[Octave-bug-tracker] [bug #63384] Error for "inv(NaN)" with LAPACK 3.11.0

[Markus Mützel]

Follow-up Comment #36, bug #63384 (project octave):

Fwiw, in Matlab R2022a:

>> x = nan (2,2)

x =

   NaN   NaN
   NaN   NaN

>> [y,rc] = inv (x)
Error using inv
Too many output arguments.
 
>> y = inv (x)
Warning: Matrix is singular, close to singular or badly scaled. Results may be
inaccurate. RCOND = NaN. 
 

y =

   NaN   NaN
   NaN   NaN

>> inv (single (x))
Warning: Matrix is singular, close to singular or badly scaled. Results may be
inaccurate. RCOND = NaN. 
 

ans =

  2×2 single matrix

   NaN   NaN
   NaN   NaN

>> inv (complex (single (x), ones (2,2)))
Warning: Matrix is singular, close to singular or badly scaled. Results may be
inaccurate. RCOND = NaN. 
 

ans =

  2×2 single matrix

      NaN +    NaNi      NaN +    NaNi
      NaN +    NaNi      NaN +    NaNi

>> inv (complex (x, ones (2,2)))
Warning: Matrix is singular, close to singular or badly scaled. Results may be
inaccurate. RCOND = NaN. 
 

ans =

      NaN +    NaNi      NaN +    NaNi
      NaN +    NaNi      NaN +    NaNi

>> version -lapack

ans =

    'Intel(R) oneAPI Math Kernel Library Version 2021.3-Product Build 20210611
for Intel(R) 64 architecture applications (CNR branch auto) supporting Linear
Algebra PACKage (LAPACK 3.9.0)'

>> version -blas

ans =

    'Intel(R) oneAPI Math Kernel Library Version 2021.3-Product Build 20210611
for Intel(R) 64 architecture applications (CNR branch auto)'



It looks like they don't return the inverse condition number. But since they
do not return anything else either, this is probably an extension we can keep
without worrying too much about compatibility.
Other than that, their results make more sense to me...



    _______________________________________________________

Reply to this item at:

  <https://savannah.gnu.org/bugs/?63384>

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