Homec4science
Diffusion LibMultiScale (features/reshape_compute_python) Browse

libmultiscale/src/coupling/bridging.cc
d132a11abad7features/reshape_comput…

Last Change
Raw File

bridging.cc
View Options

/**
* @file bridging.cc
*
* @author Guillaume Anciaux <guillaume.anciaux@epfl.ch>
*
* @date Fri Jul 11 15:47:44 2014
*
* @brief Bridging object between atomistic and finite elements
*
* @section LICENSE
*
* Copyright INRIA and CEA
*
* The LibMultiScale is a C++ parallel framework for the multiscale
* coupling methods dedicated to material simulations. This framework
* provides an API which makes it possible to program coupled simulations
* and integration of already existing codes.
*
* This Project was initiated in a collaboration between INRIA Futurs Bordeaux
* within ScAlApplix team and CEA/DPTA Ile de France.
* The project is now continued at the Ecole Polytechnique Fédérale de Lausanne
* within the LSMS/ENAC laboratory.
*
* This software is governed by the CeCILL-C license under French law and
* abiding by the rules of distribution of free software. You can use,
* modify and/ or redistribute the software under the terms of the CeCILL-C
* license as circulated by CEA, CNRS and INRIA at the following URL
* "http://www.cecill.info".
*
* As a counterpart to the access to the source code and rights to copy,
* modify and redistribute granted by the license, users are provided only
* with a limited warranty and the software's author, the holder of the
* economic rights, and the successive licensors have only limited
* liability.
*
* In this respect, the user's attention is drawn to the risks associated
* with loading, using, modifying and/or developing or reproducing the
* software by the user in light of its specific status of free software,
* that may mean that it is complicated to manipulate, and that also
* therefore means that it is reserved for developers and experienced
* professionals having in-depth computer knowledge. Users are therefore
* encouraged to load and test the software's suitability as regards their
* requirements in conditions enabling the security of their systems and/or
* data to be ensured and, more generally, to use and operate it in the
* same conditions as regards security.
*
* The fact that you are presently reading this means that you have had
* knowledge of the CeCILL-C license and that you accept its terms.
*
*/
//#define TIMER
/* -------------------------------------------------------------------------- */
#include "bridging.hh"
#include "compute_extract.hh"
#include "factory_multiscale.hh"
#include "filter_geometry.hh"
#include "geometry_manager.hh"
#include "lib_bridging.hh"
#include "lm_common.hh"
#include "trace_atom.hh"
/* -------------------------------------------------------------------------- */
__BEGIN_LIBMULTISCALE__
template <typename ContainerPoints, typename ContainerMesh, UInt Dim>
Bridging<ContainerPoints, ContainerMesh, Dim>::Bridging(const std::string &name,
ContainerPoints &cP,
ContainerMesh &cM)
: LMObject(name), DofAssociation<ContainerPoints, ContainerMesh, Dim>(
name, cP, cM),
unmatchedPointList("unmatched-point-" + name),
pointList("matched-point-" + name),
unmatchedMeshList("unmatched-fe-" + name), meshList("matched-fe-" + name),
node_shape("node_shape-" + name), contPoints(cP), contMesh(cM) {
assoc_found = 0;
smatrix = NULL;
/* registering computes for outer world */
