Hello Vasile,
Thanks for the LDT bug report. Sounds good to me. I am ready to change
this right away. I'll wait one or two days to check if some lapackers
scream .
Regarding the nonconvergence of LAPACK-3.2 DGEHQZ, this is a known issue.
(Known partly thanks to you.) My understanding of the history of the
problem is summarized:
http://www-math.ucdenver.edu/~langou/lapack-3.2/lapack_known_issues.html
We have delayed David Day's fix for the reason that a new QZ algorithm
with multishift and early aggressive deflation based on
Bo Kagstrom and Daniel Kressner. Multishift variants of the QZ algorithm
with aggressive early deflation. SIAM J. Matrix Anal. Appl.,
29(1):199-227, 2006.
is under development, so we felt that we should wait. But indeed, maybe
you are right, I'll try David Day's patch in the TESTING and maybe we will
release it. Thanks for giving us feedback.
Best wishes,
Julien.
|