Timo.
> If I remember correctly, only component_wise and custom (handwritten)
> reordering is supported for distributed meshes. Cuthill_mckee is a bad
> idea anyhow in parallel. It would create a SparsityPattern
> representing the coupling between the DoFs for all DoFs on every
> machine and then try to optimize the ordering.
Interesting that you should say so. That is exactly what I have found by
numerical experiments; I am glad your intuition "confirms" this.
In fact I find that leaving DoF numbering as it is (at least for
scalar valued equations) works best for distributed grids; which is
a very satisfying result for me :-)
Cheers.
Best,
Toby
--
Toby D. Young
Assistant Professor Philosophy & Physics
Polish Academy of Sciences
www: http://www.ippt.gov.pl/~tyoung
skype: stenografia
_______________________________________________
dealii mailing list http://poisson.dealii.org/mailman/listinfo/dealii
