations and sorting
* function integration
* function differentiation
* function minimization (single dimension only)
* polynomial root finding (single dimension only)
* chebyshev approximations
* function fitting
* dgesv, zgesv, dgels, zgels functions from LAPACK

The test scripts have been taken from the examples in the GSL
reference manual.  They both show that the bindings work and serve as
examples of how to properly call the functions within Guile.  The
bindings have been tested against Guile 1.4 and GSL 1.2; they were
produced by using SWIG 1.3.

It is available for download here:
   http://httpd.chello.nl/~a.w.peters/software/guile-numerics-0.1.0.tar.gz



The GNU Scientific Library (GSL) is a numerical library written in
ANSI C.
   http://sources.redhat.com/gsl/

Guile is an implementation of the Scheme programming language.
   http://www.gnu.org/software/guile/

SWIG is a software development tool that connects programs written in
C and C++ with a variety of high-level programming languages.
   http://swig.sf.net/

Re: Guile bindings for GSL and LAPACK

[Wartan Hachaturow]

On Fri, Sep 06, 2002 at 03:30:12PM +0200, Arno Peters wrote:

> bindings have been tested against Guile 1.4 and GSL 1.2; they were
> produced by using SWIG 1.3.

Hm. BTW, why can't we produce Python bindings using swig? If I
understand right, pygsl is being written by hands?

-- 
Regards, Wartan.
"Computers are not intelligent. They only think they are."

Re: Guile bindings for GSL and LAPACK

[Arno Peters]

[-- Attachment #1: Type: text/plain, Size: 800 bytes --]

On Fri, Sep 06, 2002 at 05:59:57PM +0400, Wartan Hachaturow wrote:
> On Fri, Sep 06, 2002 at 03:30:12PM +0200, Arno Peters wrote:
> 
> > bindings have been tested against Guile 1.4 and GSL 1.2; they were
> > produced by using SWIG 1.3.
> 
> Hm. BTW, why can't we produce Python bindings using swig? If I
> understand right, pygsl is being written by hands?

I'm sure you could.  My focus has been, and still is, producing
bindings for Guile.  This means I have written some functions to ease
the integration between GSL and Guile.  These functions would have to
be rewritten to the Python interface.  However, anybody is free to use
my work as a basis to produce bindings for Python.  Provided of course
they adhere to the boundaries set forth in the GNU GPL.

Greetings,
-- 
Arno

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: Guile bindings for GSL and LAPACK

[Wartan Hachaturow]

On Fri, Sep 06, 2002 at 04:53:25PM +0200, Arno Peters wrote:

> > Hm. BTW, why can't we produce Python bindings using swig? If I
> > understand right, pygsl is being written by hands?
> 
...
> be rewritten to the Python interface.  However, anybody is free to use
> my work as a basis to produce bindings for Python.  Provided of course
> they adhere to the boundaries set forth in the GNU GPL.

Actually, I've never had any experience with swig. I've just noticed that
your bindings are fairly complete, while pygsl's are not. So, I made
an assumption that it's easier to use swig to produce bindings.
Perhaps, pygsl people might have some arguments against swig..
If they haven't, I might have a look at this process -- since I'm pretty
interested in python/gsl integration (but not ready to hand-write all
of the wrappers :).

-- 
Regards, Wartan.
"Computers are not intelligent. They only think they are."

Re: Guile bindings for GSL and LAPACK

[Achim Gädke]

Wartan Hachaturow wrote:

 >Actually, I've never had any experience with swig. I've just noticed that
 >your bindings are fairly complete, while pygsl's are not. So, I made
 >an assumption that it's easier to use swig to produce bindings.
 >Perhaps, pygsl people might have some arguments against swig..
 >If they haven't, I might have a look at this process -- since I'm pretty
 >interested in python/gsl integration (but not ready to hand-write all
 >of the wrappers :).
 >
One of the features, I can not achieve with swig, is the use of classes,
subtyping and so on... At least, it was not possible two years before,
when I started pygsl. The aim of pygsl is to provide modularised and
objectoriented access to the gsl features.

Yes, writing an interface by hand is not very nice, but usefull for
those, who use it. If you have success with swig, just mail to
mailto:pygsl-discuss@lists.sourceforge.net .

Yours, Achim