dolfin team mailing list archive

[Anders Logg <logg@xxxxxxxxx>]

Great!

--
Anders


On Thu, May 19, 2011 at 10:37:24AM -0700, Johan Hake wrote:
> Sure!
>
> I just added the request as comment to the bug:
>
>  "Lacking check of input for function evaluation"
>
> So I remember it when I fix that one :)
>
> Johan
>
>
> On Thursday May 19 2011 10:17:43 Anders Logg wrote:
> > On Thu, May 19, 2011 at 04:35:57PM -0000, Kyle wrote:
> > > Question #158012 on DOLFIN changed:
> > > https://answers.launchpad.net/dolfin/+question/158012
> > >
> > >     Status: Answered => Open
> > >
> > > Kyle is still having a problem:
> > > Oh sorry!
> > >
> > > Yes I am using the Taylor-Hood elements.
> >
> > Then just do something like this:
> >
> > for v in vertices(mesh)
> >   ... = U(v.point())
> >
> > for e in edges(mesh)
> >   ... = U(e.midpoint())
> >
> > I noticed now that function evaluation does not work for the Point
> > class. It requires instead the coordinates as input so you need to do
> > something like
> >
> >   p = v.point()
> >   ... = U((v.x(), v.y()))
> >
> > instead.
> >
> > Note to Hake: Would it be easy to add the possibility of supplying a
> > Point argument to __call__?
> >
> > > Is this the code located in
> > > fsiproblem.py in sandbox/fsi? I looked through it but it looked a little
> > > larger than I thought it was going to be.
> >
> > Take a look in the file subproblems.py and search for "transfer". You
> > might also want to look in fsiproblem.py for how the mapping between
> > vertex numbers and edge numbers in the two meshes are computed.
> >