lisp scientific computing libraries

[R. Matthew Emerson]

What's the state of scientific computing libraries in CL?

There appear to be numerous projects that address this particular area
at different levels:

* GSLL: http://common-lisp.net/project/gsll/
* Matlisp: http://matlisp.sourceforge.net/
* lisp-matrix: http://common-lisp.net/project/lisp-matrix/

and certainly more.

If anyone on the list works in this area, I'd be interested to hear
about what libraries you are using. If you'd care to share any pros and
cons, that would also be appreciated.

(As a bit of an aside, I will mention that one particular area of
interest is quadratic programming.

The package at http://sigpromu.org/quadprog/ package looked interesting,
but it doesn't come with source any more, and is meant to work with
Matlab.

If by some chance someone is familiar with this area, perhaps you could
offer some recommendations?

Pretty much anything with a C header file would be pretty easy to use.)

[A.J. Rossini]

My N-bits on the subject (speaking as an amateur maintainer and
occassional developer of lisp-matrix, and doing a bit of non-lisp
numerics/informatics/computation professionally):


I'd probably go with GSLL if you need to "get the job done" , it has a
range of numerical domains implemented (but requires GSL,
algorithms/implementations are reasonable but not always world-class).

Matlisp works, is stable, but needs a bit of
cleanup/refinement/refactoring, and requires Lapack and a few other
Fortran libraries, and uses it's own FFI approach for Fortran. I wish
it had a Fortran-front end for CFFI implemented. But "it works".

linear-algebra (on github, and see recent posts) looks like a clean
package, lisp-only, and so far I like it very much, but havn't had as
much experience with it as I have with the other 3 you mention, so
havn't had time to suss-out any issues. I don't recall if it has
quad-programming, if not you'd have to add on the quad-programming
algorithm, but that isn't too hard (i.e. following a journal article,
for an undefined "hard").

You could also go via R, through RCL (works well) or RCLG (which I
maintain), and the quad programming packages for that system. Highly
recommend R as a non-lisp system for scientific computing. It's only
fault is that it isn't common lisp.

Lisp-matrix is a work in progress. As much as I'm fond of it, I have
a hard time recommending it.

best,
-tony

Post by R. Matthew Emerson
What's the state of scientific computing libraries in CL?
There appear to be numerous projects that address this particular area
* GSLL: http://common-lisp.net/project/gsll/
* Matlisp: http://matlisp.sourceforge.net/
* lisp-matrix: http://common-lisp.net/project/lisp-matrix/
and certainly more.
If anyone on the list works in this area, I'd be interested to hear
about what libraries you are using.  If you'd care to share any pros and
cons, that would also be appreciated.
(As a bit of an aside, I will mention that one particular area of
interest is quadratic programming.
The package at http://sigpromu.org/quadprog/ package looked interesting,
but it doesn't come with source any more, and is meant to work with
Matlab.
If by some chance someone is familiar with this area, perhaps you could
offer some recommendations?
Pretty much anything with a C header file would be pretty easy to use.)
_______________________________________________
pro mailing list
http://common-lisp.net/cgi-bin/mailman/listinfo/pro

--
best,
-tony

***@gmail.com
Muttenz, Switzerland.
"Commit early,commit often, and commit in a repository from which we
can easily roll-back your mistakes" (AJR, 4Jan05).

Drink Coffee:  Do stupid things faster with more energy!