[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Design of N-D arrays
From: |
Paul Kienzle |
Subject: |
Re: Design of N-D arrays |
Date: |
Wed, 06 Aug 2003 20:05:25 -0400 |
User-agent: |
Mozilla/5.0 (Windows; U; Win 9x 4.90; en-US; rv:1.3) Gecko/20030312 |
John W. Eaton wrote:
On 6-Aug-2003, Petter Risholm <address@hidden> wrote:
I'm also wondering whether it really makes sense to have separate 2-d
and N-d objects. Maybe they should all be N-d. Then we would have fewer types
of objects to worrry about.
Wouldn't it be nice if structures and types
could be added independently? Right now
we have the following structures:
scalar, vector, matrix, N-d
on the following types:
character, double, complex, octave_values
and also map.
Octave-forge adds a sparse matrix structure,
and galois field, symbolic and variable
precision types. It might be nice some day
to have banded matrices and unicode strings.
Each new type should implement every structure.
Each new structure should implement every type.
They don't because it is too much work. For
one thing, we are making very heavy use of
lapack, etc., which operate on doubles only.
Wouldn't it be cute if we implemented generic
forms for all our functions? Then we would
automatically be able to do things like matrix
multiplication at a given precision.
Okay, I'll stop day dreaming now.
Paul Kienzle
address@hidden