Hi freinds,
I have parallelized a code for a Thermoelastic problem based on step-40 .
Although the non-parallel code works for all the initial values and
boundary conditions, the parallel one fails giving this error :
"PETSC ERROR: Caught signal number 11 SEGV: Segmentation Violation,
probably memory access out of range"
It seems from debugger backtracethat that error comes from the
PETScWrappers::SolverCG.
Surprisingly, the problem converges for few initial value or boundary
conditions (not for all problem conditions) when I change the
preconditioner from PreconditionAMG to PreconditionJacobi.
It would be appreciated if you could give me any clue why I have this
error. The following is my solve object:
template <int dim>
unsigned int
Solid<dim>::solve ()
{
LA::MPI::Vector
completely_distributed_solution (locally_owned_dofs,
mpi_communicator);
SolverControl solver_control (dof_handler.n_dofs(), 1e-12);
#ifdef USE_PETSC_LA
LA::SolverCG solver(solver_control, mpi_communicator);
#else
LA::SolverCG solver(solver_control);
#endif
LA::MPI::PreconditionAMG preconditioner;
LA::MPI::PreconditionAMG::AdditionalData data;
#ifdef USE_PETSC_LA
data.symmetric_operator = true;
//data.strong_threshold = 0.5;
#else
/* Trilinos defaults are good */
#endif
preconditioner.initialize(tangent_matrix, data);
// LA::MPI::PreconditionJacobi preconditioner(tangent_matrix);
solver.solve (tangent_matrix, completely_distributed_solution,
system_rhs,
preconditioner);
constraints.distribute (completely_distributed_solution);
solution_update = completely_distributed_solution;
return solver_control.last_step();
}
Thanks,
--
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.
