[CIG-SHORT] Fwd: Using MUMPS solver in PETSc instead of ASM (no attachment)

[Brad Aagaard]

Sorry, my forwarding of a response from Tabrez got garbled. His response
is below.

Brad

Tabrez Ali wrote:
> I find that a parallel sparse direct solver like MUMPS is often fast
> enough. You just need to configure PETSC so that it installs it and then
> you can engage it by something like:
> 
> pylith --nodes=4 --petsc.ksp_type=preonly --petsc.pc_type=lu
> --petsc.pc_factor_mat_solver_package=mumps