From: "Tim Davis" <davis@cise.ufl.edu>
Path: news.mathworks.com!newsfeed-00.mathworks.com!webcrossing
Newsgroups: comp.soft-sys.matlab
Subject: Re: Wishlist for R2007b
Message-ID: <ef57195.13@webcrossing.raydaftYaTP>
Date: Thu, 17 May 2007 15:27:00 -0400
References: <MPG.20b52ece5ac3d2e798973b@news.mathworks.com> <1179378005.215820.316490@u30g2000hsc.googlegroups.com> <ef57195.8@webcrossing.raydaftYaTP> <ef57195.10@webcrossing.raydaftYaTP> <ef57195.11@webcrossing.raydaftYaTP>
Lines: 19
NNTP-Posting-Host: 70.171.50.145
MIME-Version: 1.0
Content-Type: text/plain; charset="ISO-8859-1"
Content-Transfer-Encoding: 8bit
Xref: news.mathworks.com comp.soft-sys.matlab:409687



Trevor wrote:
>
>
> 1. Sparse matrices of dimension greater than 2.

What do you want to do with them?

The sparse data structure for the 2D case is optimized for linear
algebra, not as a "data bucket". Thus, A(i,j) = whatever and
whatever=A(i,j) are not terribly fast ... Although they could be
better they will never be truly speedy.

ND matrices come up in tensor calculations, I think.

What the ND sparse case should look like depends strongly on what you
want to do with the beasts. If you want just sparse sub referencing
and sub assignment, that's one thing. If you want tensors, that's
another. (I'm not a tensor specialist, except in the morning when
I've had one two many tenser products via my morning coffee... ;-)