Re: [eigen] Re: Student contribution

[ Thread Index | Date Index | More Archives ]

On Sat, Jun 15, 2013 at 7:26 PM, BRUN Gauthier <brunga@xxxxxxxxxxxxxx> wrote:
> Hi,
> We have found something that could explain the bug with complexes.
> When we do
> MatrixType result(upperbid.householderU()*
> upperbid.bidiagonal().toDenseMatrix() * upperbid.householderV().adjoint());
> MatrixType result2(upperbid.householderV()*
> upperbid.bidiagonal().toDenseMatrix().adjoint() *
> upperbid.householderU().adjoint())
> We would expect result.adjoint() == result2 but we don't. When we print them
> we have different results. We do not know how to solve it.

This is an embarrassing report . The fix is in the attached file. I'll
commit it as soon as my computer has compiled and run all the unit

Thanks a lot for this finding!


Attachment: fix_householdersequence_conjugate.diff
Description: Binary data

Mail converted by MHonArc 2.6.19+