ffc team mailing list archive

[Kristian Oelgaard <k.b.oelgaard@xxxxxxxxxx>]

Quoting "Garth N. Wells" <gnw20@xxxxxxxxx>:

>
>
> Peter Brune wrote:
> > Hey -
> >
> > Is there any interest (besides me) for the implementation of the facet
> > normal in FFC?  I've found reason to not want these as propagated from
> > DOLFIN.
>
> It would be nice to compute the normal inside the generated code.
> Kristian is best placed to comment.
>
> Garth
>
> I would vote that they should be calculated in the
> > facet_determinant code-snippet, as they consist of a subset of the
> > values computed therein, and are typically required by related pieces of
> > code.  There also seems to be reasonable machinery for this from FIAT.

It shouldn't be too difficult. For affine mappings we could easily put this in
with the facet_determinant code-snippet and just remove it if it is not use. We
do this often anyway. However, how should the normal be computer for higher
order geometries and where would we put that?

Kristian

> > Thanks,
> >
> > - Peter
> >
> >
> > ------------------------------------------------------------------------
> >
> > _______________________________________________
> > FFC-dev mailing list
> > FFC-dev@xxxxxxxxxx
> > http://www.fenics.org/mailman/listinfo/ffc-dev
>
>
> _______________________________________________
> FFC-dev mailing list
> FFC-dev@xxxxxxxxxx
> http://www.fenics.org/mailman/listinfo/ffc-dev
>