[Octave-bug-tracker] [bug #57475] svds warns about ill-condition input with NaNs rather than issuing error

[Rik]

Update of bug #57475 (project octave):

                  Status:                    None => Confirmed              
                 Release:                   5.1.0 => dev                    
                 Summary: svds returns unusual error => svds warns about
ill-condition input with NaNs rather than issuing error

    _______________________________________________________

Follow-up Comment #1:

For reference, the behavior on the development version of Octave which is
about to become version 6.1 is a little less scary:


 x = svds(nan*ones(3),2)
warning: svds: returning fewer singular values than requested
warning: called from
    svds at line 230 column 7

x = [](0x1)


Still, it would be better if the warning message was more precise about the
problem (NaN).  For reference, Matlab issues the same error for either of your
examples (badly-conditioned matrix with Info or NaN present).




    _______________________________________________________

Reply to this item at:

  <https://savannah.gnu.org/bugs/?57475>

_______________________________________________
  Message sent via Savannah
  https://savannah.gnu.org/