dolfin team mailing list archive

[Jed Brown <jed@xxxxxxxx>]

On Wed 2008-09-24 19:36, Nuno David Lopes wrote:
> Also another doubt, (i'm guessing  that is my math ignorance only)
> is it usual that  PETSc::gmres with hypre::amg preconditioner doesn't work.
> (After reading that it is so powerfull...)
> In a Stokes 566000^2  subsystem it simply blows up with all of the  RAM memory 
> (16Gb).   
> (I tried it with simpler problems but with bigger systems   and it worked 
> perfectly, converging in <10 iterations.)

The Stokes problem is indefinite, hence normal preconditioners are not
expected to work.  Reordering unknowns and diagonal shifts can help, but
you really want a Schur complement preconditioner or Vanka smoothers for
multigrid (not conveniently available).  The easiest thing is to
assemble the matrix

  P1 = (A  B')         or         P2 = (A  0)
       (0  M )                         (0  M)

(where M is the mass matrix) to precondition the Stokes system

  (A  B') (u)  =  (f)
  (B  0 ) (p)     (0)

Unfortunately, Dolfin does not have the plumbing in place for separate
preconditioning matrices so you will need to work with the KSP directly.

Jed

Attachment: pgpmIkMxct5tl.pgp
Description: PGP signature