dolfin team mailing list archive

[Anders Logg <logg@xxxxxxxxx>]

On Fri, May 30, 2008 at 10:02:29AM +0200, Anne Voigt wrote:
> Hello all,
> 
> I'm working with FFC for quite a while and I like the easy way to handle
> multilinear forms with the help of FFC.
> 
> But now I have a problem where I'm not sure if it is possible to
> programme this with FFC.
> 
> I'm working on an optimal control problem where I have to solve an
> integral over just a part of the whole twodimensional domain (x,y) \in
> \Omega. So what I need is the bilinear form
> 
> \int_{y_0}^{y_1} u*v dy where [y_0,y_1] \in \Omega
> 
> Is it possible to implement something like that in FFC and how can I do
> that?  
> 
> Hope somebody can help me?!

Yes, that's easy (but not so well documented I admit).

The simplest way is to just supply a SubDomain as argument to the assembler:

  assemble(A, form, mesh, sub_domain);

Create a SubDomain which overloads the inside() function, returning
True when you are inside your subdomain, for example

  class MySubDomain : public SubDomain
  {
    bool inside(const real* x, bool on_boundary) const
    {
      return x[0] + x[1] > 1.0;
    }
  };  

If you can't specify your domain based on an expression, you need to
create a MeshFunction that labels which cells are included in your
domain.

It's also possible to have different integrals over different
subdomains, integrals over subsets of the boundary etc.

-- 
Anders