Homec4science
Diffusion akantu (features/terms-to-add-in-python) Browse

rAKA/src/model/#non_linear_solver_newton_raphson.cc#
5eb7cda6aae0features/terms-to-add…

Last Change
Raw File

#non_linear_solver_newton_raphson.cc#
View Options

/**
* @file non_linear_solver_newton_raphson.cc
*
* @author Nicolas Richart <nicolas.richart@epfl.ch>
*
* @date creation: Tue Sep 15 2015
* @date last modification: Wed Feb 21 2018
*
* @brief Implementation of the default NonLinearSolver
*
* @section LICENSE
*
* Copyright (©) 2015-2018 EPFL (Ecole Polytechnique Fédérale de Lausanne)
* Laboratory (LSMS - Laboratoire de Simulation en Mécanique des Solides)
*
* Akantu is free software: you can redistribute it and/or modify it under the
* terms of the GNU Lesser General Public License as published by the Free
* Software Foundation, either version 3 of the License, or (at your option) any
* later version.
*
* Akantu is distributed in the hope that it will be useful, but WITHOUT ANY
* WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR
* A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more
* details.
*
* You should have received a copy of the GNU Lesser General Public License
* along with Akantu. If not, see <http://www.gnu.org/licenses/>.
*
*/
/* -------------------------------------------------------------------------- */
#include "non_linear_solver_newton_raphson.hh"
#include "communicator.hh"
#include "dof_manager_default.hh"
#include "solver_callback.hh"
#include "sparse_solver_mumps.hh"
/* -------------------------------------------------------------------------- */
namespace akantu {
/* -------------------------------------------------------------------------- */
NonLinearSolverNewtonRaphson::NonLinearSolverNewtonRaphson(
DOFManagerDefault & dof_manager,
const NonLinearSolverType & non_linear_solver_type, const ID & id,
UInt memory_id)
: NonLinearSolver(dof_manager, non_linear_solver_type, id, memory_id),
dof_manager(dof_manager),
solver(std::make_unique<SparseSolverMumps>(
dof_manager, "J", id + ":sparse_solver", memory_id)) {
this->supported_type.insert(_nls_newton_raphson_modified);
this->supported_type.insert(_nls_newton_raphson_contact);
this->supported_type.insert(_nls_newton_raphson);
this->supported_type.insert(_nls_linear);
this->checkIfTypeIsSupported();
this->registerParam("threshold", convergence_criteria, 1e-10, _pat_parsmod,
"Threshold to consider results as converged");
this->registerParam("convergence_type", convergence_criteria_type,
_scc_solution, _pat_parsmod,
"Type of convergence criteria");
this->registerParam("max_iterations", max_iterations, 10, _pat_parsmod,
"Max number of iterations");
this->registerParam("error", error, _pat_readable, "Last reached error");
this->registerParam("nb_iterations", n_iter, _pat_readable,
"Last reached number of iterations");
this->registerParam("converged", converged, _pat_readable,
"Did last solve converged");
this->registerParam("force_linear_recompute", force_linear_recompute, true,
_pat_modifiable,
"Force reassembly of the jacobian matrix");
}
/* -------------------------------------------------------------------------- */
NonLinearSolverNewtonRaphson::~NonLinearSolverNewtonRaphson() = default;
/* ------------------------------------------------------------------------ */
void NonLinearSolverNewtonRaphson::solve(SolverCallback & solver_callback) {
this->dof_manager.updateGlobalBlockedDofs();
solver_callback.predictor();
if (non_linear_solver_type == _nls_linear and
solver_callback.canSplitResidual())
solver_callback.assembleMatrix("K");
this->assembleResidual(solver_callback);
if (this->non_linear_solver_type == _nls_newton_raphson_modified ||
(this->non_linear_solver_type == _nls_linear &&
this->force_linear_recompute)) {
solver_callback.assembleMatrix("J");
this->force_linear_recompute = false;
}
this->n_iter = 0;
this->converged = false;
//this->dump();
if (this->convergence_criteria_type == _scc_residual and
this->non_linear_solver_type != _nls_newton_raphson_contact) {
this->converged = this->testConvergence(this->dof_manager.getResidual());
if (this->converged)
return;
}
do {
if (this->non_linear_solver_type == _nls_newton_raphson or
this->non_linear_solver_type == _nls_newton_raphson_contact)
solver_callback.assembleMatrix("J");
this->solver->solve();
solver_callback.corrector();
// EventManager::sendEvent(NonLinearSolver::AfterSparseSolve(method));
if (this->convergence_criteria_type == _scc_residual) {
this->assembleResidual(solver_callback);
this->converged = this->testConvergence(this->dof_manager.getResidual());
} else {
this->converged =
this->testConvergence(this->dof_manager.getGlobalSolution());
}
if (this->convergence_criteria_type == _scc_solution and
not this->converged)
this->assembleResidual(solver_callback);
//this->dump();
this->n_iter++;
AKANTU_DEBUG_INFO(
"[" << this->convergence_criteria_type << "] Convergence iteration "
<< std::setw(std::log10(this->max_iterations)) << this->n_iter
<< ": error " << this->error << (this->converged ? " < " : " > ")
<< this->convergence_criteria);
} while (not this->converged and this->n_iter < this->max_iterations);
// this makes sure that you have correct strains and stresses after the
// solveStep function (e.g., for dumping)
if (this->convergence_criteria_type == _scc_solution)
this->assembleResidual(solver_callback);
if (this->converged) {
// this->sendEvent(NonLinearSolver::ConvergedEvent(method));
} else if (this->n_iter == this->max_iterations) {
AKANTU_CUSTOM_EXCEPTION(debug::NLSNotConvergedException(
this->convergence_criteria, this->n_iter, this->error));
AKANTU_DEBUG_WARNING("[" << this->convergence_criteria_type
<< "] Convergence not reached after "
<< std::setw(std::log10(this->max_iterations))
<< this->n_iter << " iteration"
<< (this->n_iter == 1 ? "" : "s") << "!");
}
return;
}
/* -------------------------------------------------------------------------- */
bool NonLinearSolverNewtonRaphson::testConvergence(const Array<Real> & array) {
AKANTU_DEBUG_IN();
const Array<bool> & blocked_dofs = this->dof_manager.getGlobalBlockedDOFs();
UInt nb_degree_of_freedoms = array.size();
auto arr_it = array.begin();
auto bld_it = blocked_dofs.begin();
Real norm = 0.;
for (UInt n = 0; n < nb_degree_of_freedoms; ++n, ++arr_it, ++bld_it) {
bool is_local_node = this->dof_manager.isLocalOrMasterDOF(n);
if ((!*bld_it) && is_local_node) {
norm += *arr_it * *arr_it;
}
}
dof_manager.getCommunicator().allReduce(norm, SynchronizerOperation::_sum);
norm = std::sqrt(norm);
AKANTU_DEBUG_ASSERT(!Math::isnan(norm),
"Something went wrong in the solve phase");
this->error = norm;
return (error < this->convergence_criteria);
}
/* -------------------------------------------------------------------------- */
} // akantu
c4science · Help