On Mon, 28 Jan 2008, Umar Farooq wrote:

> The two matrices have a different non-zero structure...entries in diferent
> indicies.

Oh, I understand - yes, this should be different depending on how many
processors you use.  libMesh numbers degrees of freedom based on its
element numbering, which in turn is based on how the mesh is
partitioned, which differs depending on how many partitions are used.

The matrix considered as an operator on finite element functions
hasn't changed, just the indexing of those functions' coefficients
has.

> What iterator should be used? local_elements?

It's up to you.  If you're testing node positions to find your
boundaries, local_nodes is reasonable; if you're using boundary ids
(or if you want to be able to use integrated rather than nodal penalty
terms later) then the best thing to do is loop over
active_local_elements then loop over nodes in each element.

> Im not using a refined mesh.

Maybe not now, but use active_local_elements anyway.  There's no
difference between that and local_elements on an unrefined mesh, but
if you started using refinement later local_elements could be a bug
waiting to happen.
---
Roy

-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
Libmesh-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/libmesh-devel

Reply via email to