Re: speeding up wrapper for "new" datatype
- From: AJ Rossini <blindglobe@xxxxxxxxx>
- Date: Sun, 12 Apr 2009 11:56:22 -0700 (PDT)
On Apr 12, 6:37 pm, "Thomas M. Hermann" <tmh.pub...@xxxxxxxxx> wrote:
On Apr 12, 9:44 am, Tamas K Papp <tkp...@xxxxxxxxx> wrote:
I was cleaning up code that interfaces to BLAS/LAPACK/UMFPack
libraries, and it occured to me that I could unify the interface for
various special matrices using CLOS. The idea is that I create a
matrix class, and specialize that as needed, including dense matrices
which are really just wrappers around 2d arrays.
I've already started a library that does this, but it isn't public.
I'd like to collaborate on this with you, please send me an email and
I will send you my latest documentation. I've sent a copy of the
document to Liam Healy, the maintainer of GSLL, to get his thoughts.
My basic idea is to define the interface, create a reference
implementation entirely in CL and then generate backends to the
I don't know what you've been using for unit testing, but Liam and I
have been collaborating on floating point extensions to lisp-unit.
They are publicly available at:
Looking forward to working with you on this.
I've got some similar code for lisp-matrix (on github) which starts to
generalize the framework there for symmetric matrices, and there are a
few other specialized types I'll be working on as I need them (have
the statistical modeling framework to perfect, and the types appear as
Have to admit that I'm using LIFT with some numerical equality
extensions, and am looking at wrapping the GSLL stuff as well (since
for me, it's one of the "big 4", i.e. Tamas' lisp-centric work, rif's
foriegn storage work, and matlisp-gui-alikes being the other 3 that
I'd like to unify).
Of course, I've got my own work, a lisp-centric R replacement with
roots in LispStat, which is driving my numeric needs, so the numerics
are coming slower than I'd like.
- Prev by Date: Re: New posting on Codeaholic: [newLISP] restart-router
- Next by Date: Re: New posting on Codeaholic: [newLISP] restart-router
- Previous by thread: Re: speeding up wrapper for "new" datatype
- Next by thread: Re: speeding up wrapper for "new" datatype