[petsc-users] Assembling primal Schur matrix in FETI-DP method

Jed Brown jedbrown at mcs.anl.gov
Mon Nov 21 09:26:58 CST 2011


On Mon, Nov 21, 2011 at 09:19, Barry Smith <bsmith at mcs.anl.gov> wrote:

> Note there is never a reason to use KSP to do the solve because for the
> many solves needed a direct solver will always win over using an iterative
> solver
>

Depends. If multigrid works well on the subdomain problems, then it will
definitely win for large subdomains. (Or you might not even be able to
afford that amount of memory.)

Yes, you could use smaller subdomains, but most people only write FETI-DP
as a 2-level method and the size of the coarse space is a serious problem.


> and since the result is dense it doesn't make sense to do this computation
> with huge matrices.
>

But you are only solving against a few right hand sides. Between 4 and 200,
for example, while the subdomain size might be more than 100k.

If you want to use an external LU solver (it will not be faster than
> PETSc's so why bother).
>

This must be a joke. ;-)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20111121/837a65de/attachment.htm>


More information about the petsc-users mailing list