ffc team mailing list archive

[Anders Logg <logg@xxxxxxxxx>]

On Mon, May 08, 2006 at 11:01:30AM +0200, Garth N. Wells wrote:
> On Mon, 2006-05-08 at 10:20 +0200, Anders Logg wrote:
> > On Mon, May 08, 2006 at 07:56:45AM +0200, Garth N. Wells wrote:
> > > On Sun, 2006-05-07 at 23:03 +0200, Anders Logg wrote:
> > > > I haven't tried, and I also haven't updated the additional formats,
> > > > including PyDOLFIN.
> > > > 
> > > > I plan to extend the Poisson solver module with general Robin
> > > > boundary conditions and try to gain some confidence that everything
> > > > works. I looked over your additions and everything seems to be in
> > > > order.
> > > > 
> > > 
> > > Do you plan to introduce the normal vector to the domain in FFC, or will
> > > you treat it as a function?
> > 
> > I would like to treat it as a Function (like we do with MeshSize), but
> > since it's discontinuous across the facets, a special solution might be
> > needed.
> > 
> > > At the moment, functions on boundaries are defined suing the trace of
> > > the basis functions for a cell. It would be useful and more general to
> > > define functions using a basis defined on the facet.  Would this be
> > > difficult? Functions could, for example, be discontinuous from one facet
> > > to the next (in which case the normal vector could be treated as a
> > > function).
> > 
> > I haven't thought about this in too much detail. Are there other
> > possibly discontinuous functions than the normal that one would like
> > to specify on the boundary? If so, we should add general support for
> > that and have the normal as a special predefined function. We could
> > probably do something like overloading eval() in Function with
> > something that takes facet as an extra argument.
> > 
> 
> There are definitely other functions that are discontinuous. A simple
> example is applying a boundary condition on the face of a polygonal
> domain. When assuming that the boundary condition function comes from a
> continuous space, it "creeps" around the corner of the face. You can see
> this is src/demo/pde/poisson.

ok, I'll think about how to implement this.

/Anders


> Garth 
> 
> 
> > /Anders
> > 
> > 
> > > Garth   
> > > 
> > > > /Anders
> > > > 
> > > > 	On Sun, May 07, 2006 at 06:58:48PM +0200, Garth N. Wells wrote:
> > > > > Great! Does it work with the "-f blas" option?
> > > > > 
> > > > > Garth
> > > > > 
> > > > > On Sun, 2006-05-07 at 15:57 +0200, FFC wrote:
> > > > > > One or more new changesets pushed to the primary FFC repository.
> > > > > > A short summary of the last three changesets is included below.
> > > > > > 
> > > > > > changeset:   387:7e888efabfdc4987f08c5602f570a92a2dfb3747
> > > > > > tag:         tip
> > > > > > user:        "Anders Logg <logg@xxxxxxxxx>"
> > > > > > date:        Sun May  7 15:55:37 2006 +0200
> > > > > > files:       ChangeLog src/demo/Poisson.form src/ffc/compiler/compiler.py src/ffc/compiler/elementtensor.py src/ffc/compiler/finiteelement.py src/ffc/compiler/form.py src/ffc/compiler/mixedelement.py src/ffc/compiler/monomialintegration.py src/ffc/compiler/projection.py src/ffc/compiler/referencetensor.py src/ffc/format/dolfin.py src/reference/Elasticity.h src/reference/EnergyNorm.h src/reference/Heat.h src/reference/Mass.h src/reference/NavierStokes.h src/reference/NeumannProblem.h src/reference/Optimization.h src/reference/Poisson.h src/reference/PoissonSystem.h src/reference/Projection.h src/reference/Stokes.h src/reference/TensorWeightedPoisson.h
> > > > > > description:
> > > > > > Add support for boundary integrals.
> > > > > > 
> > > > > > Changes pulled from Garth's tree with some additional updates,
> > > > > > mainly fixes for mixed elements and projections to make all
> > > > > > tests in src/reference compile correctly.
> > > > > > 
> > > > > > 
> > > > > > changeset:   386:45a447e390950e45145bf0e3fb4092104505ee40
> > > > > > user:        "Garth N. Wells <g.n.wells@xxxxxxxxxx>"
> > > > > > date:        Thu May  4 02:41:50 2006 +0200
> > > > > > files:       src/demo/Poisson.form src/ffc/format/dolfin.py
> > > > > > description:
> > > > > > Add missing "break;" in DOLFIN output.
> > > > > > 
> > > > > > 
> > > > > > changeset:   385:8301d56d241aacf9e7ab15921710ff6ca83d5ba4
> > > > > > user:        "Garth N. Wells <g.n.wells@xxxxxxxxxx>"
> > > > > > date:        Tue May  2 19:44:01 2006 +0200
> > > > > > files:       src/ffc/compiler/compiler.py src/ffc/compiler/elementtensor.py src/ffc/format/dolfin.py
> > > > > > description:
> > > > > > Deal with terms in geometry tensor for boundaries.
> > > > > > 
> > > > > > 
> > > > > > ----------------------------------------------------
> > > > > > For more details, visit http://www.fenics.org/hg/ffc
> > > > > > 
> > > > > > _______________________________________________
> > > > > > FFC-dev mailing list
> > > > > > FFC-dev@xxxxxxxxxx
> > > > > > http://www.fenics.org/cgi-bin/mailman/listinfo/ffc-dev
> > > > > 
> > > > > 
> > > > > _______________________________________________
> > > > > FFC-dev mailing list
> > > > > FFC-dev@xxxxxxxxxx
> > > > > http://www.fenics.org/cgi-bin/mailman/listinfo/ffc-dev
> > > > 
> > > > _______________________________________________
> > > > FFC-dev mailing list
> > > > FFC-dev@xxxxxxxxxx
> > > > http://www.fenics.org/cgi-bin/mailman/listinfo/ffc-dev
> > > 
> > 
> > _______________________________________________
> > FFC-dev mailing list
> > FFC-dev@xxxxxxxxxx
> > http://www.fenics.org/cgi-bin/mailman/listinfo/ffc-dev
>