ffc team mailing list archive

[Anders Logg <logg@xxxxxxxxx>]

On Tue, Oct 30, 2007 at 09:55:42PM +0100, Bartosz Sawicki wrote:
> > The reason for why are there are no examples using the Nedelec elements 
> > is probably that we have not finished implementing and testing them yet. 
> > In other words, they are not working properly yet.
> 
> Ok, but there are probably some particular examples where it works. 
> Maybe I can help to improve the functionality.
> 
> >> My first, the simplest try was:
> >>
> >> element = VectorElement("Nedelec", "tetrahedron", 1, 3)
>  >
> > Further, the Nedelec elements are vector-elements by  default, so you want
> > 
> >    element = FiniteElement("Nedelec", "tetrahedron", 1)
> > 
> > here. (With the notation you used above, you get a vector of 3 Nedelec 
> > vector elements.)
> 
> Yeeeap :), it was the problem. Now fcc compiles correctly.
> Thank you very much.

It might compile, but I wouldn't count on getting anything sensible
(but you might be lucky). As Marie points out, the Nedelec elements
need some more work.

/Anders

> >> v = TestFunction(element)
> >> t = TrialFunction(element)
> >> c = VectorConstant("tetrahedron")
> >> a = dot(curl(v),curl(t))*dx
> >> L = c*dx
> >>   
> > 
> > In addition L should be a linear functional. Thus c*dx where c is a 
> > vector does not make sense in the ffc-regime. Maybe you want something like
> > 
> >    L = dot(curl(v), c)*dx 
> > instead?
> 
> Yes, of course.
> On the right side of PDE equation I have only source vector funtion (c), 
> so as far as I understand weak formulation there should be:
> L = dot(v, c)*dx
> 
> regards,
> Bartosz Sawicki
> 
> 
> _______________________________________________
> FFC-dev mailing list
> FFC-dev@xxxxxxxxxx
> http://www.fenics.org/mailman/listinfo/ffc-dev