FilterManager::getManager().addObject(unmatchedMeshList);
FilterManager::getManager().addObject(unmatchedPointList);
FilterManager::getManager().addObject(pointList);
FilterManager::getManager().addObject(meshList);
}
/* -------------------------------------------------------------------------- */
template <typename ContainerPoints, typename ContainerMesh, UInt Dim>
Bridging<ContainerPoints, ContainerMesh, Dim>::~Bridging() {
if (smatrix)
delete smatrix;
smatrix = NULL;
}
/* -------------------------------------------------------------------------- */
template <typename ContainerPoints, typename ContainerMesh, UInt Dim>
void Bridging<ContainerPoints, ContainerMesh, Dim>::init() {
DUMPBYPROC("selecting DOFs in bridging zone", DBG_INFO_STARTUP, 0);
if (this->in_group_A) {
this->buildPointList();
this->buildPositions(unmatchedPointList);
}
if (this->in_group_B)
this->buildContinuumDOFsList();
DUMP(this->getID() << " : Generating Communication Scheme", DBG_INFO_STARTUP);
MPI_Barrier(MPI_COMM_WORLD);
DUMP(this->getID() << " : exchange coarse geometries", DBG_MESSAGE);
this->exchangeGeometries(this->unmatchedPointList, this->unmatchedMeshList);
MPI_Barrier(MPI_COMM_WORLD);
DUMP(this->getID() << " : exchange points positions", DBG_MESSAGE);
this->exchangePositions();
MPI_Barrier(MPI_COMM_WORLD);
DUMP(this->getID() << " : build shape matrix", DBG_MESSAGE);
if (this->in_group_B) {
DUMP(this->getID() << " : build shape matrix (parsing "
<< this->local_points << " points by position)",
DBG_INFO);
this->buildShapeMatrix(this->local_points);
this->local_points = this->assoc_found;
DUMP(this->getID() << " : I will manage " << this->local_points << " atoms",
DBG_INFO);
}
MPI_Barrier(MPI_COMM_WORLD);
DUMP(this->getID() << " : ExchangeAssociation", DBG_MESSAGE);
ContainerArray<UInt> managed;
this->exchangeAssociationInformation(managed);
if (this->in_group_A)
for (UInt i = 0; i < this->nb_zone_B; ++i)
DUMP("second test Do I communicate with " << i << " : "
<< this->com_with[i],
DBG_INFO);
if (this->in_group_B)
for (UInt i = 0; i < this->nb_zone_A; ++i)
DUMP("second test Do I communicate with " << i << " : "
<< this->com_with[i],
DBG_INFO);
MPI_Barrier(MPI_COMM_WORLD);
DUMP(this->getID()
<< " : generate duo vector and detect double atom assignement",
DBG_MESSAGE);
std::vector<std::vector<UInt>> unassociated;
if (this->in_group_A)
this->createDuoVectorA("Atom", managed, unassociated);
DUMP(this->getID() << " : doing exchange rejected", DBG_MESSAGE);
this->exchangeRejectedContinuumOwners(unassociated);
MPI_Barrier(MPI_COMM_WORLD);
DUMP(this->getID() << " : doing filter rejected", DBG_MESSAGE);
this->filterRejectedContinuumOwners(unassociated);
MPI_Barrier(MPI_COMM_WORLD);
DUMP(this->getID() << " : filtering position & association vector",
DBG_MESSAGE);
if (this->in_group_A) {
this->filterPointListForUnmatched();
}
if (this->in_group_B) {
this->filterArray(this->positions, Dim);
this->filterAssoc();
this->buildNodeShape();
}
this->deleteGrid();
DUMP(this->getID() << " : all done for scheme generation", DBG_MESSAGE);
MPI_Barrier(MPI_COMM_WORLD);
}
/* -------------------------------------------------------------------------- */
template <typename ContainerPoints, typename ContainerMesh, UInt Dim>
void Bridging<ContainerPoints, ContainerMesh, Dim>::buildContinuumDOFsList() {
unmatchedMeshList.setParam("GEOMETRY", this->geomID);
unmatchedMeshList.buildManual(contMesh);
