Re: [Libmesh-users] Hanging nodes, constraints ignored during mesh.read()

[Vijay S. M. <vi...@gm...>]

hmm well thanks for the quick reply John. I haven't has enough time to
play around too much with the AMR capabilities but I've been playing
around with it for an eigenvalue problem. It would be useful for me to
reuse the same mesh as the eigenproblem because the mesh resolves the
discontinuities in the solution decently and the transient solution
does not change much spatially. I currently just wrote the final mesh
in eigenvalue simulation to a file and re-read that for my transient
and proceeded. Of course, things don't quite look the same because now
I've lost all information about the constraints.

Anyway, I have another question based on your answer though. Is there
a routine to make a given non-conforming mesh to level-0 conforming
mesh so that it can be read correctly ? If this requires a lot of code
addition, then maybe I'll just try to run the eigenproblem and
transient in one run so that the mesh changes are preserved in memory.

Vijay

On Fri, Aug 7, 2009 at 3:38 PM, John
Peterson<pet...@cf...> wrote:
> On Fri, Aug 7, 2009 at 4:27 PM, Vijay S. Mahadevan<vi...@gm...> wrote:
>> Hi,
>>
>> I read a mesh file (.msh) generated after some AMR sequence during a
>> separate run. The resulting mesh has few hanging nodes and associated
>> constraints. When I read this mesh and start the simulation anew, the
>> number of constrained dofs in the mesh is 0. Is there anyway, I can
>> ask DofMap to calculate the constraints based on whether there is a
>> hanging node or not ?
>
> I haven't looked closely but I'm pretty sure there is no support for
> reading in a family of refined gmsh meshes, and there is definitely no
> support for doing calculations on a single non-conforming mesh read
> from file.
>
> When reading in xda style meshes, we still assume there is a
> conforming level-0 mesh, since this required for most of the
> projection routines used in libmesh, and the entire hierarchy of the
> grid is written to xda files to support this.
>
> --
> John
>