dolfin team mailing list archive

[Anders Logg <logg@xxxxxxxxx>]

On Wed, Jun 18, 2008 at 01:33:06PM +0200, Dag Lindbo wrote:
> Anders Logg wrote:
> > On Tue, Jun 17, 2008 at 10:52:06PM +0200, Dag Lindbo wrote:
> >> Hello again!
> >>
> >> Is it correct that Periodic BCs won't work for vector-valued systems?
> >> There is a comment to suggest this in PeriodicBC.cpp. Is there a
> >> reasonable workaround for the needy?
> >>
> >> /Dag
> > 
> > I don't know of an easy workaround.
> > 
> > But maybe you can take a look at the code in PeriodicBC.cpp to see if
> > it's possible to understand what it does and then extend it? I guess
> > the extension will be nontrivial, but the code is fairly well
> > documented so it should not be impossible.
> 
> Paper submission deadline approaching fast so I don't really have a
> choice but to fix it!

Nice.

> It currently works for "simple" forms (i.e. scalar non-mixed), so it
> seems natural to have
> PeriodicBC::apply(GenericMatrix& A, GenericVector& b, const Form& form)
> pick out the "simple" finite elements and dof-maps from the form and
> call apply(A, b, dof_map, form) for each. Any thoughts?

I'm not sure. Perhaps you may just use the top-level dof-map to handle
everything at once.

-- 
Anders

Attachment: signature.asc
Description: Digital signature