Hi Bruno,
AFAIK, there is a simple solution: make initial vector (or subspace)
perpendicular to those constrained entries.
That is, if you do Lancoz, set random initial vector and then zero out
constrained DoFs.
Then being Krylov-based method it should form subspaces {x, Ax, A^2x,...}
orthogonal to those constrained DoFs, so
you should not get any issues.
Cheers,
Denis.
On Thursday, November 30, 2017 at 4:24:05 PM UTC+1, Bruno Turcksin wrote:
>
>
> Hi all,
>
> In step-36, there is an explanation on how Dirichlet boundary conditions
> introduce spurious eigenvalues because some dofs are constrained. However,
> there is no mention of hanging nodes. So I am wondering if I can treat them
> as shown for the Dirichlet boundary, i.e, the only difference between a
> hanging node and a Dirichlet is what happens in
> ConstraintMatrix::distribute(). I also wonder if there is a way to avoid
> having these spurious eigenvalues computed or if the only way to deal with
> them is to redo the calculation after changing the entries in the matrix.
>
> Best,
>
> Bruno
>
--
The deal.II project is located at http://www.dealii.org/
For mailing list/forum options, see
https://groups.google.com/d/forum/dealii?hl=en
---
You received this message because you are subscribed to the Google Groups
"deal.II User Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/d/optout.
