[petsc-dev] About parallel of ILU

[Ed Bueler]

At a practical level you may be looking for parallel ways to use
PETSc's native (but serial) ILU.  The simplest parallel extension is
processor-block Jacobi which applies ILU on the processor-owned degrees of
freedom but with no overlap:

-pc_type bjacobi -sub_pc_type ilu

The overlapping version is with ASM:

-pc_type asm -sub_pc_type ilu

where -pc_asm_overlap controls the overlap distance.  Finally one can use
the serial ILU as a smoother in geometric multigrid, for example in
combination with overlapping ASM:

-pc_type mg -mg_levels_pc_type asm -mg_levels_sub_pc_type ilu

Apologies if this was all assumed to be known in your original question.
Indeed, application of parallel direct solvers in PETSc often uses
hyper/mumps/superlu, though I'm not familiar with ILU in these packages.

Ed

Dear Professor,
> I'm writing about this mail about the ILU algorithm in PETSc. I have
> noticed that the ILU in PETSs cannot be paralleled, and the Hypre providing
> the parallel implement of ILU.
> My questions is : why the PETSc does not provide the parallel implement of
> ILU directly, instead of linking to the Hypre to obtain the Parallel ILU
> algorithm?
> PS: the attachment is the Parallel ILU algorithm in Hypre.
> I'm looking forward to hear you!
> Best,
> Gang Chen
> Sichuan Univertsity,
> School of Mathematics,
> Chengdu, China


-- 
Ed Bueler
Dept of Mathematics and Statistics
University of Alaska Fairbanks
Fairbanks, AK 99775-6660
306C Chapman
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20210112/995db8f0/attachment.html>