Thanks Roy, for your reply.
I have additional questions related to the aspect of boundary
conditions with AMR.
Is there a consistent way of handling the issue of applying dirichlet
boundary conditions on a mesh that is being refined/coarsened?
For instance, if a portion of the boundary has dirichlet bc applied
to it, and gets refined, then the new nodes created inside that
portion also should have the same bc applied to it. One way to do
this is to check for nodes on that boundary and apply bc on them. Do
you have a recommendation on this?
Similarly, with the case of Neumann bcs...?
On Jan 16, 2007, at 2:12 PM, Roy Stogner wrote:
> On Tue, 16 Jan 2007, Manav Bhatia wrote:
>> Kindly you help me with a hint about the basic concept of the
>> MeshSmoother class?
>> In my understanding, it does not affect the boundary nodes, but
>> changes some of the domain interior node locations after a
>> coarsening. What is the need for this?
> Mesh generators, as well as mesh-distorting formulations for
> high-deformation mechanics problems, often produce meshes with
> elements that are more distorted than optimal for a finite element
> solver, or even too distorted to solve on at all. Running such a mesh
> through a good smoother algorithm is intended to fix that.
> A mesh smoother may affect boundary as well as interior nodes, just so
> long as the domain of the mesh remains relatively unchanged. In
> practice what I've seen is that smoothers will allow nodes to "slide"
> along flat boundaries, but will keep nodes on corners, edges, or
> curves pinned in place.