On Tue, Jun 2, 2015 at 4:00 PM, Justin Chang <[email protected]> wrote:
> MTH did not converge with the default -ksp_rtol 1e-6 so I had to raise the > tolerance to 1e-5 in order to get a solution. Attached are the outputs for > TH and MTH > This really looks like it was never solving this system. Matt > Last one with the svd did not work with the way the AMG PC was hard-coded > into FEniCS. Here's the list of preconditioners my installation of FEniCS > supports: > > Preconditioner | Description > --------------------------------------------------------------- > default | default preconditioner > ilu | Incomplete LU factorization > icc | Incomplete Cholesky factorization > sor | Successive over-relaxation > petsc_amg | PETSc algebraic multigrid > jacobi | Jacobi iteration > bjacobi | Block Jacobi iteration > additive_schwarz | Additive Schwarz > amg | Algebraic multigrid > hypre_amg | Hypre algebraic multigrid (BoomerAMG) > hypre_euclid | Hypre parallel incomplete LU factorization > hypre_parasails | Hypre parallel sparse approximate inverse > none | No preconditioner > > I would need to change a few things here and there if we really insisted > on seeing if svd works. > > On Tue, Jun 2, 2015 at 3:00 PM, Jed Brown <[email protected]> wrote: > >> Justin Chang <[email protected]> writes: >> >> > I originally solve that example problem using LU. >> >> I'd like to learn why LU didn't notice that the system is singular. >> (The checks are not reliable, but this case should be pretty obviously >> bad.) >> >> > But when I solve this one: >> > >> > >> http://fenicsproject.org/documentation/dolfin/1.5.0/python/demo/documented/stokes-iterative/python/documentation.html >> > >> > By simply running their code as is for TH and adding the one like I >> > mentioned for MTH, I get the following outputs when I pass in >> -ksp_monitor >> > -ksp_view and -log_summary >> >> Thanks, Justin. Could you please run with >> >> -ksp_type gmres -ksp_gmres_restart 1000 -ksp_monitor_true_residual >> -ksp_monitor_singular_value >> >> Also, can you run a tiny problem size (like less than 1000 dofs) with >> >> -pc_type svd -pc_svd_monitor >> >> Thanks. >> > > -- What most experimenters take for granted before they begin their experiments is infinitely more interesting than any results to which their experiments lead. -- Norbert Wiener
