Changeset 58468

Timestamp:

Dec 20, 2009, 2:31:24 PM (13 years ago)

Author:

Thomas Klimpel

Message:

Continue of regression tests update
(OK, that one was really bad. Both hseqr and trevc lead to ambiguous overloads between the real and the complex version.

• For trevc, the complex version is missing a const before the matrix T, because LAPACK declared T as a input/output parameter in the complex case. The reason for this is that the diagonal of T is internally overwritten, but restored before the routine exists. The assert on "m" for trevc is in the wrong place, because "m" is not an input parameter -> it should be possible to fix this problem in the generator.
• For hseqr, the real version takes two vectors for the real and imaginary parts of the eigenvectors. This conflicts with the user-defined workspace version of the complex hseqr. Only way out seems to be two different names.)

(No files)