dolfin team mailing list archive

[Anders Logg <logg@xxxxxxxxx>]

On Tue, Oct 06, 2009 at 09:10:37AM -0500, Andy Ray Terrel wrote:
> On Tue, Oct 6, 2009 at 8:48 AM, Anders Logg <logg@xxxxxxxxx> wrote:
> > On Tue, Oct 06, 2009 at 02:26:32PM +0100, Garth N. Wells wrote:
> >>
> >>
> >> Johan Hake wrote:
> >> > Hello!
> >> >
> >> > dolfin::Form is not working properly yet after the changes. When assembling a
> >> > form of only coefficients, no basisfunctions, we are not able to extract the
> >> > mesh.
> >> >
> >>
> >> Functionals have been a bit tricky since we introduced FunctionSpace.
> >> Perhaps their constructor should take a Mesh?
> >
>
> +1
>
> I've been trying to work around this for the last day.  My problem
> being that I often use functionals to guide my algorithm, I use to be
> able to just define them and then call them in the loop.  Now I have
> to wait until the coeffiecient function has a mesh attached to it,
> making the logic more difficult.

This should be fixed now.

In C++, one has to supply the mesh as the first argument to the
constructor of the form when the form is a functional:

  Lift::Functional L(mesh, p);

In Python, one has to supply the mesh as a keyword argument to the
assemble function:

  lift = assemble(L, mesh=mesh)

--
Anders

Attachment: signature.asc
Description: Digital signature