[R] lapack routine dgesdd, error code 1

[Prof Brian Ripley]

You have to read the LAPACK code.  It says

*  INFO    (output) INTEGER
*          = 0:  successful exit.
*          < 0:  if INFO = -i, the i-th argument had an illegal value.
*          > 0:  DBDSDC did not converge, updating process failed.

and 1 > 0 so this was a convergence failure inside the SVD code.

Such things are almost impossible to reproduce elsewhere as they depend on
the BLAS and the optimizations done by the compiler.

On Tue, 24 Feb 2004, karl cottenie wrote:

> Hello R-users,
> 
> during one of my analyses that involve a SVD, I get the following error
> message:
> 
> Error in La.svd(x, nu, nv, method) : error code 1 from Lapack routine
> dgesdd
> 
> With a search on the R web site, I only found references to error codes
> 17 and 3 for this particular routine. I also found the Lapack web site,
> but could not find a list of the possible error messages. If somebody
> knows what this error message means, or where I can find more
> information, that would be great.
> 
> thanks in advance,
> 
> Karl
> 

-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595