From: Roy S. <roy...@ic...> - 2011-03-23 05:34:40
|
On Wed, 23 Mar 2011, Joh...@un... wrote: > I'd like to add a variable to a DofMap, that lives not on the whole > mesh but only on its surface. Is subdomain the right thing, that > should work or do I have to do something completely different? I don't think we support this with fully-coupled solves right now. Per-subdomain variables work if you want a variable that only lives on a subset of the interior. If you want a loosely coupled variable that only lives on a surface then you create a BoundaryMesh and create a separate system on it. But we don't have any way to solve such variables in the same (non)linear System as the interior Mesh equations. I don't think there's anything in the libMesh architecture that would *prevent* adding such a capability, but the addition would require some serious library internal changes. --- Roy |