[Lapack] dsyev and dsyevd

[Bedrich Sousedik]

Dear Gabriel,

thank you for letting me know.
Good luck with your code!

Bedrich


On Wed, Feb 16, 2011 at 8:27 AM, GG Balint-Kurti
<Gabriel.Balint-Kurti@Domain.Removed> wrote:
> Dear Bedrich,
>
> Thank you again for your response. I now think that my problems are of my
> own making and are not caused by any failure of the dsyev routine.
>
> Again -- thank you for your trouble.
>
> Gabriel Balint-Kurti
>
> --On 15 February 2011 09:17 -0700 Bedrich Sousedik
> <bedrich.sousedik@Domain.Removed> wrote:
>
> > Dear Gabriel,
> >
> > this indeed seems unexpected.
> > Can you please send me your sample (8x8) matrix and a piece of your code
> > (small extract that would reproduce the problem you are running into).
> >
> > Thank you,
> >
> > Bedrich
> >
> >
> > On Mon, Feb 14, 2011 at 4:07 PM, GG Balint-Kurti
> > <Gabriel.Balint-Kurti@Domain.Removed> wrote:
> > > Dear LAPACK,
> > >
> > > I have been using subroutines dsyev and dsyevd to compute the
> > > eigenvectors and eigenvalues of a symmetric 8x8 matrix. the matrix
> > > formed by these eigenvectors should form a unitary matrix and therefore
> > > be the basis of a unitary transformation. When I test the unitariy of
> > > the resulting matrix I find that it is quite inaccurate. In particular I
> > > find that the scalar product of two of the eigenvectors is 3.8D-5. this
> > > is far less accurate than I would like. As I need to perform at least
> > > 40,000 unitary transformations in my dynamical calculations such a large
> > > error allows a large build-up of errors and renders my calculations
> > > invalid.
> > >
> > > All my variables are double precision. Can you advise as to how I can
> > > improve the accuracy of my eigenvectors.
> > >
> > > Gabriel Balint-Kurti
> > > Professor of Theoretical Chemistry
> > >
> > > ----------------------
> > > Gabriel Balint-Kurti
> > > Centre for Computational Chemistry
> > > School of Chemistry
> > > University of Bristol
> > > Bristol, BS8 1TS
> > > Tel: +44 117 928 7662
> > > fax: +44 117 925 1295
> > > Gabriel.Balint-Kurti@Domain.Removed
> > > _______________________________________________
> > > Lapack mailing list
> > > Lapack@Domain.Removed
> > > http://lists.eecs.utk.edu/mailman/listinfo/lapack
>
>
> ----------------------
> Gabriel Balint-Kurti
> Centre for Computational Chemistry
> School of Chemistry
> University of Bristol
> Bristol, BS8 1TS
> Tel: +44 117 928 7662
> fax: +44 117 925 1295
> Gabriel.Balint-Kurti@Domain.Removed