[dune-pdelab] (no subject)

[Steffen Müthing]

Hi Carlos,

> Am 23.07.2015 um 16:51 schrieb Carlos Cueto Camejo <carlos.cueto at ovgu.de>:
> 
> 
> Dear PDELab developer,
> 
> attached you can find a picture of computation I did using grid adaptivity in parallel with two processors. These is example07 from pdelab-howto rewritten in parallel. As you can see in the picture I get a refinement where the grid is partitioned for the parallel computation. This doesn't happen in the sequential computation and I would like to avoided. Just imagine how many more element I would get if instead of computing with 2 processors I do with 12. Does anyone knows why this happens and how to corrected?

the picture looks like your error estimator isn’t doing the right thing at the processor boundaries, causing excessive refinement.
You could try outputting the error estimator (just use a P0 space for it) to look at the per-cell values in ParaView.

It might be necessary to do some communication as part of the error estimator calculation, but that is just a guess…

Hope that helps,

Steffen

> 
> Best regards and thanks in advance,
> Carlos
> 
> <example07_parallel.jpg>_______________________________________________
> dune-pdelab mailing list
> dune-pdelab at dune-project.org
> http://lists.dune-project.org/mailman/listinfo/dune-pdelab

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 495 bytes
Desc: Message signed with OpenPGP using GPGMail
URL: <https://lists.dune-project.org/pipermail/dune-pdelab/attachments/20150724/872a414b/attachment.sig>