[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: limitations of octave 2.9.9 without suitesparse?

From: David Bateman
Subject: Re: limitations of octave 2.9.9 without suitesparse?
Date: Thu, 23 Nov 2006 23:45:32 +0100
User-agent: Thunderbird (X11/20060921)

Norbert Hauser wrote:
> Hello,
> I wonder which functions of sparse matrices are not supported without 
> suitesparse.

The major one is the \ operator

CHOLMOD is needed for positive definite matrix
UMFPACK for non-symmetric, and
CXSPARSE for singular or under-, over-determined problems

These also rely on CAMD, COLAMD and CCOLAMD (for cholmod). You can still
treat upper, lower or banded matrices however. Other things you'll miss
are the factorizations chol (cholmod), lu (umfpack) and qr (cxsparse).
You'll also miss all of the sparse reordering functions, symamd (amd),
colamd (colamd), dmperm (cxsparse). I think the only other things that
might be effected are the inv and det functions. Apart from that, most
of the other stuff will work...

> Since I have not been able to compile octave on top of suitesparse, I tried 
> what happens without suitesparse. All sparse functions, that I tried so far, 
> seem to work without problems. Do I have to be careful about the result of 
> certain functions?

There is a bug in 2.9.9, that is fixed in the CVS, where if you try to
use one the above missing features, and then the matrix_type function is
called on the resulting matrix (in C++ only), then a seg-fault occurs.
You can induce this with missing sparse dependencies by typing "make
check" in the top-level of your 2.9.9 build tree.


> Norbert
> _____________________________________________________________________
> Der WEB.DE SmartSurfer hilft bis zu 70% Ihrer Onlinekosten zu sparen!

reply via email to

[Prev in Thread] Current Thread [Next in Thread]