Re: [Getfem-users] interpolated fem again

[Andriy Andreykiv]

Dear Yves,

       What I want to do with interpolated fem is the following. I want to 
simulate a composite material (something like a car distribution belt), 
where the matrix of the composite is simulated with hexahedral elements 
while the fibers, that enforce the composite material are simulated with 
trusses. I want those trusses not to be conformal with the mesh of the 
matrix material, hence I want to impose a weak constraint that displacement 
of the fibers should be equal to the displacement of the matrix material  in 
the location where those trusses fibers cross the matrix (this way I would 
attach the fibers inside the matrix  material). Hence, again this is 
conceptually similar to fictitious domain method for fluid-structure 
interection, where velocity of the fluid on the boundary of the structure 
should be equal the velocity of the structure and this constraint is imposed 
in a weak sence.
       I probably can reformulate my vectorial problem into a  set of 
scalar problems and then use interpolated fem, and then assemble a vectorial 
problem again. Would that be a solution?

Thank you in advance,
                                       Andriy


----- Original Message ----- 
From: "Yves Renard" <address@hidden>
To: "Andriy Andreykiv" <address@hidden>
Cc: <address@hidden>
Sent: Wednesday, July 11, 2007 6:44 PM
Subject: Re: [Getfem-users] memory leakage somewhere in 
getfem::mesh_im_level_set (and interpolated_fem again)


Le mardi 10 juillet 2007 10:32, vous avez écrit :
> Dear Yves,
>
>     Thank you very much. There is no memory leakage any more.
> By the way, may be you want to know. I noticed that you can't compile some
> of the /contrib files anymore with the latest getfem from SVN. For
> instance, in xfem_contact you have the line
>
> 278:    pfp1 = fem_precomp(pf1, &pai1->integration_points());
>
> while fem_precomp has 3 arguments, so it gives an error. So, I editted it
> like this
>
>     pfp1 = fem_precomp(pf1,
> &pai1->integration_points(),mim.int_method_of_element(cv1));
>
> and it worked.
>
> After I continued running "make check", there were other compilation
> errors, but I didn't go through them anymore.

Ok, I made the corrections. Thank you.

> Another question if I may. In getfem_interpolated_fem.cc it says that
> extension of interpolated_fem to vectorial problems is straightforward and
> there is a short explanation. Would you be so kind to give a bit more
> elaborate explanation on what should there be done in order to do this. Do
> you think that somebody from outside of Getfem development crew can do it?

In fact, if you just want to use your interpolated fem on a vectorial 
problem,
it should work as usual (I think, but I never tried) whenever your original
fem is scalar. To make the things working for original vectorial elements,
real_base_value and grad_base_values functions have to be adapted. But you
have to take into account the fact that the original element can be a direct
product of scalar element or an intrinsic vectorial element (like Raviart
Thomas elements for instance). This is not completely obvious, in fact.

Can you describe a little bit what you expect to do with a vectorial
interpolated fem ? May be there is a better solution than using this
fonctionality.

Yves.

-------------------------------------------------------------------------
 Yves Renard (address@hidden)       tel : (33) 04.72.43.80.11
 Pole de Mathematiques,                       fax : (33) 04.72.43.85.29
 Institut Camille Jordan - CNRS UMR 5208
 INSA de Lyon, Universite de Lyon
 20, rue Albert Einstein
 69621 Villeurbanne Cedex, FRANCE
 http://math.univ-lyon1.fr/~renard
-------------------------------------------------------------------------