DUMP("we have found " << unmatchedMeshList->getContainerElems().size()
<< " concerned elements",
DBG_INFO_STARTUP);
STARTTIMER("Filling spatial-grid");
auto contElems = unmatchedMeshList->getContainerElems();
auto contNodes = unmatchedMeshList->getContainerNodes();
std::vector<UInt> nodes;
Geometry &geom = *GeometryManager::getManager().getGeometry(this->geomID);
Cube cube = geom.getBoundingBox();
DUMP("geometry of the grid => \n" << cube, DBG_INFO_STARTUP);
if (this->grid)
this->deleteGrid();
this->grid = new SpatialGridElem<UInt, Dim>(cube, this->grid_division);
UInt nb_elem = 0;
for (auto &&el : contElems) {
std::vector<Real> node_coords;
auto &&nodes = el.globalIndexes();
UInt nb = nodes.size();
for (UInt i = 0; i < nb; ++i) {
RefNode nd = el.getNode(i);
#ifndef LM_OPTIMIZED
Real mass_node = nd.mass();
#endif
LM_ASSERT(mass_node > 0, "invalid mass" << nodes[i] << " " << mass_node);
auto X = nd.position0();
for (UInt i = 0; i < Dim; ++i)
node_coords.push_back(X[i]);
}
this->grid->addFiniteElement(nb_elem, node_coords);
++nb_elem;
}
STOPTIMER("Filling spatial-grid");
DUMP("we have found " << nb_elem << " concerned elements", DBG_INFO_STARTUP);
DUMP("in average " << this->grid->getAverageEltByBlock()
<< " elements par block",
DBG_INFO_STARTUP);
}
/* -------------------------------------------------------------------------- */
template <typename ContainerPoints, typename ContainerMesh, UInt Dim>
void Bridging<ContainerPoints, ContainerMesh, Dim>::buildNodeList() {
meshList.computeAlteredConnectivity(contMesh.getContainerNodes());
UInt nb = meshList.size();
if (!nb) {
DUMP("We found no nodes in bridging zone", DBG_INFO_STARTUP);
return;
}
DUMP("We found " << nb << " nodes concerned into "
<< meshList.getContainerElems().size() << " elements",
DBG_INFO_STARTUP);
}
/* -------------------------------------------------------------------------- */
template <typename ContainerPoints, typename ContainerMesh, UInt Dim>
void Bridging<ContainerPoints, ContainerMesh, Dim>::buildPointList() {
unmatchedPointList.setParam("GEOMETRY", this->geomID);
unmatchedPointList.buildManual(contPoints);
this->local_points = unmatchedPointList->size();
if (!this->local_points)
DUMP("We found no atom in the bridging zone", DBG_INFO);
DUMP("We found " << this->local_points << " atom in the bridging zone",
DBG_INFO_STARTUP);
}
/* -------------------------------------------------------------------------- */
template <typename ContainerPoints, typename ContainerMesh, UInt Dim>
void Bridging<ContainerPoints, ContainerMesh, Dim>::buildShapeMatrix(
UInt nb_atoms) {
this->assoc.assign(nb_atoms, UINT_MAX);
if (unmatchedMeshList->getContainerElems().size() == 0) {
DUMP("elem_rec is empty!", DBG_WARNING);
// delete grid;
return;
}
std::vector<UInt> nb_atoms_per_element;
nb_atoms_per_element.resize(unmatchedMeshList->getContainerElems().size());
/* construction du array d'assoc atoms elem */
STARTTIMER("Construction association");
for (UInt i = 0; i < nb_atoms; ++i) {
#ifndef TIMER
if (i % 100000 == 0)
DUMP("construction de l'association - atome " << i << "/" << nb_atoms,
DBG_INFO);
#endif
LM_ASSERT(i * Dim < this->positions.size(), "overflow detected");
Vector<Dim> x;
for (UInt k = 0; k < Dim; ++k)
x[k] = this->positions[i * Dim + k];
std::vector<UInt> &subset_elts = this->grid->findSet(x);
std::vector<UInt>::iterator it = subset_elts.begin();
for (it = subset_elts.begin(); it != subset_elts.end(); ++it) {
