[Pkg-scicomp-devel] Bug#569436: libitpp: FTBFS: Test `svd_test' FAILED

[Kumar Appaiah]

reassign 569436 lapack 3.2.1-2
severity 569436 grave
thanks

On Thu, Feb 11, 2010 at 08:02:21PM +0100, Lucas Nussbaum wrote:
> During a rebuild of all packages in sid, your package failed to build on
> amd64.
> 
> Relevant part:
> > make[3]: Entering directory `/build/user-libitpp_4.0.7-1-amd64-Zz34fy/libitpp-4.0.7/tests'
[snip]
> > ------------------------------------------------------------------------------
> > Test `svd_test' FAILED!!!
> > ------------------------------------------------------------------------------

Lucas, thanks for the bug report. I investigated the cause of this
bug, and it appears that the SVD does, indeed, give a wrong
result. Since the SVD is done by Lapack's zgesvd function, I reassign
the bug report to Lapack, and set the severity to grave, since several
programs such as Octave also use this function (Indeed, I did verify
that Octave also produces incorrect results).

Lapack maintainers, here are a few methods by which you can reproduce
the error.

- I used an example program from Intel, which I don't attach
  here since it is copyrighted:

  http://software.intel.com/sites/products/documentation/hpc/mkl/lapack/mkl_lapack_examples/zgesvd_ex.f.htm

  The program lists the expected output. Indeed, the output is correct
  in a Lenny system, but it is wrong in my chroot, where the FTBFS
  happens. There, I get the output as:

 ( -0.86,  0.00) (  0.40,  0.00) (  0.32,  0.00)
 ( -0.35, -0.21) ( -0.24, -0.21) ( -0.63,  0.60)
 (  0.15,  0.61) (  0.61,  0.61) ( -0.36,  0.10)

 which differs from the output given within the code.

- I used Octave in the chroot, and compared it with a "working"
  Octave, for this:

A = [1 2 3; 4 5 6; 7 8 9] * 1j;
[U, S, V] = svd(A);
disp(U);

The correct U is:

  -0.000000 - 0.214837i  -0.000000 - 0.887231i  -0.402839 - 0.066237i
   0.000000 - 0.520587i   0.000000 - 0.249644i   0.805678 + 0.132474i
   0.000000 - 0.826338i   0.000000 + 0.387943i  -0.402839 - 0.066237i

But in the chroot, I get:

  -0.00000 - 0.88741i  -0.00000 - 0.88723i  -0.39566 - 0.10061i
   0.00000 - 0.25007i  -0.00000 - 0.24964i   0.79131 + 0.20122i
   0.00000 + 0.38726i   0.00000 + 0.38794i  -0.39566 - 0.10061i

Again, I don't think the net fault lies with lapack, since libitpp
built and passed all tests on the buildds with this very version of
lapack; I couldn't think of what changes caused this to happen,
though.

Please let me know what else I can do to help.

Thanks!

Kumar
-- 
If you shed tears when you miss the sun, you also miss the stars.
- Rabindranath Tagore (Stray Birds, 1916)
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 836 bytes
Desc: Digital signature
URL: <http://lists.alioth.debian.org/pipermail/pkg-scicomp-devel/attachments/20100211/752bd292/attachment.pgp>