Re: [Libmesh-users] interaction between subdomain_id and dof constraints?

[Roy S. <roy...@ic...>]

On Tue, 27 Apr 2010, Boyce Griffith wrote:

> I am trying to use the DofMap to impose Dirichlet-type boundary
> conditions within an explicit nonlinear mechanics solver which is
> embedded in a larger fluid-structure interaction solver.  I constrain
> the velocity and force DOFs at Dirichlet boundaries to be zero by adding
> a DofConstraintRow for each of the constrained DOFs.

This should be a good method for homogeneous constraints on LAGRANGE
(or HIERARCHIC, or pretty much anything but CLOUGH) elements.  I don't
know if anyone's ever actually tested using it before, though.

> This seems to work well as long as I do not also have subdomains in my
> mesh; however, as soon as I start labeling different subdomains, the dof
> constraints no longer impose the boundary conditions.  Is this expected
> behavior?  Is there something special that I should be doing in the case
> of multiple subdomains?

This is very surprising - unless you're adding subdomain-specific
variables, in general the only part of the code that cares about
subdomains is the I/O (to read them in and write them out), the
refinement (to assign them to new child elements), and your own user
code!  Could you verify this problem in a small example?

> By the way, subdomain_id seems to be the most direct way to assign
> different material properties to different parts of the mesh.  Are
> there any alternatives which I should use instead?

Subdomain ids are ideal unless you have a very large number of
different materials or you have continuously varying material
properties; in those cases you'd want to use a separate ExplicitSystem
with properties as variable fields.
---
Roy