[petsc-users] Rectangular matrices turned into square?

[Barry Smith]

 Are you talking about the below?  If so, I suspect that the code is actually generating a square matrix, not a rectangular one. Hypre definitely cannot handle rectangular matrices.

   Barry



PC Object: 2 MPI processes
  type: hypre
    HYPRE BoomerAMG preconditioning
    HYPRE BoomerAMG: Cycle type V
    HYPRE BoomerAMG: Maximum number of levels 25
    HYPRE BoomerAMG: Maximum number of iterations PER hypre call 1
    HYPRE BoomerAMG: Convergence tolerance PER hypre call 0.
    HYPRE BoomerAMG: Threshold for strong coupling 0.25
    HYPRE BoomerAMG: Interpolation truncation factor 0.
    HYPRE BoomerAMG: Interpolation: max elements per row 0
    HYPRE BoomerAMG: Number of levels of aggressive coarsening 0
    HYPRE BoomerAMG: Number of paths for aggressive coarsening 1
    HYPRE BoomerAMG: Maximum row sums 0.9
    HYPRE BoomerAMG: Sweeps down         1
    HYPRE BoomerAMG: Sweeps up           1
    HYPRE BoomerAMG: Sweeps on coarse    1
    HYPRE BoomerAMG: Relax down          symmetric-SOR/Jacobi
    HYPRE BoomerAMG: Relax up            symmetric-SOR/Jacobi
    HYPRE BoomerAMG: Relax on coarse     Gaussian-elimination
    HYPRE BoomerAMG: Relax weight  (all)      1.
    HYPRE BoomerAMG: Outer relax weight (all) 1.
    HYPRE BoomerAMG: Using CF-relaxation
    HYPRE BoomerAMG: Not using more complex smoothers.
    HYPRE BoomerAMG: Measure type        local
    HYPRE BoomerAMG: Coarsen type        Falgout
    HYPRE BoomerAMG: Interpolation type  classical
  linear system matrix = precond matrix:
  Mat Object:   2 MPI processes
    type: mpiaij
    rows=2112, cols=2112, bs=2
    total: nonzeros=29056, allocated nonzeros=29056
    total number of mallocs used during MatSetValues calls =0
      using I-node (on process 0) routines: found 531 nodes, limit used is 5

> On Jul 12, 2017, at 2:55 PM, Adam Denchfield <adenchfi at hawk.iit.edu> wrote:
> 
> Hi,
> Below I've attached the text file with my ksp_view. It's a time-dependent problem so I included ten time-steps. The information doesn't seem to be much different in the multiple time steps.
> 
> ______________
> 
> 
> 
> Regards,
> Adam Denchfield
> Peer Career Coach - Career Services
> Illinois Institute of Technology
> Bachelors of Science in Applied Physics (2018)
> Email: adenchfi at hawk.iit.edu
> My LinkedIn My ResearchGate Profile
> 
> On Wed, Jul 12, 2017 at 2:48 PM, Barry Smith <bsmith at mcs.anl.gov> wrote:
> 
>   Adam,
> 
>     Please send the output from KSP view. We don't "change" the matrix but we might have some intermediate matrix of that other side, or it could be a bug in our output.
> 
>    Barry
> 
> > On Jul 12, 2017, at 2:45 PM, Adam Denchfield <adenchfi at hawk.iit.edu> wrote:
> >
> > Hi,
> >
> > I'm using the FEniCS package, which uses DOLFIN/petsc4py (and thus petsc). If I'm understanding the libraries right, I'm assembling rectangular matrices and sending them to be solved in a KrylovSolver. However, with -ksp_view, I'm seeing 2112x2112 matrices, not 1062x2112 matrices. Does PETSc do conversions like this automatically during preconditioning or something? Trying to identify if this would be a behaviour on PETSc's part.
> >
> > Regards,
> > Adam Denchfield
> > Peer Career Coach - Career Services
> > Illinois Institute of Technology
> > Bachelors of Science in Applied Physics (2018)
> > Email: adenchfi at hawk.iit.edu
> > My LinkedIn My ResearchGate Profile
> 
> 
> <output_ksp.txt>