getfem-users
[Top][All Lists]

## Re: [Getfem-users] defining diffusion coefficients

 From: Arvind Ajoy Subject: Re: [Getfem-users] defining diffusion coefficients Date: Fri, 26 Sep 2008 14:22:28 +0530

Dear Marco

This might be useful.

m = gf_mesh('cartesian',[0:0.1:1],[0:0.1:1])
mf = gf_mesh_fem(m,1);
gf_mesh_fem_set(mf,'fem',gf_fem('FEM_QK(2,1)'));
mim = gf_mesh_im(m,gf_integ('IM_NC_PARALLELEPIPED(2,1)'));
F = gf_mesh_fem_get(mf,'eval',{'(x.^2 + y.^2)'});
gf_plot(mf,F)
size(F)
gf_mesh_get(m,'nbpts')
gf_mesh_fem_get(mf,'nbdof')

The gf_mesh_fem_get(mf,'eval', ()) allows you to define a quantity on the
mesh.

Regards
Arvind

On Fri, Sep 26, 2008 at 1:31 PM, marco trevisan wrote:
Hello,
I am working wit the Matlab interface on a simple diffusion-transport
problem on a rectangular domain. I am using a rectangular mesh and Q1
elements, so the nodes of the mesh and the dofs are the same quantity.
Since the diffusion coefficient A is not constant but varies strongly
within a specified "path" in the domain, I would like to know if there
is a method of defining this coefficient as a scalar function or
matrix over the domain and then converting it to a convenient object (
which, I suppose, is a vector of size [ 1 , number_of_dof ] ) .