On Mon, 5 Jan 2009, Yujie wrote:

> Because there is a mesh interface
> between two processors, I find some dof number is less than first dof
> number.

Right.

> I guess these dof numbers should belong to the processor with low ID, right?

That's the situation right now; we may change the rule assigning
boundary nodes to processors in the future.  In any case, those dof
numbers will belong to the processor given by their processor_id().

> If it is, to current processor, I just omit these dof numbers?

That depends on what you're trying to do.  But rest assured that if
they're non-local on one processor, they'll be local on another - so
if you want to do something with them exactly once, omit them on the
non-local processor.

> They are not counted in dofs using n_local_dofs(), right?

Right.

---
Roy

------------------------------------------------------------------------------
_______________________________________________
Libmesh-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/libmesh-users

Reply via email to