R SIG Debian - Dec 2014 - BLAS/LAPACK routine 'DLASCL' gave error code -4 in liblapack3 Version 3.5.0

Dear all

Some computations gave a matrix. The single value decomposition of it worked
fine previously, after an system upgrade I get the following error:
-----
> La.svd(x,256,256)
Error in La.svd(x, nu, nv) :
  BLAS/LAPACK routine 'DLASCL' gave error code -4
-----
Erroneous Matrix:
ftp://usys-ftp.ethz.ch/ITES/STEP/nussbaum/R/matrix-lapack-error4.RData /.csv
Operating System: Debian 8
Previous version:    liblapack3   3.4.1 (running without error)
New version:          liblapack3   3.5.0-4 (gives error above)
How do I track down this error code?
Is this a bug in liblapack3 or what did change?
Thank you!
M. Nussbaum


	[[alternative HTML version deleted]]

Hi Madlene,

On 16 December 2014 at 13:10, Nussbaum  Madlene wrote:
| 
| Dear all
| 
| Some computations gave a matrix. The single value decomposition of it worked
fine previously, after an system upgrade I get the following error:
| -----
| > La.svd(x,256,256)
| Error in La.svd(x, nu, nv) :
|   BLAS/LAPACK routine 'DLASCL' gave error code -4
| -----
| Erroneous Matrix:
ftp://usys-ftp.ethz.ch/ITES/STEP/nussbaum/R/matrix-lapack-error4.RData /.csv
| Operating System: Debian 8
| Previous version:    liblapack3   3.4.1 (running without error)
| New version:          liblapack3   3.5.0-4 (gives error above)
| How do I track down this error code?
| Is this a bug in liblapack3 or what did change?

Might be best to talk to the maintainer(s) of liblapack3, who may or may not
be conversant in R so a reproducible example in just C may help. 

I just checked https://packages.debian.org/search?keywords=lapack and 3.5.0-4
is the most recent version.

In the meantime you can of course also try switching to Atlas or OpenBLAS and
see how your matrix does there.

Dirk

| Thank you!
| M. Nussbaum
| 
| 
| 	[[alternative HTML version deleted]]
| 
| _______________________________________________
| R-SIG-Debian mailing list
| R-SIG-Debian at r-project.org
| https://stat.ethz.ch/mailman/listinfo/r-sig-debian

-- 
http://dirk.eddelbuettel.com | @eddelbuettel | edd at debian.org

Dear Madlene,

while this list serves as a forum for discussions about R backports present on 
CRAN, this would be something preferably filed in the Debian bug tracking 
system bts.debian.org - could you file a bug there? But please don't ask me
on
which package - maybe Dirk has a good answer to that one...

Cheers,

Johannes

P.S.: https://www.debian.org/Bugs/Reporting

Am Dienstag, 16. Dezember 2014, 13:10:15 schrieb Nussbaum 
Madlene:
> Dear all
> 
> Some computations gave a matrix. The single value decomposition of it
worked
> fine previously, after an system upgrade I get the following error: -----
> 
> > La.svd(x,256,256)
> 
> Error in La.svd(x, nu, nv) :
>   BLAS/LAPACK routine 'DLASCL' gave error code -4
> -----
> Erroneous Matrix:
> ftp://usys-ftp.ethz.ch/ITES/STEP/nussbaum/R/matrix-lapack-error4.RData
> /.csv Operating System: Debian 8
> Previous version:    liblapack3   3.4.1 (running without error)
> New version:          liblapack3   3.5.0-4 (gives error above)
> How do I track down this error code?
> Is this a bug in liblapack3 or what did change?
> Thank you!
> M. Nussbaum
> 
> 
> 	[[alternative HTML version deleted]]
> 
> _______________________________________________
> R-SIG-Debian mailing list
> R-SIG-Debian at r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-debian
-- 
PD Dr. Johannes Ranke
Kronacher Str. 8
79639 Grenzach-Wyhlen

On 16 December 2014 at 14:45, Johannes Ranke wrote:
| this list serves as a forum for discussions about R backports present on CRAN,

The mandate, as I recall, is a little broader: "usage of R on Debian-based
systems"

As such, I found the post perfectly adequate. We redirect from r-help and
other lists to here because it is the one focus point for all things related
to Debian / Ubuntu / ... and R.

Dirk

-- 
http://dirk.eddelbuettel.com | @eddelbuettel | edd at debian.org