Re: sys2tf seems broken in 2.1.38

[Nicholas Piper]

On Fri, 01 Nov 2002, John W. Eaton wrote:

> What system are you using?  What lapack do you have?  If you are using
> some vendor-supplied version of lapack, it could be out of date
> compared to what Octave needs.  Octave assumes that you have the
> latest (or something like ATLAS that is based on the latest) version
> of lapack.  I think in this particular case, Octave is calling DGESVD
> once with LWORK = -1 and expecting it to return the proper length of
> the work array.  So it looks like your version of lapack doesn't
> handle that.

I didn't realise LAPACK wasn't bundled! I just happened to have a
/very/ old version already installed, back from when this machine was
a slink box.

I spent many hours today trying to build a new one (and I still don't
quite understand the real relationship between BLAS (reference
vs. vendor; I think I can't have the intel version without paying ?),
LAPACK and ATLAS (makes a good BLAS for your platform ? But seems to
use the installed BLAS library if you have one ?...). In the end, I
used "alien" to grab the libraries out of the modern debian packages
for BLAS and LAPACK.

And now step() works just fine!

Thank you for your help.

Is there a way Octave can warn users if they have an LAPACK too old ?
Or is this a case of my not having RTFM'd ?

 Nick

-- 
Part 3 MEng Cybernetics; Reading, UK       http://www.nickpiper.co.uk/
GPG Encrypted mail welcome!                             1024D/3ED8B27F
Choose life. Be Vegan :-) Please reduce needless cruelty + suffering !



-------------------------------------------------------------
Octave is freely available under the terms of the GNU GPL.

Octave's home on the web:  http://www.octave.org
How to fund new projects:  http://www.octave.org/funding.html
Subscription information:  http://www.octave.org/archive.html
-------------------------------------------------------------