RefElem el = unmatchedMeshList->getContainerElems().get(*it);
UInt j = *it;
DUMP("is atom " << i << " within element " << j << "?", DBG_ALL);
if (el.contains(x)) {
LM_ASSERT(this->assoc[i] == UINT_MAX,
"this should not happen. "
<< "I found twice the same atom while filtering");
DUMP("associating atom " << i << " and element " << j, DBG_ALL);
this->assoc[i] = j;
++assoc_found;
break;
}
}
if (it == subset_elts.end()) {
this->assoc[i] = UINT_MAX;
}
}
STOPTIMER("Construction association");
/* build number of atoms per element */
for (UInt i = 0; i < nb_atoms; ++i) {
if (this->assoc[i] != UINT_MAX)
++nb_atoms_per_element[this->assoc[i]];
}
/* filter element list and node list for the unassociated node and elements */
filterContainerElems(nb_atoms_per_element);
buildNodeList();
smatrix = new ShapeMatrix<Dim>(meshList.getContainerElems().size());
UInt i = 0;
for (auto &&el : meshList.getContainerElems()) {
DUMP("declare sub matrix number " << i << " which will be of size "
<< nb_atoms_per_element[i] << "x"
<< el.nNodes(),
DBG_ALL);
LM_ASSERT(nb_atoms_per_element[i] > 0, "this should not happen any more");
// if (nb_atoms_per_element[i]>0)
smatrix->setSub(i, nb_atoms_per_element[i], el.nNodes());
++i;
}
// Allocate the shapematrix structure (contiguous)
smatrix->allocate();
/* set the indirection (alloc tab) plus the shapes */
std::vector<Real> shapes;
UInt atom_index = 0;
for (i = 0; i < nb_atoms; ++i) {
if (this->assoc[i] == UINT_MAX)
continue;
Real X[3] = {0.0, 0.0, 0.0};
for (UInt k = 0; k < Dim; ++k)
X[k] = this->positions[i * Dim + k];
auto el = meshList.getContainerElems().get(this->assoc[i]);
el.computeShapes(shapes, X);
auto &&nodes = el.globalIndexes();
IF_TRACED(X, "checking for atom in trouble");
for (UInt j = 0; j < shapes.size(); ++j)
smatrix->fill(atom_index, nodes[j], meshList.subIndex2Index(nodes[j]),
this->assoc[i], shapes[j]);
++atom_index;
}
}
/* -------------------------------------------------------------------------- */
template <typename ContainerPoints, typename ContainerMesh, UInt Dim>
void Bridging<ContainerPoints, ContainerMesh, Dim>::buildLocalPBCPairs() {
UInt npairs = pbc_pairs.size();
for (UInt pair = 0; pair < npairs; ++pair) {
UInt ind1 = pbc_pairs[pair].first;
UInt ind2 = pbc_pairs[pair].second;
UInt i1 = meshList.subIndex2Index(ind1);
UInt i2 = meshList.subIndex2Index(ind2);
if (i1 == UINT_MAX || i2 == UINT_MAX)
continue;
std::pair<UInt, UInt> p((UInt)(i1), (UInt)(i2));
local_pbc_pairs.push_back(p);
}
DUMP("Detected " << local_pbc_pairs.size() << " local pairs", DBG_MESSAGE);
smatrix->alterMatrixIndexesForPBC(local_pbc_pairs);
}
/* -------------------------------------------------------------------------- */
template <typename ContainerPoints, typename ContainerMesh, UInt Dim>
void Bridging<ContainerPoints, ContainerMesh, Dim>::buildNodeShape() {
if (this->local_points == 0)
return;
UInt nb_coupled_nodes = meshList.size();
node_shape.assign(nb_coupled_nodes, 0);
LM_ASSERT(smatrix, "shapematrix object was not allocated "
<< " eventhough local atoms were detected");
smatrix->buildNodeShape(node_shape);
buildLocalPBCPairs();
cumulPBC(node_shape);
#ifndef LM_OPTIMIZED
for (UInt i = 0; i < nb_coupled_nodes; ++i) {
if (node_shape[i] <= 1e-1) {
DUMP(i << " " << node_shape[i], DBG_MESSAGE);
LM_FATAL("Please check your geometry:"
<< " one or more nodes has zero nodal shape values."
<< " Changing the FE element size may resolve this issue!"
