Page Menu
Home
c4science
Search
Configure Global Search
Log In
Files
F90754767
facet_synchronizer.cc
No One
Temporary
Actions
Download File
Edit File
Delete File
View Transforms
Subscribe
Mute Notifications
Award Token
Subscribers
None
File Metadata
Details
File Info
Storage
Attached
Created
Mon, Nov 4, 11:21
Size
8 KB
Mime Type
text/x-c
Expires
Wed, Nov 6, 11:21 (2 d)
Engine
blob
Format
Raw Data
Handle
22119573
Attached To
rAKA akantu
facet_synchronizer.cc
View Options
/**
* @file facet_synchronizer.cc
*
* @author Nicolas Richart <nicolas.richart@epfl.ch>
* @author Marco Vocialta <marco.vocialta@epfl.ch>
*
* @date creation: Wed Nov 05 2014
* @date last modification: Fri Jan 26 2018
*
* @brief Facet synchronizer for parallel simulations with cohesive elments
*
*
* 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 "facet_synchronizer.hh"
/* -------------------------------------------------------------------------- */
#if defined(AKANTU_MODULE)
#define AKANTU_MODULE_SAVE_ AKANTU_MODULE
#undef AKANTU_MODULE
#endif
#define AKANTU_MODULE facet_synchronizer
namespace akantu {
/* -------------------------------------------------------------------------- */
FacetSynchronizer::FacetSynchronizer(
Mesh & mesh, const ElementSynchronizer & element_synchronizer,
const ID & id, MemoryID memory_id)
: ElementSynchronizer(mesh, id, memory_id) {
auto spatial_dimension = mesh.getSpatialDimension();
element_to_prank.initialize(mesh, _spatial_dimension = spatial_dimension - 1,
_ghost_type = _ghost, _with_nb_element = true,
_default_value = rank);
// Build element to prank
for (auto && scheme_pair :
element_synchronizer.communications.iterateSchemes(_recv)) {
auto proc = std::get<0>(scheme_pair);
const auto & scheme = std::get<1>(scheme_pair);
for (auto && elem : scheme) {
const auto & facet_to_element =
mesh.getSubelementToElement(elem.type, elem.ghost_type);
Vector<Element> facets = facet_to_element.begin(
facet_to_element.getNbComponent())[elem.element];
for (UInt f = 0; f < facets.size(); ++f) {
const auto & facet = facets(f);
if (facet == ElementNull) {
continue;
}
if (facet.ghost_type == _not_ghost) {
continue;
}
auto & facet_rank = element_to_prank(facet);
if ((proc < UInt(facet_rank)) || (UInt(facet_rank) == rank)) {
facet_rank = proc;
}
}
}
}
ElementTypeMapArray<UInt> facet_global_connectivities(
"facet_global_connectivities", id, memory_id);
facet_global_connectivities.initialize(
mesh, _spatial_dimension = spatial_dimension - 1, _with_nb_element = true,
_with_nb_nodes_per_element = true);
mesh.getGlobalConnectivity(facet_global_connectivities);
// \TODO perhaps a global element numbering might be useful here...
for (auto type : facet_global_connectivities.elementTypes(_spatial_dimension =
_all_dimensions,
_element_kind = _ek_not_defined, _ghost_type = _not_ghost)) {
auto & conn = facet_global_connectivities(type, _not_ghost);
auto conn_view = make_view(conn, conn.getNbComponent());
std::for_each(conn_view.begin(), conn_view.end(), [&](auto & conn) {
std::sort(conn.storage(), conn.storage() + conn.size());
});
}
/// init facet check tracking
ElementTypeMapArray<bool> facet_checked("facet_checked", id, memory_id);
std::map<UInt, ElementTypeMapArray<UInt>> recv_connectivities;
/// Generate the recv scheme and connnectivities to send to the other
/// processors
for (auto && scheme_pair :
element_synchronizer.communications.iterateSchemes(_recv)) {
facet_checked.initialize(mesh, _spatial_dimension = spatial_dimension - 1,
_ghost_type = _ghost, _with_nb_element = true,
_default_value = false);
auto proc = scheme_pair.first;
const auto & elements = scheme_pair.second;
auto & facet_scheme = communications.createScheme(proc, _recv);
// this creates empty arrays...
auto & connectivities_for_proc = recv_connectivities[proc];
connectivities_for_proc.setID(
id + ":connectivities_for_proc:" + std::to_string(proc));
connectivities_for_proc.initialize(
mesh, _spatial_dimension = spatial_dimension - 1,
_with_nb_nodes_per_element = true, _ghost_type = _ghost);
// for every element in the element synchronizer communication scheme,
// check the facets to see if they should be communicated and create a
// connectivity array to match with the one other processors might send
for (auto && element : elements) {
const auto & facet_to_element =
mesh.getSubelementToElement(element.type, element.ghost_type);
Vector<Element> facets = facet_to_element.begin(
facet_to_element.getNbComponent())[element.element];
for (UInt f = 0; f < facets.size(); ++f) {
auto & facet = facets(f);
// exclude no valid facets
if (facet == ElementNull) {
continue;
}
// exclude _ghost facet from send scheme and _not_ghost from receive
if (facet.ghost_type != _ghost) {
continue;
}
// exclude facet from other processors then the one of current
// interest in case of receive scheme
if (UInt(element_to_prank(facet)) != proc) {
continue;
}
auto & checked = facet_checked(facet);
// skip already checked facets
if (checked) {
continue;
}
checked = true;
facet_scheme.push_back(facet);
auto & global_conn =
facet_global_connectivities(facet.type, facet.ghost_type);
Vector<UInt> conn =
global_conn.begin(global_conn.getNbComponent())[facet.element];
std::sort(conn.storage(), conn.storage() + conn.size());
connectivities_for_proc(facet.type, facet.ghost_type).push_back(conn);
}
}
}
std::vector<CommunicationRequest> send_requests;
/// do every communication by element type
for (auto && type : mesh.elementTypes(spatial_dimension - 1)) {
for (auto && pair : recv_connectivities) {
auto proc = std::get<0>(pair);
const auto & connectivities_for_proc = std::get<1>(pair);
auto && tag = Tag::genTag(proc, type, 1337);
send_requests.push_back(
communicator.asyncSend(connectivities_for_proc(type, _ghost), proc,
tag, CommunicationMode::_synchronous));
}
auto nb_nodes_per_facet = Mesh::getNbNodesPerElement(type);
communicator.receiveAnyNumber<UInt>(
send_requests,
[&](auto && proc, auto && message) {
auto & local_connectivities =
facet_global_connectivities(type, _not_ghost);
auto & send_scheme = communications.createScheme(proc, _send);
auto conn_view = make_view(local_connectivities, nb_nodes_per_facet);
auto conn_begin = conn_view.begin();
auto conn_end = conn_view.end();
for (const auto & c_to_match :
make_view(message, nb_nodes_per_facet)) {
auto it = std::find(conn_begin, conn_end, c_to_match);
if (it != conn_end) {
auto facet = Element{type, UInt(it - conn_begin), _not_ghost};
send_scheme.push_back(facet);
} else {
AKANTU_EXCEPTION("No local facet found to send to proc "
<< proc << " corresponding to " << c_to_match);
}
}
},
Tag::genTag(rank, type, 1337));
}
}
} // namespace akantu
#if defined(AKANTU_MODULE_SAVE_)
#undef AKANTU_MODULE
#define AKANTU_MODULE AKANTU_MODULE_SAVE_
#undef AKANTU_MODULE_SAVE_
#endif
Event Timeline
Log In to Comment