Re: [Getfem-users] Test- and Trial space on different meshes

[julien pommier]

Hello Klaus,

> is it possible to assemble a mass matrix where Trial and Test Functions
> are defined on different meshes?
>
> For instance the Test Function is defined on a fine, the Trial Function
> on a coarse mesh \Omega_H, which may be a subset of the fine one
> \Omega_h. Integration is performed over the fine mesh: \int_{\Omega_h}
> \partial\phi \partial\psi* where \partial\psi* is the restriction of
> \psi defined on \Omega_H to \Omega_h.
>
> I know that with asm_mass_matrix it is possible to define two different
> functions mf_u1 and mf_u2 to integrate, but what if they are defined on
> different meshes?
>   
The asm_mass_matrix function will not allow you to use directly two 
mesh_fem with different meshes, you have to go through an 
"interpolated_fem". What you are looking for is explained in the chapter 
"Interpolation of a finite element method on non-matching meshes":
http://www-gmm.insa-toulouse.fr/getfem/getfemuser/getfemuser_15.html

--
Julien