csytrf_rook test fails for dimensions 127, 260

[zotkevich]

Hi,

For new *_rook functionality I see csr test fails with extended set of input dimensions. Dimensions are 127 and 260. Output is following:

UPLO = 'U', N = 137, NB = 8, type 11, test 4, ratio = 48.634
UPLO = 'U', N = 260, NB = 8, type 2, test 4, ratio = 55.406
CSR: 2 out of 744 tests failed to pass the threshold

Error occurs in pivot condition number checking (TEST 4 in cchksy_rook.f).

As far as I understand condition number of a pivot should be bounded by constant value and can not exceed theoretical limit by design of algorithm.

It looks like something wrong with CSYTRF_ROOK function.

Folks, could you please comment?

[admin]

Dear Alex,
Thank you for your bug report.
Just to confirm that we have recorded the issue.
The bug number is 123 - http://www.netlib.org/lapack/Errata/index2.html
We sent your Bug report to Berkeley.
We hope to fix the issue very soon.

[admin]

Dear Alex,
the bug has been fixed in our repository. The bug fix will be included in the coming release of LAPACK.

Revision Log

Fixed BUG number 123, that is testing of 'rook' routines, specifically TEST 4 for UPLO='U' case that test bounds of block diagonal matrices. TEST 4 wasfailing for UPLO='U', since 2-by-2 was passed incorrectly to the -EVX routine. Also, during the fix the TEST 4 for both UPLO='U' and 'L' was modified. Instead of computing the ratio of eigenvalues for 2-by-2 matrices, using the -EVX routines, now we compute the ratio of singular values using *GESVD routines. Affected test routines: TESTING/LIN/schksy_rook.f TESTING/LIN/dchksy_rook.f TESTING/LIN/cchksy_rook.f TESTING/LIN/zchksy_rook.f TESTING/LIN/cchkhe_rook.f TESTING/LIN/zchkhe_rook.f