Path: news.mathworks.com!not-for-mail
From: <HIDDEN>
Newsgroups: comp.soft-sys.matlab
Subject: Re: transpose of interpn() operations
Date: Mon, 4 Aug 2008 08:49:02 +0000 (UTC)
Organization: Xoran Technologies
Lines: 29
Message-ID: <g76fpu$7lt$1@fred.mathworks.com>
References: <g6vt59$k9o$1@fred.mathworks.com> <g70566$jh0$1@fred.mathworks.com>
Reply-To: <HIDDEN>
NNTP-Posting-Host: webapp-03-blr.mathworks.com
Content-Type: text/plain; charset="ISO-8859-1"
Content-Transfer-Encoding: 8bit
X-Trace: fred.mathworks.com 1217839742 7869 172.30.248.38 (4 Aug 2008 08:49:02 GMT)
X-Complaints-To: news@mathworks.com
NNTP-Posting-Date: Mon, 4 Aug 2008 08:49:02 +0000 (UTC)
X-Newsreader: MATLAB Central Newsreader 1440443
Xref: news.mathworks.com comp.soft-sys.matlab:483453




> Why not? Learn what a sparse matrix is,
> and you can write that matrix directly,
> then performing any such operations
> with ease.

Believe me, I'm no stranger to sparse matrices, but it 
won't do the trick. 

For the applications I work with, a typical size for V 
might be 400x400x300.

Even if I stuck with tensorial linear interpolation, and 
even if sparse matrices supported single precision floats, 
the size of A would easily run to about 2GB and strain 
MATLAB's memory limits. On top of that, I would have to 
convert V to type double to perform the multiplication.

Aside from memory issues, the time required to generate A 
is problematic. In typical applications, you may need to 
repeat these operations with several sets of interpolation 
coordinate data. This means you will have to regenerate A 
multiple times. 

Anyway, if I'm reaching for the moon, so be it. I'm just 
surprised that there hasn't been more interest in a coded 
routine for A.'. It seems to me like something that would 
often be needed.