dolfin team mailing list archive

[Anders Logg <logg@xxxxxxxxx>]

Alessio Quaglino wrote:
> Ok very simple thanks (I thought it might have been dependent to the
> finite element shape). I still have one question about the new Funcion
> interface. When I define a User-defined Function, only eval() is used for
> the FEM assembly?

Yes, and note that there are two versions, one for general tensor-valued 
functions and a simple one for scalar functions. You can choose to 
overload any of these.

> I ask this because I have some problems in understanding
> if overloading this method is enough in order to define a source term.

Yes, that should be enough.

> Also this method is defined as const, hence if this happens to be the
> case, I have to redisign a part of my source class. Thanks.

Yes, it needs to be const. If your class needs to modify some data, then 
make the data mutable.

/Anders


> Alessio
>
>
> > Just do the following:
> >
> > Mesh mesh; Function f;
> >
> > real* vertex_values = new real[mesh.numVertices()];
> > f.interpolate(vertex_values);
> >
> > Then vertex_values will be an array with the values at all vertices.
> >
> > For a scalar function, you simply get all the vertex values and for a
> > vector (or in general tensor-valued) function, you first get all the
> > values for component 0, then all values for component 1 etc.
> >
> > /Anders
> >
> >
> > Alessio Quaglino wrote:
> > > > On Tue, Jun 26, 2007 at 02:56:52PM +0200, Alessio Quaglino wrote:
> > > > > I'm trying to get my code working with the new dolfin 0.7.0-1 but it
> > > > > seems
> > > > > there are a several problems. For example, a part of the
> > > > > dolfin::Function
> > > > > interface has disappeared. In particular, it is not possible to get
> > > > > the
> > > > > value corresponding to a vertex or the finite element space where the
> > > > > function is living. Have these functionalities been moved somewhere
> > > > > else?
> > > > >
> > > > > Regards,
> > > > > Alessio Quaglino
> > > > Use Function::interpolate() to access the values. Often, one can
> > > > avoid ever needing to access the values directly, but if they should
> > > > be needed, then it is possible to either interpolate the Function to a
> > > > given Cell (getting the expansion coefficients in the local finite
> > > > element basis) or get the values at all vertices.
> > > >
> > > > So, try to avoid accessing the values but if you really need the
> > > > values at vertices you can still get them.
> > > >
> > > > /Anders
> > > > _______________________________________________
> > > > DOLFIN-dev mailing list
> > > > DOLFIN-dev@xxxxxxxxxx
> > > > http://www.fenics.org/mailman/listinfo/dolfin-dev
> > > Ok I see, however with interpolate() I can either get all the Vertex
> > > values or the values at a given Cell. While the latter case is
> > > unefficient
> > > for my purpose  because I'd pass through the same Vertex several times,
> > > in
> > > the former I get a vector with the values at all vertices but no
> > > information about how to access such a vector.
> > >
> > > Regards,
> > > Alessio
> > >
> > >
> > > _______________________________________________
> > > DOLFIN-dev mailing list
> > > DOLFIN-dev@xxxxxxxxxx
> > > http://www.fenics.org/mailman/listinfo/dolfin-dev
>
>
> _______________________________________________
> DOLFIN-dev mailing list
> DOLFIN-dev@xxxxxxxxxx
> http://www.fenics.org/mailman/listinfo/dolfin-dev