Barry and Jed,
 
Nice to know that! Is there a corresponding function call to set RCM?
 
Thanks,
Qin



On Friday, February 14, 2014 10:20 AM, Jed Brown <[email protected]> wrote:
  
Qin Lu <[email protected]> writes:


> Hello,
>  
> Is the RCM reordering available for ILU preconditioner in PETSc? The manual 
> says RCM works with LU factorization, but it does not mention it for 
> preconditioners. 

It works for all factorization preconditioners (ILU, ICC, LU, Cholesky).

~/petsc/src/ksp/ksp/examples/tutorials$ ./ex2 -m 100 -n 100 -ksp_view -pc_type 
ilu -pc_factor_mat_ordering_type rcm
KSP Object: 1 MPI processes
  type: gmres
    GMRES: restart=30, using Classical (unmodified) Gram-Schmidt 
Orthogonalization with no iterative refinement
    GMRES: happy breakdown tolerance 1e-30
  maximum iterations=10000, initial guess is zero
  tolerances:  relative=9.80296e-07, absolute=1e-50, divergence=10000
  left preconditioning
  using PRECONDITIONED norm type for convergence test
PC Object: 1 MPI processes
  type: ilu
    ILU: out-of-place factorization
    0 levels of fill
    tolerance for zero pivot 2.22045e-14
    using diagonal shift on blocks to prevent zero pivot [INBLOCKS]
    matrix ordering: rcm
    factor fill ratio given 1, needed 1
      Factored matrix follows:
        Mat Object:         1 MPI processes
          type: seqaij
          rows=10000, cols=10000
          package used to perform factorization: petsc
          total: nonzeros=49600, allocated nonzeros=49600
          total number of mallocs used during MatSetValues calls =0
            not using I-node routines
  linear system matrix = precond matrix:
  Mat Object:   1 MPI processes
    type: seqaij
    rows=10000, cols=10000
    total: nonzeros=49600, allocated nonzeros=50000
    total number of mallocs used during MatSetValues calls =0
      not using I-node routines
Norm of error 0.00314611 iterations 84

Reply via email to