[petsc-dev] DMPlex redistribution

[Matthew Knepley]

On Mon, Mar 27, 2023 at 8:52 PM Adrian Croucher <a.croucher at auckland.ac.nz>
wrote:

> hi Matt,
> On 28/03/23 12:16, Matthew Knepley wrote:
>
>
> Yes, that would cause this. Your routine looks right to me, but it should
> be easy to find a problem. Put a check
> in at 3460 that h != dim, and also check that the height of minc_leaves(i)
> is not 0.
>
> Thanks, I just tried that, and I am getting lots of both of those (h ==
> dim, and height of minc_leaves(i) == 0). But I'm a little unclear about the
> significance of them.
>
> Okay, that means that you have created overlap before this.

> In your original message you said:
>
> I do not expect cells that are shared until we create an overlap.
>
> What I'm doing is distributing the DM with overlap 1, then adding the MINC
> cells etc. if needed, then redistributing (again with overlap 1). It seems
> to have worked ok but could the fact that it's already been distributed
> with overlap 1 be causing the error with redistribution?
>
> Yes, I think so.  The partitioning groups cells, and does not expect cells
in the overlap. We could generalize that, which is probably
the right thing to do when redistributing meshes with overlap. I will look
at why it is complaining. Maybe I can just put a guard instead
of an error in there.

  Thanks,

    Matt

> - Adrian
>
> --
> Dr Adrian Croucher
> Senior Research Fellow
> Department of Engineering Science
> University of Auckland, New Zealand
> email: a.croucher at auckland.ac.nz
> tel: +64 (0)9 923 4611
>
>

-- 
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

https://www.cse.buffalo.edu/~knepley/ <http://www.cse.buffalo.edu/~knepley/>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20230327/4423612c/attachment.html>