> I am facing the doing of parallel computation for some special governing
> equation, however mainly based on stokes one. I am still getting in
> problem for it. I have been modified step-17 for stokes equation with
> using the combination of GMRES+ILU in the Petsc Library. However,
> inherently , Petsc library can not support ILU preconditioning method in
> the parallel version.
That's true, but it supports what it calls a BJacobi preconditioner: a block
Jacobi version in which an ILU of the diagonal blocks of the matrix is
performed (as opposed to an ILU of the entire matrix). You should give that a
try.
> Could you give me a advice to deal with a parallel computation for stokes
> equation in deal.II.?
We're working on a parallel version of step-31 (which has as its central part
a Stokes solver). It's not ready yet, though, and may take a few more weeks.
The parallel part is going to be based on Trilinos rather than PETSc.
Take a look at the current state of step-31 in the subversion repository. If
you have experience with Trilinos, we would certainly appreciate any help!
Best
Wolfgang
-------------------------------------------------------------------------
Wolfgang Bangerth email: [EMAIL PROTECTED]
www: http://www.math.tamu.edu/~bangerth/
_______________________________________________
