MatFeti

Barry Smith bsmith at mcs.anl.gov
Tue Sep 22 13:24:40 CDT 2009


    Yes, long dead code. It should be all removed but ideas could be  
used for new code.


    Barry

On Sep 22, 2009, at 1:15 PM, Dmitry Karpeev wrote:

> I've found something that looks like somewhat outdated code (e.g.,  
> it uses SLESxxx) for FETI-DP:
> src/mat/impls/feti
>
> It appears that this matrix class actually implements a FETI-DP  
> based preconditioner.
> Is that right?
> Perhaps some (or all) of it can be rolled into a more general  
> PCFieldSplit-type framework?
> Any ideas?
>
> Thanks.
> Dmitry.




More information about the petsc-dev mailing list