[petsc-users] Multi level DM Shell

[Matthew Knepley]

On Mon, Feb 3, 2014 at 4:13 PM, Luc Berger-Vergiat <lb2653 at columbia.edu>wrote:

> Would there be a better option than a DMShell to do define fieldsplits of
> a matrix already assembled?
> The matrix comes from an FEM code and has boundary conditions applied to
> it before it is passed to PETSc.
> A DMShell seems like a good option since it does not require any
> particular knowledge on the underlying mesh of the problem.
> But is there another DM type supporting the DMCreateSubDM function, can
> work on a matrix coming from an FEM unstructured quad mesh with BCs applied
> to it?
> Maybe DMPlex? though I am not sure that FEM meshes can be described as
> Hasse Diagram all the time especial quad meshes...
>

Yes, DMPlex can handle that, but that is not the point for DMCreateSubDM().
You could have a DMShell, but define your
layout with a PetscSection if you wanted. I assume "with BC applied" means
that constrained dofs are eliminated from the
system. If you need help setting up the PetscSection, I can walk you
through it.

   Matt


> Best,
> Luc
>
> On 3 févr. 2014, at 16:59, Matthew Knepley wrote:
>
> Yes
>
>   Matt
>
>
> On Mon, Feb 3, 2014 at 3:32 PM, Luc Berger-Vergiat <lb2653 at columbia.edu>wrote:
>
>> I'm fairly sure I know the answer to this question after looking up the
>> source code but is it possible to create sub-splits while using a DMShell?
>> To be more specific I have a matrix which I decompose into 4 fields and I
>> would like to do something like
>>
>> -ksp_type gmres
>> -pc_type fieldsplit
>> -pc_fieldsplit_type schur
>> -pc_fieldsplit_schur_factorization_type full
>> -pc_fieldsplit_schur_precondition self
>> -pc_fieldsplit_0_fields 0,1
>> -pc_fieldsplit_1_fields 2,3
>> ...
>>
>> Best,
>> Luc
>>
>
>
>
> --
> What most experimenters take for granted before they begin their
> experiments is infinitely more interesting than any results to which their
> experiments lead.
> -- Norbert Wiener
>
>
>


-- 
What most experimenters take for granted before they begin their
experiments is infinitely more interesting than any results to which their
experiments lead.
-- Norbert Wiener
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20140203/0c225078/attachment.html>