Re: [Libmesh-users] Dirichlet BCs on a triangulation

[David K. <dav...@ba...>]

>> Every boundary node should be on a boundary side of some element - by
>> what other definition is it a boundary node?
>
> A triangle can have a node which is on the boundary but no side
> which is on the boundary.  However, penalty BC should handle this
> case with no problem.

Yep, I was getting confused, I thought I had to do something special/ 
different with triangles because non boundary triangles can have a  
node on the boundary, but of course I don't because this node is also  
on a boundary side of another triangle. Thanks for pointing out my  
misconception.

It turns out the weird behaviour I was seeing was caused because I  
was imposing of the _wrong_ boundary conditions for the PDE I was  
solving, not because of the way I was imposing the BCs.

Thanks,
David