--- /dev/null
+//---------------------------------------------------------------------------
+// $Id: no_flux_constraints.cc 24175 2011-08-24 12:14:21Z kronbichler $
+// Version: $Name$
+//
+// Copyright (C) 2009, 2010 by the deal.II authors
+//
+// This file is subject to QPL and may not be distributed
+// without copyright and license information. Please refer
+// to the file deal.II/doc/license.html for the text and
+// further information on this license.
+//
+//---------------------------------------------------------------------------
+
+
+// check that the ConstraintMatrix with hanging nodes and no-normal-flux
+// constraints on an adaptively refined hyper_cube are the same independet
+// of the number of CPUs
+
+#include "../tests.h"
+#include <deal.II/base/logstream.h>
+#include <deal.II/base/tensor.h>
+#include <deal.II/distributed/tria.h>
+#include <deal.II/grid/grid_generator.h>
+#include <deal.II/dofs/dof_handler.h>
+#include <deal.II/dofs/dof_tools.h>
+#include <deal.II/dofs/dof_renumbering.h>
+#include <deal.II/grid/tria_accessor.h>
+#include <deal.II/grid/tria_iterator.h>
+#include <deal.II/dofs/dof_accessor.h>
+#include <deal.II/distributed/tria.h>
+#include <deal.II/fe/fe_system.h>
+#include <deal.II/fe/fe_q.h>
+#include <deal.II/fe/mapping_q.h>
+#include <deal.II/lac/trilinos_vector.h>
+#include <deal.II/numerics/vectors.h>
+
+#include <fstream>
+#include <sstream>
+
+template<int dim>
+void test()
+{
+ Assert (dim == 3, ExcNotImplemented());
+ unsigned int myid = Utilities::System::get_this_mpi_process (MPI_COMM_WORLD);
+ unsigned int numprocs = Utilities::System::get_n_mpi_processes (MPI_COMM_WORLD);
+
+ parallel::distributed::Triangulation<dim> triangulation(MPI_COMM_WORLD);
+
+ GridGenerator::hyper_cube (triangulation, -1.0, 1.0);
+ triangulation.refine_global (3);
+
+ for (typename Triangulation<dim>::active_cell_iterator
+ cell = triangulation.begin_active();
+ cell != triangulation.end(); ++cell)
+ if (!cell->is_ghost() && !cell->is_artificial())
+ if (cell->center().norm() < 0.3)
+ {
+ cell->set_refine_flag();
+ }
+
+ triangulation.prepare_coarsening_and_refinement();
+ triangulation.execute_coarsening_and_refinement();
+
+ if (myid == 0)
+ deallog << "#cells = " << triangulation.n_global_active_cells()
+ << std::endl;
+
+ // create FE_System and fill in no-normal flux
+ // conditions on boundary 1 (outer)
+ static const FESystem<dim> fe(FE_Q<dim> (1), dim);
+ DoFHandler<dim> dofh(triangulation);
+ dofh.distribute_dofs (fe);
+ DoFRenumbering::hierarchical(dofh);
+
+ if (myid == 0)
+ deallog << "#dofs = " << dofh.locally_owned_dofs().size()
+ << std::endl;
+
+ IndexSet relevant_set;
+ DoFTools::extract_locally_relevant_dofs (dofh, relevant_set);
+
+ ConstraintMatrix constraints;
+ constraints.reinit(relevant_set);
+ DoFTools::make_hanging_node_constraints (dofh, constraints);
+ std::set<unsigned char> no_normal_flux_boundaries;
+ no_normal_flux_boundaries.insert (0);
+ const unsigned int degree = 1;
+ VectorTools::compute_no_normal_flux_constraints (dofh, 0,
+ no_normal_flux_boundaries,
+ constraints,
+ MappingQ<dim>(degree));
+ constraints.close();
+
+ if (myid==0)
+ system("rm -rf no_flux_constraints_03/cm_?.dot");
+
+ MPI_Barrier(MPI_COMM_WORLD);
+
+ { //write the constraintmatrix to a file on each cpu
+ char fname[] = "no_flux_constraints_03/cm_0.dot";
+ fname[26]+=myid;
+ std::ofstream file(fname);
+ constraints.print(file);
+ }
+ MPI_Barrier(MPI_COMM_WORLD);
+ sleep(1);
+ if (myid==0)
+ {
+ //sort and merge the constraint matrices on proc 0, generate a checksum
+ //and output that into the deallog
+ system("cat no_flux_constraints_03/cm_?.dot|sort -n|uniq >no_flux_constraints_03/cm");
+ system("md5sum no_flux_constraints_03/cm >no_flux_constraints_03/cm.check");
+ {
+ std::ifstream file("no_flux_constraints_03/cm.check");
+ std::string str;
+ while (!file.eof())
+ {
+ std::getline(file, str);
+ deallog << str << std::endl;
+ }
+ }
+ }
+
+ // print the number of constraints. since
+ // processors might write info in different
+ // orders, copy all numbers to root processor
+ std::vector<unsigned int> n_constraints_glob (numprocs);
+ unsigned int n_constraints = constraints.n_constraints();
+ MPI_Gather (&n_constraints, 1, MPI_UNSIGNED,
+ &n_constraints_glob[0], 1, MPI_UNSIGNED, 0, MPI_COMM_WORLD);
+ if (myid == 0)
+ for (unsigned int i=0; i<numprocs; ++i)
+ deallog << "#constraints on " << i << ": " << n_constraints_glob[i]
+ << std::endl;
+
+ // dummy assembly: put 1 in all components of
+ // the vector
+ TrilinosWrappers::MPI::Vector vector;
+ vector.reinit (dofh.locally_owned_dofs(), MPI_COMM_WORLD);
+ {
+ const unsigned int dofs_per_cell = fe.dofs_per_cell;
+ std::vector<unsigned int> local_dof_indices (dofs_per_cell);
+ Vector<double> local_vector (dofs_per_cell);
+ for (unsigned int i=0; i<dofs_per_cell; ++i)
+ local_vector(i) = 1.;
+ typename DoFHandler<dim>::active_cell_iterator
+ cell = dofh.begin_active(),
+ endc = dofh.end();
+ for (; cell!=endc; ++cell)
+ if (cell->subdomain_id() == triangulation.locally_owned_subdomain())
+ {
+ cell->get_dof_indices (local_dof_indices);
+ constraints.distribute_local_to_global (local_vector,
+ local_dof_indices,
+ vector);
+ }
+ vector.compress (Add);
+ }
+
+ // now check that no entries were generated
+ // for constrained entries on the locally
+ // owned range.
+ const std::pair<unsigned int,unsigned int> range = vector.local_range();
+ for (unsigned int i=range.first; i<range.second; ++i)
+ if (constraints.is_constrained(i))
+ Assert (vector(i)==0, ExcInternalError());
+
+ if (myid==0)
+ deallog << "OK" << std::endl;
+}
+
+
+int main(int argc, char *argv[])
+{
+ {
+ Utilities::System::MPI_InitFinalize mpi_init (argc, argv);
+ unsigned int myid = Utilities::System::get_this_mpi_process (MPI_COMM_WORLD);
+
+
+ deallog.push(Utilities::int_to_string(myid));
+
+ if (myid == 0)
+ {
+ std::ofstream logfile(output_file_for_mpi("no_flux_constraints_03").c_str());
+ deallog.attach(logfile);
+ deallog.depth_console(0);
+ deallog.threshold_double(1.e-10);
+
+ deallog.push("3d");
+ test<3>();
+ deallog.pop();
+ }
+ else
+ test<3>();
+ }
+
+ return 0;
+}