Multispace Laplace problem

Rene Velasquez

unread,

Mar 1, 2019, 8:28:09 PM3/1/19

to bempp

Hi everyone,

I'm having trouble solving a multispace Laplace problem, I can't even solve the equation because I have a problem with the blocked operator. I don't know if I misunderstood some concept or if I missing something in the code. I share with you the code and the grids. Any help would be appreciated. Thanks in advance!

Salutations,

René.

cylinder.msh

cylinder_2.msh

matrix.msh

S2-E1-CE 01-03-Copy1 (1).ipynb

Elwin

unread,

Mar 4, 2019, 5:28:31 AM3/4/19

to bempp

Dear René,

You will have to give a closer look at the function spaces you are using.

First, you are using discontinuous spaces, where continuous spaces would suffice as well: “P1” instead of “DP1”. This will reduce the number of DOFs.

The error message that is given means that the number of DOFs does not match in the blocked matrix. Specifically, all matrices in the same block column need to have the same number of columns. That is, all operators in each column of the block matrix need to have the same domain space. In your case, the third column has both a K2 and a K3 operator, which have a different number of DOFs in their domain spaces.

Actually, the same problem happens in the first and second column as well, where you use the domain spaces DS and DS2, which just happen to have exactly the same number of DOFs. Therefore, the test is passed whereas the spaces are still not compatible.

I would suggest reviewing the formulation carefully, including all function spaces.

Best,

Elwin

Rene Velasquez

unread,

Jun 29, 2019, 11:47:47 PM6/29/19

to bempp

Hi Elwin,

I change the formulation of the problem,

and I don't have the same error as before. But I get this message:

And I don't know if it's a computational problem. I thoght if I reduce the number of elements in the mesh it would solve the problem., but it wasn't the case. Normally I work with meshes of 5500 elements, and I reduce it to 3000. But the problem it's still there. What are your recomendations? (Also I change all the discontinous spaces to continous!)

Timo Betcke

unread,

Jul 2, 2019, 10:22:23 AM7/2/19

to bempp

Hi,

this usually points to a memory issue. Did you try it with significantly reduced grid size (e.g. a few hundred)?