ffc team mailing list archive

[Anders Logg <logg@xxxxxxxxx>]

On Fri, Jan 29, 2010 at 09:08:44PM +0100, Kristian Oelgaard wrote:
>
>
> On 29 January 2010 20:21, Anders Logg <logg@xxxxxxxxx> wrote:
> >On Fri, Jan 29, 2010 at 04:13:13PM +0100, Kristian Oelgaard wrote:
> >>I think it is almost there, but we need to rethink the way we
> >>manipulate the degrees of finite elements and forms, I think we have
> >>discussed this before but I don't recall if we ever reached a
> >>conclusion.  Try running the QuadratureElement.ufl demo.
> >
> >I don't see exactly what the problem is. Is it the automatic degree
> >selection?
> >
> >I've added some more debug output and the following is now printed for
> >the QuadratureElement demo:
> >
> > Found quadrature element(s) with the following degree(s): [3]
> > Selecting quadrature degree based on quadrature element: 3
> > quadrature_degree: auto --> 3
> >
> >That looks correct to me. Shouldn't this result in the same points
> >being chosen for both the QuadratureElement and the quadrature loop?
>
> I see now that it is only a problem for tensor representation. For
> the linear form the first term v*u*dx makes it a form of degree 4,
> and tensor representation will use that degree for the entire form
> if I remember correctly?

Yes, and that's not much of a problem since it happens at compile
time.

But for the FFC demo the representation is chosen as quadrature since
it involves quadrature elements, so if it fails it shouldn't be
related to tensor representation?

--
Anders

Attachment: signature.asc
Description: Digital signature