On Tue, 24 Jun 2008, Benjamin Kirk wrote:

> Big picture, the concept of serialized vectors needs to be wholesale
> replaced with vectors+ghost padding.  PETSc offers a way to do this, and
> Trilinos has something similar.

Keep in mind, for anywhere we still need to use serialized vectors
(Derek's case not included), the padding's going to have to be more
complicated than just "include ghost dofs".  For that HP selection
code of mine, for example, you might need to go several levels of
elements out from the current partition to get enough DoFs to do a
proper coarsening, and for System::project_vector() there may be no
relation between the current dof numbering and the numbering of the
old dofs we're trying to project onto them.

> What I think is needed is that we implement this functionality in
> DistributedVector<> and then derive PetscVector<>/EpetraVector<> from this
> class.

Assuming you mean NumericVector not DistributedVector, that sounds
like an excellent idea.
---
Roy

-------------------------------------------------------------------------
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services for
just about anything Open Source.
http://sourceforge.net/services/buy/index.php
_______________________________________________
Libmesh-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/libmesh-devel

Reply via email to