<< std::endl);
}
}
#endif // LM_OPTIMIZED
FilterManager::getManager().addObject(node_shape);
}
/* -------------------------------------------------------------------------- */
template <typename ContainerPoints, typename ContainerMesh, UInt Dim>
void Bridging<ContainerPoints, ContainerMesh, Dim>::filterContainerElems(
std::vector<UInt> &nb_atome_par_element) {
typename ContainerMeshSubset::ContainerElems new_t(this->getID() + ":elems");
std::vector<int> new_index;
for (UInt i = 0; i < nb_atome_par_element.size(); ++i) {
DUMP("filtering old index element " << i, DBG_ALL);
if (nb_atome_par_element[i] > 0) {
DUMP("for container, elem " << i << " becomes " << new_t.size(), DBG_ALL);
new_index.push_back(new_t.size());
DUMP("maintenant nb_atome_par_element["
<< new_t.size() << "]= " << nb_atome_par_element[i]
<< " (i=" << i << ")",
DBG_ALL);
nb_atome_par_element[new_t.size()] = nb_atome_par_element[i];
new_t.push_back(unmatchedMeshList->getContainerElems().get(i));
} else
new_index.push_back(new_t.size());
}
nb_atome_par_element.resize(new_t.size());
meshList.clear();
// rebuild the element container
for (UInt i = 0; i < new_t.size(); ++i)
meshList.getContainerElems().push_back(new_t.get(i));
// rebuild the association list
for (UInt i = 0; i < this->assoc.size(); ++i) {
if (this->assoc[i] != UINT_MAX) {
this->assoc[i] = new_index[this->assoc[i]];
}
}
}
/* -------------------------------------------------------------------------- */
template <typename ContainerPoints, typename ContainerMesh, UInt Dim>
void Bridging<ContainerPoints, ContainerMesh,
Dim>::filterPointListForUnmatched() {
pointList.clear();
for (UInt i = 0; i < this->assoc.size(); ++i)
if (this->assoc[i] != UINT_MAX) {
DUMP("for container, atom " << i << " becomes " << pointList.size(),
DBG_ALL);
pointList.push_back(unmatchedPointList->get(i));
} else {
DUMP("atom filtered " << i << " " << unmatchedPointList->get(i),
DBG_DETAIL);
}
pointList.copyContainerInfo(unmatchedPointList);
}
/* -------------------------------------------------------------------------- */
template <typename ContainerPoints, typename ContainerMesh, UInt Dim>
void Bridging<ContainerPoints, ContainerMesh, Dim>::filterArray(
ContainerArray<Real> &array, UInt stride) {
UInt index = 0;
for (UInt i = 0; i < this->assoc.size(); ++i)
if (this->assoc[i] != UINT_MAX) {
DUMP("for container, atom " << i << " becomes " << index, DBG_ALL);
LM_ASSERT(i < array.size() && index < array.size(),
"overflow detected: nmax = " << array.size() << " , index = "
<< index << " and i = " << i);
for (UInt k = 0; k < stride; ++k)
array[stride * index + k] = array[stride * i + k];
++index;
}
}
/* -------------------------------------------------------------------------- */
template <typename ContainerPoints, typename ContainerMesh, UInt Dim>
void Bridging<ContainerPoints, ContainerMesh, Dim>::allocateBuffer(
UInt size, UInt stride) {
if (size > buffer.size()) {
buffer.assign(stride * size, 0);
DUMP("allocating buffer of size " << size * stride, DBG_INFO);
}
}
/* -------------------------------------------------------------------------- */
template <typename ContainerPoints, typename ContainerMesh, UInt Dim>
void Bridging<ContainerPoints, ContainerMesh, Dim>::interpolatePointData(
std::vector<Real> &data_atom, std::vector<Real> &data_node, UInt stride) {
if (this->local_points == 0) {
DUMP("Warning : atomic container is empty:"
<< " cannot interpolate mesh fields on control points (check "
"geometries ?!)",
DBG_WARNING);
return;
}
smatrix->interpolate(data_atom, data_node, stride);
}
/* -------------------------------------------------------------------------- */
template <typename ContainerPoints, typename ContainerMesh, UInt Dim>
