[petsc-users] Parallel mesh partitioning

[Thomas Witkowski]

Zitat von Matthew Knepley <knepley at gmail.com>:

> On Fri, Jan 28, 2011 at 7:30 AM, Thomas Witkowski <
> thomas.witkowski at tu-dresden.de> wrote:
>
>> Does anybody of you nows a mesh/graph partitioner which works in parallel
>> and can guaranty to create only connected parts (subdomains)? I tried
>> ParMETiS and Zoltan. But both have no possibility to (optionally) ensure
>> this.
>>
>
> Why do you need connected partitions?

Because my FEM code does not support disconnected domain!


> Also, Zoltan uses ParMetis.

Yes and no. Zoltan can make use of ParMetis, but it has a lot of build  
in algorithms for geometric, graph and hypergraph-based partitioning.

Thomas

>
>   Thanks,
>
>      Matt
>
>
>> Regards,
>>
>> Thomas
>
> --
> 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
>