Re: [ff3d-users] questions...

[Stephane Del Pino]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On samedi 24 Avril 2004 18:48, address@hidden wrote:
> Hello everybody,
Hello Jean-Michel.

> I am very new in the world of ff3d and I have a questions. I hope somebody
> will be so nice to help me :) I have tried to introduced in a code written
> in ff3d language an object using the language of pov-ray but I have seen
> that it is not very satisfying because I can't obtain a finite element
> mesh.
No. There is no volume mesher in ff3d. This is why you can solve your problems 
using fictitious domain like methods. Those methods do not require a 3D mesh 
that fits your geometry. You can find few details about it in the manual 
(that is under construction).

> So I have 3 questions: 
> 1) It is possible to read and use from a ff3d code a mesh constructed from
> GMSH?
Yes! In fact ff3d can read unstructured meshes using the 'mesh' format. 
Unfortunately this is still undocumented --- I am trying to find the time to 
do so :-(

> 2) If yes, how can I reference the elements and part of the surface to take
> into account the boundary conditions?
This is very simple as the example will show.

> 3) Can somebody send me an example in such a way I can understand better
> how it is possible to do these things? (obviously if it is possible... )
> Thanks a lot to everybody will try to help me.
First, everything work exactly as it does in the fictitious domain case. The 
only notable difference is that one has to replace
        solve(u) in Omega by Mesh
by
        solve(u) in Mesh
since now the Mesh is the computational domain!

If one wants to manage materials referenced by the mesh, one can defined 
"reference" functions that way:

function f1 = reference(elements, Mesh, 1:g, 2:h, ...);
or
function f2 = reference(vertices, Mesh, 1:g, 2:h, ...);

The first case is related to function defined using elements references and 
the second vertices references. Both functions are built using the mesh 
'Mesh' and there definition reads:

Associate the function (or constant) g to reference 1 and h to reference 2. I 
think this is quite simple ...
! Note that if some references are not associated to any function, the default 
value 0 will be return when f will be evaluated!

To specify boundary conditions, proceed like that:

solve(u) in Mesh
{
  pde(u) -div(grad(u)) = f;
  u = u0 on 1,2,3;
  dnu(u) = u1 on 4;
}

where 1,2,3 and 4 are boundary elements references.

I think that answers the question.

Best regards,
Stephane.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (GNU/Linux)

iD8DBQFAjB1YbZyhKhZwdc0RAjfCAJ0SGJ966c3aT0l9Hw0zN+kaj9z5qgCg4xy9
ZyzddyRwa1mRKdCBeJ/uq6I=
=ehKf
-----END PGP SIGNATURE-----