#11248: SVD matrix decomposition may return a non-invertible "unitary" matrix
------------------------------+---------------------------------------------
Reporter: rbeezer | Owner: jason, was
Type: defect | Status: new
Priority: major | Milestone: sage-4.7
Component: linear algebra | Keywords:
Work_issues: | Upstream: N/A
Reviewer: | Author:
Merged: | Dependencies:
------------------------------+---------------------------------------------
Comment(by fbissey):
Replying to [comment:16 rbeezer]:
> Replying to [comment:14 slabbe]:
> > So, does this means it is beyond Sage? I mean, should a fix be made in
Sage or should it be done upstream?
>
> Its beyond me, as in - its beyond my knowledge.
>
> I do not think the true source of the error has been located. It is not
Sage library code. It appears in numpy, which wraps the LAPACK function
ZGESDD. But I do not understand the relationships between LAPACK, ATLAS
and BLAS, and especially on the Mac. Palmieri says ATLAS on the mac is
"system supplied." It still could be the Sage build that does something
wrong.
>
Let me try to shed some light. There are 3 libraries: BLAS also sometimes
f77BLAS,
CBLAS and LAPACK. BLAS and LAPACK are standard libraries which means they
have a defined interface and behavior. Anyone can create an implementation
of Blas or Lapack, it has to conform to a standard.
f77blas is a fortran implementation cblas a C implementation. Blas is low
level.
Lapack is a higher level library and is fortran based. It uses f77blas
functions.
Atlas is a particular implementation of cblas, it also provides a f77blas
implementation. It is highly optimized by autotunning (if all goes right).
Furthermore if you provide the netlib reference lapack implementation it
can produce an optimized lapack library based on some of its optimized
blas algorithm.
[http://math-
atlas.sourceforge.net/devel/atlas_devel/atlas_devel.html#SECTION00060000000000000000]
Of course lapack also has to link to f77blas, if you have a subtle problem
there it will appear in lapack.
--
Ticket URL: <http://trac.sagemath.org/sage_trac/ticket/11248#comment:19>
Sage <http://www.sagemath.org>
Sage: Creating a Viable Open Source Alternative to Magma, Maple, Mathematica,
and MATLAB
--
You received this message because you are subscribed to the Google Groups
"sage-trac" group.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to
[email protected].
For more options, visit this group at
http://groups.google.com/group/sage-trac?hl=en.
