Dear all,
I am hitting an AssertDimension (partition_blocks[partition],
task_info.n_blocks); when initializing MatrixFree object for the finest
level of the mesh.
This happens at 2nd h-adaptive step with 3 a-priori global refinements.
The part of the code during setup of each level is rather standard GMG with
Laplace operator:
typename MatrixFree<dim,LevelNumberType>::AdditionalData
mg_additional_data;
mg_additional_data.tasks_parallel_scheme = MatrixFree<dim,
LevelNumberType>::AdditionalData::partition_color;
mg_additional_data.level_mg_handler = level;
mg_additional_data.mapping_update_flags = update_gradients |
update_JxW_values;
ConstraintMatrix level_constraints;
IndexSet relevant_dofs;
DoFTools::extract_locally_relevant_level_dofs(dof_handler, level,
relevant_dofs);
level_constraints.reinit(relevant_dofs);
level_constraints.add_lines(mg_constrained_dofs.get_boundary_indices
(level));
level_constraints.close();
std::shared_ptr<MatrixFree<dim,LevelNumberType>> mg_level_data(new
MatrixFree<dim,LevelNumberType>());
mg_level_data->reinit (mapping, dof_handler, level_constraints,
quadrature_formula,
mg_additional_data);
backtrace from this reinit is:
matrix_free.h:1668
1665 quad_hp.emplace_back (quad[q]);
1666 internal_reinit(StaticMappingQ1<dim>::mapping,
dof_handler,constraint,
1667 locally_owned_set, quad_hp, additional_data);
-> 1668 }
matrix_free.templates.h:166
163 // constraint_pool_data. It also reorders the way cells are
gone through
164 // (to separate cells with overlap to other processors from
others
165 // without).
-> 166 initialize_indices (constraint, locally_owned_set);
matrix_free.templates.h:689
686 (size_info, task_info, renumbering, irregular_cells,
687 dof_handlers.active_dof_handler == DoFHandlers::hp);
688 else
-> 689 dof_info[0].make_thread_graph_partition_color
690 (size_info, task_info, renumbering, irregular_cells,
691 dof_handlers.active_dof_handler == DoFHandlers::hp);
dof_info.templates.h:1011
1008 break;
1009 }
1010 }
-> 1011 AssertDimension (partition_blocks[partition],
task_info.n_blocks);
Any ideas?
Regards,
Denis.
--
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.
