[petsc-users] Projection preconditioner as PCMG

Jed Brown jedbrown at mcs.anl.gov
Tue Mar 6 09:04:59 CST 2012


On Tue, Mar 6, 2012 at 08:50, Abdul Hanan Sheikh <hanangul12 at yahoo.co.uk>wrote:

> I had to implement the projection preconditioner (P = I - A*R*(A_H)^1 *P)
> along with an standard preconditioner (M).
> And I read in some thread of users list that multilevel preconditioner can better
> handled by PCMG too, thus
> I had tried to implement the later (M) as pre smoother, where as the
> earlier (P) as CGC in PCMG framework.
>
> 1. Does it make sense ?
> or the coarse grid correction of PCMG only does this part: R*(A_H)^1 *P
> ??
> 2. How can I adapt my projection preconditioner (P) as (P1 = I -
> A*R*(A_H)^1 *P + R*(A_H)^1 *P ) ?


It's still not clear exactly what you want the method to do. PCMG can do
additive or multiplicative cycles. For more general composition, see

http://www.mcs.anl.gov/petsc/petsc-current/docs/manualpages/PC/PCCOMPOSITE.html
http://www.mcs.anl.gov/petsc/petsc-current/docs/manualpages/PC/PCGALERKIN.html

I see the man pages are lacking details, the user's manual has more details.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20120306/daffdce1/attachment.htm>


More information about the petsc-users mailing list