void Bridging<ContainerPoints, ContainerMesh, Dim>::cumulPBC(
ContainerArray<Real> &data, UInt stride) {
UInt npairs = local_pbc_pairs.size();
DUMP("Detected " << npairs << " local pairs", DBG_INFO);
for (UInt pair = 0; pair < npairs; ++pair) {
UInt i1 = local_pbc_pairs[pair].first;
UInt i2 = local_pbc_pairs[pair].second;
for (UInt k = 0; k < stride; ++k) {
data[i2 * stride + k] += data[stride * i1 + k];
DUMP("cumul pbc : [" << i2 * stride + k << "] = " << data[i2 * stride + k]
<< " += [" << i2 * stride + k
<< "] = " << data[stride * i1 + k] << " := "
<< data[stride * i1 + k] + data[stride * i2 + k],
DBG_DETAIL);
}
}
for (UInt pair = 0; pair < npairs; ++pair) {
UInt i1 = local_pbc_pairs[pair].first;
UInt i2 = local_pbc_pairs[pair].second;
for (UInt k = 0; k < stride; ++k)
data[stride * i1 + k] = data[stride * i2 + k];
}
}
/* -------------------------------------------------------------------------- */
template <typename ContainerPoints, typename ContainerMesh, UInt Dim>
void Bridging<ContainerPoints, ContainerMesh, Dim>::leastSquarePointData(
std::vector<Real> &v_out, std::vector<Real> &data_atom, UInt stride,
UInt nmax) {
LM_TOIMPLEMENT;
}
/* -------------------------------------------------------------------------- */
template <typename ContainerPoints, typename ContainerMesh, UInt Dim>
void Bridging<ContainerPoints, ContainerMesh, Dim>::solveLeastSquare(
ContainerArray<Real> &mesh_data, ContainerArray<Real> &atomic_data,
UInt stride) {
// build right hand side of leastsquare system
mesh_data.assign(stride * meshList.size(), 0);
smatrix->buildLeastSquareRHS(mesh_data, atomic_data, stride);
cumulPBC(mesh_data, stride);
LM_ASSERT(node_shape.size() != 0,
"node_shape is empty this should not happen, check geometry");
// solve least square system
for (UInt i = 0; i < meshList.size(); ++i)
for (UInt k = 0; k < stride; ++k) {
DUMP(mesh_data[i * stride + k]
<< " / " << node_shape[i] << " = "
<< mesh_data[i * stride + k] / node_shape[i],
DBG_DETAIL);
mesh_data[i * stride + k] /= node_shape[i];
}
}
/* -------------------------------------------------------------------------- */
template <typename ContainerPoints, typename ContainerMesh, UInt Dim>
void Bridging<ContainerPoints, ContainerMesh, Dim>::averageOnElements(
ContainerArray<Real> &data_atomic, ContainerArray<Real> &data_mesh,
UInt stride) {
data_mesh.assign(stride * meshList.getContainerElems().size(), 0);
smatrix->averageOnElements(data_atomic, data_mesh, stride);
}
/* -------------------------------------------------------------------------- */
template <typename ContainerPoints, typename ContainerMesh, UInt Dim>
void Bridging<ContainerPoints, ContainerMesh, Dim>::clearAll() {
if (smatrix)
delete smatrix;
smatrix = NULL;
meshList.getContainerElems().clear();
// nodeList.clear();
assoc_found = 0;
this->assoc.clear();
pointList.clear();
}
/* -------------------------------------------------------------------------- */
template <typename ContainerPoints, typename ContainerMesh, UInt Dim>
void Bridging<ContainerPoints, ContainerMesh, Dim>::buildLeastSquareMatrix(
math::Matrix &mat) {
smatrix->buildLeastSquareMatrix(&mat);
cumulPBC(mat);
}
/* -------------------------------------------------------------------------- */
template <typename ContainerPoints, typename ContainerMesh, UInt Dim>
void Bridging<ContainerPoints, ContainerMesh, Dim>::cumulPBC(
math::Matrix &mat) {
UInt npairs = local_pbc_pairs.size();
for (UInt pair = 0; pair < npairs; ++pair) {
UInt i1 = local_pbc_pairs[pair].first;
UInt i2 = local_pbc_pairs[pair].second;
for (UInt j = 0; j < mat.n(); ++j) {
mat(i2, j) += mat(i1, j);
}
}
for (UInt pair = 0; pair < npairs; ++pair) {
