octave-bug-tracker
[Top][All Lists]
Advanced

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

[Octave-bug-tracker] [bug #37336] Incorrect result of rcond with Netlib


From: Sébastien Villemot
Subject: [Octave-bug-tracker] [bug #37336] Incorrect result of rcond with Netlib LAPACK
Date: Thu, 13 Sep 2012 08:12:27 +0000
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:10.0.7) Gecko/20100101 Firefox/10.0.7 Iceweasel/10.0.7

Follow-up Comment #2, bug #37336 (project octave):

Sorry for the missing pieces of information.

By Netlib LAPACK I mean the reference implementation of LAPACK, distributed on
www.netlib.org.

By Atlas LAPACK I mean the version created by Atlas: many of its routines
directly come from Netlib LAPACK, but some are replaced by optimized
versions.

For Netlib LAPACK, I use version 3.4.1 (from Debian wheezy/sid amd64, package
liblapack-dev/liblapack3 version 3.4.1-6).

For Atlas LAPACK I use version 3.4.8 (from Debian wheezy/sid amd64, package
libatlas3-base version 3.8.4-9).

Note that in Debian we recently switched the default LAPACK flavour: the
default used to be Atlas LAPACK, but we recently switched (in atlas 3.8.4-8)
to Netlib LAPACK (for reasons which are off-topic here). If you are using
Debian wheezy/sid (as I think you do), you can check your current
LAPACK flavour with:

 sudo update-alternatives --display liblapack.so.3

and change it with:

 sudo update-alternatives --config liblapack.so.3

Finally, please find attached the C program which directly does the
calculation using LAPACK. You need to link it against libmatio (available in
package libmatio-dev), with something like:

 gcc rcond-test.c -o rcond-test -lmatio -llapack

The program expects burp.mat in the current directory.


(file #26545)
    _______________________________________________________

Additional Item Attachment:

File name: rcond-test.c                   Size:1 KB


    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/bugs/?37336>

_______________________________________________
  Message posté via/par Savannah
  http://savannah.gnu.org/




reply via email to

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