[petsc-users] parallel complex preconditioner needed

[Craig Tanis]

This is a frequency domain EM simulation.  

I think my main problem is interpreting the options available..  PETSc apparently has a built-in direct solver, which is how ILU works on a single processor, right? 

Are the parallel direct solvers in some 3rd party package,like hypre, that I need to install separately?  Do I somehow specify a direct solver to use for the built-in ILU ?

thanks,
-craig


On Jul 20, 2010, at 6:30 AM, Jed Brown wrote:

> On Mon, 19 Jul 2010 14:50:03 -0400, Craig Tanis <craig at tanistech.com> wrote:
>> I'm looking for a parallel preconditioner for a complex linear system.
>> For smaller serial runs, the built-in ILU has been great.
>> 
>> I've used hypre (euclid) on real matrices, but that suite is
>> apparently only for reals.  What are my options for these big complex
>> systems?
> 
> What problem are you solving?  Many of the direct solvers work with
> complex scalars, the algebraic multigrid packages are currently lacking
> in this regard.  If the problem is sufficiently small, you can use a
> parallel direct solver, otherwise you'll need some sort of domain
> decomposition method (e.g. -pc_type asm).  If ILU is working well for
> you, direct subdomain solves are probably not important.  You may still
> need a coarse level, there are many ways to do this, but they are
> problem dependent.
> 
> Jed