UInt i1 = local_pbc_pairs[pair].first;
for (UInt j = 0; j < mat.n(); ++j) {
if (i1 == j)
mat(i1, j) = 1.;
else
mat(i1, j) = 0.;
}
}
}
/* -------------------------------------------------------------------------- */
template <typename ContainerPoints, typename ContainerMesh, UInt Dim>
void Bridging<ContainerPoints, ContainerMesh, Dim>::copySlaveValues(
std::vector<Real> &v, UInt stride) {
UInt npairs = local_pbc_pairs.size();
for (UInt pair = 0; pair < npairs; ++pair) {
UInt i1 = local_pbc_pairs[pair].first;
UInt i2 = local_pbc_pairs[pair].second;
for (UInt k = 0; k < stride; ++k) {
v[i1 * stride + k] = v[i2 * stride + k];
}
}
}
/* -------------------------------------------------------------------------- */
template <typename ContainerPoints, typename ContainerMesh, UInt Dim>
void Bridging<ContainerPoints, ContainerMesh, Dim>::extractShapeMatrix(
math::Matrix &mat) {
smatrix->extractShapeMatrix(&mat);
}
/* -------------------------------------------------------------------------- */
/* Parallel Methods */
/* -------------------------------------------------------------------------- */
template <typename DomainA, typename DomainC, UInt Dim>
void Bridging<DomainA, DomainC, Dim>::communicateBufferAtom2Continuum(
UInt stride, ContainerArray<Real> &buf) {
if (this->local_points != 0) {
this->distributeVectorA2B("communicate buffer", buf, stride);
}
}
/* -------------------------------------------------------------------------- */
template <typename DomainA, typename DomainC, UInt Dim>
void Bridging<DomainA, DomainC, Dim>::communicateBufferContinuum2Atom(
UInt stride, ContainerArray<Real> &buf) {
if (this->local_points != 0) {
this->distributeVectorB2A("communicate buffer", buf, stride);
}
}
/* -------------------------------------------------------------------------- */
template <typename DomainA, typename DomainC, UInt Dim>
void Bridging<DomainA, DomainC, Dim>::synchSumBuffer(
UInt stride, ContainerArray<Real> &buf) {
if (this->local_points != 0) {
this->synchronizeVectorBySum("synchsumvector", buf, stride);
}
}
/* -------------------------------------------------------------------------- */
template <typename ContainerA, typename ContainerB, UInt Dim>
void Bridging<ContainerA, ContainerB, Dim>::filterRejectedContinuumOwners(
std::vector<std::vector<UInt>> &unassociated_atoms) {
if (this->comm_group_A == this->comm_group_B)
return;
if (this->in_group_B) {
DUMP("local points", DBG_MESSAGE);
if (this->local_points == 0)
return;
UInt offset = 0;
// recompute indexes of continuum atoms
std::vector<int> current_indexes(this->assoc.size());
{
UInt indirection = 0;
for (UInt at = 0; at < this->assoc.size(); ++at) {
current_indexes[at] = -1;
if (this->assoc[at] == UINT_MAX)
continue;
current_indexes[at] = indirection;
++indirection;
}
LM_ASSERT(indirection == this->local_points, "this should not happend "
<< indirection << " "
<< this->local_points);
}
// remove unassociated atoms
{
offset = 0;
UInt cpt = 0;
for (UInt i = 0; i < this->nb_zone_A; ++i) {
if (this->com_with[i]) {
DUMP("atomic processor " << i << " unassociated "
<< unassociated_atoms[i].size() << " atoms ",
DBG_INFO_STARTUP);
UInt size = unassociated_atoms[i].size();
for (UInt j = 0; j < size; ++j) {
// get element that owned the atom
UInt at_index = offset + unassociated_atoms[i][j];
LM_ASSERT(this->assoc[at_index] != UINT_MAX,
"this should not happend !! " << at_index);
LM_ASSERT(current_indexes[at_index] != -1,
"this should not happend !! " << at_index);
UInt el = this->assoc[at_index];
this->smatrix->removeAtom(current_indexes[at_index], el);
this->assoc[at_index] = UINT_MAX;
++cpt;
DUMP("unassociated atom at position "
<< at_index << " whereas offset = " << offset
