← Back to team overview

dolfin team mailing list archive

Re: Improved handling of subdomains

 

On 03/11/2011 11:44 AM, Johannes Ring wrote:
On Fri, Mar 11, 2011 at 10:49 AM, Marie E. Rognes<meg@xxxxxxxxx>  wrote:
On 03/11/2011 10:43 AM, Anders Logg wrote:

On Fri, Mar 11, 2011 at 10:40:40AM +0100, Marie E. Rognes wrote:

On 03/11/2011 01:24 AM, Anders Logg wrote:

One can now attach subdomains to a form:

   a.cell_domains = cell_domains
   a.exterior_facet_domains = exterior_facet_domains
   a.interior_facet_domains = interior_facet_domains

As a consequence, one does no longer need to send these arguments to
the assembler. Another consequence is we can simplify the
VariationalProblem class and remove the extra functions that contain
these arguments (will fix tomorrow). It also means one can use
VariationalProblem for cases where different domains are needed for a
and L.

It is still possible to send the arguments to the assembler so it's
backwards compatible. It is also (now) possible to specify the domains
using MeshData with names "cell_domains", "exterior_facet_domains" and
"interior_facet_domains".

Arguments to the assembler will override subdomains attached to a
form which in turn overrides subdomains specified as part of the mesh.


Brilliant. I'll apply this to the error control stuff, and then we
only need to figure out the adaption of mesh functions.

My buildbot currently fails when running one of the unit tests. Don't
know why because it works fine here.

Try running

   test/unit/fem/python$ python test.py

I'll restart it again and see if it works this time.

I'll wait until things make it into main, and bot is green again then ;-)

My buildbot also suddenly broke at the parallel-assembly-solve system test.
What is the deal there?

I guess you have merged from main after I exposed a hidden failure in
this system test a few days ago. Garth made some updates yesterday but
the test is still failing.

Ok, thanks.

--
Marie




References