ffc team mailing list archive

[Anders Logg <logg@xxxxxxxxx>]

On Mon, Apr 14, 2008 at 09:18:48AM +0200, Martin Sandve Alnæs wrote:
> 2008/4/14, Kristian Oelgaard <k.b.oelgaard@xxxxxxxxxx>:
> > Quoting Anders Logg <logg@xxxxxxxxx>:
> >
> >
> > > This should be fixed now.
> >  >
> >  > I wasn't able to compile 1D vector elements. Looks like a bug in FFC.
> >
> >
> > Depends how you look at it. If you do:
> >
> >  element = VectorElement("Lagrange", "interval", 2)
> >
> >  FFC will try to generate a vector element with just cell_dimension = 1 component
> >  i.e. effectively a FiniteElement (this will fail because MixedElement expects a
> >  list).
> >
> >  If you do e.g.,
> >
> >  element = VectorElement("Lagrange", "interval", 2, 3)
> >
> >  things work fine.
> >
> >  We could modify VectorElement to return the FiniteElement for the special case
> >  of shape = "interval" and no specified number of components, but on the other
> >  hand I think it is more logical if it doesn't work.
> 
> I agree. Either they are treated as equivalent, or it should fail like
> in FFC today. The latter case could print an informative message.

Yes, I was a little too quick. There is no "natural" vector element in
1D, so it's not a bug.

I'll add a better error message.

-- 
Anders