<< " next offset is "
<< offset + this->nb_points_per_proc[i],
DBG_INFO_STARTUP);
}
offset += this->nb_points_per_proc[i];
}
}
if (cpt) {
DUMP("removed " << cpt
<< " atoms from unassociation. Now local atoms is "
<< this->local_points,
DBG_MESSAGE);
this->local_points -= cpt;
DUMP("and becomes " << this->local_points, DBG_MESSAGE);
}
}
// little check
{
UInt cpt_tmp = 0;
for (UInt i = 0; i < this->assoc.size(); ++i)
if (this->assoc[i] != UINT_MAX)
cpt_tmp++;
if (cpt_tmp != this->local_points)
LM_FATAL("biip this should not happend " << cpt_tmp << " "
<< this->local_points);
}
// now renumber indirection in shapematrix
{
UInt indirection = 0;
for (UInt at = 0; at < this->assoc.size(); ++at) {
if (this->assoc[at] == UINT_MAX)
continue;
LM_ASSERT(current_indexes[at] != -1, "This should not happend");
this->smatrix->changeAtomIndirection(this->assoc[at],
current_indexes[at], indirection);
++indirection;
}
this->smatrix->swapAtomIndirections();
}
// //now i set the duo vector
this->createDuoVectorB("FE");
// std::stringstream sstr;
// sstr << "duoFE-" << this->getID();
// DuoDistributedVector & duo =
// this->createDuoVector(this->local_points,this->total_points);
// UInt counter = 0;
// offset = 0;
// for (UInt i = 0 ; i < this->nb_zone_A ; ++i)
// if (this->com_with[i]){
// for (UInt j = 0 ; j < this->nb_points_per_proc[i] ; ++j)
// if (this->assoc[j+offset] != UINT_MAX){
// duo.setDuoProc(counter,i);
// ++counter;
// }
// DUMP("compteur is " << counter << " after treating atoms from " << i
// << " offset is " << offset,DBG_INFO_STARTUP);
// offset += this->nb_points_per_proc[i];
// }
}
}
/* -------------------------------------------------------------------------- */
template <typename DomainA, typename DomainC, UInt Dim>
void Bridging<DomainA, DomainC, Dim>::updateForMigration() {
DUMPBYPROC("update migration for " << this->getID(), DBG_INFO, 0);
STARTTIMER("syncMigration resize");
// update local number of atoms in atomic part
if (this->in_group_A) {
this->local_points = this->pointList.size();
this->positions.resize(Dim * this->local_points);
}
STOPTIMER("syncMigration resize");
STARTTIMER("syncMigration duo");
this->synchronizeMigration(this->comm_group_A, this->comm_group_B);
STOPTIMER("syncMigration duo");
// pointList.setRelease(this->contA.getRelease());
// meshList.setRelease(this->contB.getRelease());
}
/* -------------------------------------------------------------------------- */
/* LMDESC Bridging
This class implements a bridging zone where
point and elements are associated. \\ \\
For debugging pruposes, several set of DOf are registered to the central
system:
\begin{itemize}
\item unmatched-fe-\${couplerID} : the set of nodes and elements
that are contained in the geometry provided by GEOMETRY keyword.
\item unmatched-point-\${couplerID} : the set of points
that are contained in the geometry provided by GEOMETRY keyword.
\item matched-fe-\${couplerID} : the set of nodes and elements
that are containing at least one point of unmatched-point-\${couplerID}.
\item matched-point-\${couplerID} : the set of points
that are contained in at least one element of unmatched-fe-\${couplerID}.
\end{itemize}
*/
/* LMHERITANCE filter_geometry dof_association */
template <typename ContainerA, typename ContainerB, UInt Dim>
void Bridging<ContainerA, ContainerB, Dim>::declareParams() {
this->addSubParsableObject(unmatchedMeshList);
DofAssociation<ContainerA, ContainerB, Dim>::declareParams();
}
/* -------------------------------------------------------------------------- */
DECLARE_BRIDGING_ATOMIC_CONTINUUM_TEMPLATE(Bridging)
DECLARE_BRIDGING_DD_CONTINUUM_TEMPLATE(Bridging)
__END_LIBMULTISCALE__
c4science · Help