And one iteration of undamped Jacobi after the solve should fix this. On Wed, Jul 6, 2022 at 8:42 AM Karabelas, Elias (elias.karabe...@uni-graz.at) <elias.karabe...@uni-graz.at> wrote:
> Dear Matt, > > thanks for the fast response. That makes perfect sense to me. > > Best regards > Elias > > Am 06.07.22 um 14:35 schrieb Matthew Knepley: > > On Wed, Jul 6, 2022 at 7:46 AM Karabelas, Elias ( > elias.karabe...@uni-graz.at) <elias.karabe...@uni-graz.at> wrote: > >> Dear all, >> >> I don't know if this is a bug, but I observed that when using GMRES with >> AGG-PCGAMG as preconditioner Dirichlet boundary conditions don't seem to be >> exactly fulfilled. >> >> My Matrix has zero rows and cols with 1 on the diagonal where I have >> dirichlet-bcs in my FE-mesh and I would expect that the eqs in this rows >> can be exactly fulfilled (as u_i = g_i) there. >> > I would not expect aggregation to be exact here, but only within the > iteration tolerance. If instead you eliminate those variables, you can > maintain algebraic exactness. > This is what we do in examples, like SNES ex56. > > Thanks, > > Matt > >> However, when I solve A*x = b with the above solver I only get u_i = g_i >> + error in that part of the vector. Switching from pc_gamg_type agg to >> pc_gamg_type classical cures this problem, but the classical is not >> advertised in the user manual. >> >> These are the options I'm currently using: >> >> -ksp_type gmres >> -ksp_pc_side right >> -pc_type gamg >> -pc_gamg_type agg [or classical] >> -pc_gamg_sym_graph 1 >> -pc_gamg_square_graph 1 >> -pc_gamg_agg_nsmooths 1 >> -pc_gamg_threshold 0.01 >> -pc_mg_cycles v >> >> Iteration counts are basically the same. >> >> Best regards >> >> Elias >> >> -- >> Dr. Elias Karabelas >> Research Associate >> University of Graz >> Institute of Mathematics and Scientific Computing >> Heinrichstraße 36 >> A-8010 Graz >> Austria >> >> Phone: +43 316 380 8546 >> Email: elias.karabe...@uni-graz.at >> Web: https://ccl.medunigraz.at/ >> >> > > -- > 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 > > https://www.cse.buffalo.edu/~knepley/ > <http://www.cse.buffalo.edu/~knepley/> > > > -- > Dr. Elias Karabelas > Research Associate > University of Graz > Institute of Mathematics and Scientific Computing > Heinrichstraße 36 > A-8010 Graz > Austria > > Phone: +43 316 380 8546 > Email: elias.karabe...@uni-graz.at > Web: https://ccl.medunigraz.at/ > >
