Thank you Raul and Mike.

I wrote:

> Eigenvalues again. For structural dynamics it is the generalised 
> Eigenvalue problem that is needed, the solution to
>
> (A - w B) r = 0
>
> where A and B are real square matrices, w the eigenvalue and r is the 
> eigenvector.

Raul asked:

> For your case are A and B known, and w and r unknown?  Or something
  different?

Yes, A and B are known, and w and r are unknown.

Mike suggested:

> There's a dgegv.f in ~\addons\math\lapack .  Its preamble comments
  say it's deprecated in favour of dggev... so are you looking for an
  implementation of dggev.ijs?

Yes, I read that too. Though dgegv does what I want, maybe dggev is
  more accurate.

I have subsequently found that at the moment I don't need either of
them, because for the typical problems I am attempting B is always a
unit matrix, so all I need is dgeev, which I've got.

Thanks for your help. Sorry for the false alarm.

Graham


----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to