Hi Julian, Well, it seems then that my lack of optimism in my suggestion was well founded. I've done a bit more digging and I think that this might be a bug in the library. I don't see the member AdditionalData <https://github.com/dealii/dealii/blob/master/include/deal.II/lac/trilinos_solver.h#L342> struct in the TrilinosWrappers::SolverBase class being initialised anywhere. Rather it appears that each Solver keeps it own AdditionalData <https://github.com/dealii/dealii/blob/master/source/lac/trilinos_solver.cc#L609> that is not reachable by the base class, and that stored in the base class reverts to the default values chosen in the constructor. I'll open up an issue on the GitHub repository. Would you like to submit a patch for the bug? I'd be happy to help you do so.
Best, J-P On Tuesday, May 30, 2017 at 7:33:13 AM UTC+2, Julian A wrote: > > If I do that I see the usual output from the deallog but not the output > from the solver library. I thought that it's convenient to see the output > to check if the solver picks up the correct settings. If I do the steps > manually (that dealii does to create the solver) and cast the objects > correctly I get the expected log from Aztec. > > On Tue 30. May 2017 at 07:10, Jean-Paul Pelteret <> wrote: > >> Hi Julian, >> >> I had a quick look in the source code to verify that this flag is being >> passed >> to the solver >> <https://github.com/dealii/dealii/blob/master/source/lac/trilinos_solver.cc#L468> >> >> (it is). I've not used this option before so I'm afraid I don't have many >> ideas. The only thing that comes to mind is that the following option, if >> set, may suppress the output: >> >> deallog.depth_console(0); >> >> If you do have this set, you could (presumably) raise the console depth >> either globally or only around where you use the solver in order to get >> minimal extra output. I'm not particularly confident that my suggestion >> will work since I didn't see the deal.II internal logstream being attached >> to the solver anywhere in the code. But its not much effort to try this out >> anyway. >> >> Regards, >> Jean-Paul >> >> >> On Friday, May 26, 2017 at 7:30:08 PM UTC+2, Julian A wrote: >>> >>> Hi, >>> >>> i'm using the following code to setup a solver >>> >>> namespace LA { >>> using namespace dealii::LinearAlgebraTrilinos; >>> } >>> >>> >>> SolverControl solver_control(100, 1e-12); >>> LA::SolverCG::AdditionalData cgdata; >>> cgdata.output_solver_details = true; >>> LA::SolverCG solver(solver_control, cgdata); >>> solver.solve(system_matrix, completely_distributed_solution, system_rhs, >>> prec); >>> >>> The solver works and outputs the correct solution. Although no output >>> from AztecOO does occur. I traced the passing of >>> cgdata.output_solver_details >>> with gdb and it gets handed over the the solver. >>> >>> When i run step-33 which uses AztecOO directly and not via the >>> interface, i receive the output that i expect. I also tested different >>> AztecOO solvers (GMRES) in my code part but that doesn't produce verbose >>> output either. >>> >>> Is this expected or suppressed somewhere? I'm basically using step-40 to >>> play around with different options. >>> >>> What are other options to retrieve or log the residual at each timestep? >>> >>> Regards >>> >> -- >> The deal.II project is located at http://www.dealii.org/ >> For mailing list/forum options, see >> https://groups.google.com/d/forum/dealii?hl=en >> --- >> You received this message because you are subscribed to a topic in the >> Google Groups "deal.II User Group" group. >> To unsubscribe from this topic, visit >> https://groups.google.com/d/topic/dealii/kiLFKj94ubc/unsubscribe. >> To unsubscribe from this group and all its topics, send an email to >> [email protected]. >> For more options, visit https://groups.google.com/d/optout. >> > -- The deal.II project is located at http://www.dealii.org/ For mailing list/forum options, see https://groups.google.com/d/forum/dealii?hl=en --- You received this message because you are subscribed to the Google Groups "deal.II User Group" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/d/optout.
