1 // Copyright (C) 2007-2015 CEA/DEN, EDF R&D
3 // This library is free software; you can redistribute it and/or
4 // modify it under the terms of the GNU Lesser General Public
5 // License as published by the Free Software Foundation; either
6 // version 2.1 of the License, or (at your option) any later version.
8 // This library is distributed in the hope that it will be useful,
9 // but WITHOUT ANY WARRANTY; without even the implied warranty of
10 // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
11 // Lesser General Public License for more details.
13 // You should have received a copy of the GNU Lesser General Public
14 // License along with this library; if not, write to the Free Software
15 // Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17 // See http://www.salome-platform.org/ or email : webmaster.salome@opencascade.com
19 // Author : Anthony Geay (CEA/DEN)
21 #include "OverlapDEC.hxx"
22 #include "CommInterface.hxx"
23 #include "ParaMESH.hxx"
24 #include "ParaFIELD.hxx"
25 #include "MPIProcessorGroup.hxx"
26 #include "OverlapElementLocator.hxx"
27 #include "OverlapInterpolationMatrix.hxx"
32 \anchor OverlapDEC-det
35 \section OverlapDEC-over Overview
37 The \c OverlapDEC enables the \ref InterpKerRemapGlobal "conservative remapping" of fields between
38 two parallel codes. This remapping is based on the computation of intersection volumes on
39 a \b single \b processor \b group. On this processor group are defined two field-templates called A
40 and B. The computation is possible for 3D meshes, 2D meshes, 3D-surface meshes, 1D meshes and
41 2D-curve meshes. Dimensions must be similar for the distribution templates A and B.
43 The main difference with \ref InterpKernelDEC-det "InterpKernelDEC" is that this
44 \ref para-dec "DEC" works with a *single* processor group, in which processors will share the work.
45 Consequently each processor manages two \ref MEDCouplingFieldTemplatesPage "field templates" (A and B)
46 called source and target.
47 Furthermore all processors in the processor group cooperate in the global interpolation matrix
48 computation. In this respect \c InterpKernelDEC is a specialization of \c OverlapDEC.
50 \section ParaMEDMEMOverlapDECAlgorithmDescription Algorithm description
52 Let's consider the following use case that is ran in ParaMEDMEMTest_OverlapDEC.cxx to describes
53 the different steps of the computation. The processor group contains 3 processors.
54 \anchor ParaMEDMEMOverlapDECImgTest1
55 \image html OverlapDEC1.png "Example split of the source and target mesh among the 3 procs"
57 \subsection ParaMEDMEMOverlapDECAlgoStep1 Step 1 : Bounding box exchange and global interaction between procs computation.
59 In order to reduce as much as possible the amount of communications between distant processors,
60 every processor computes a bounding box for A and B. Then a AllToAll communication is performed
62 every processor can compute the \b global interactions between processor.
63 This computation leads every processor to compute the same global TODO list expressed as a list
64 of pair. A pair ( x, y ) means that proc \b x fieldtemplate A can interact with fieltemplate B of
65 proc \b y because the two bounding boxes interact.
66 In the \ref ParaMEDMEMOverlapDECImgTest1 "example above" this computation leads to the following
67 a \b global TODO list :
69 \b (0,0),(0,1),(1,0),(1,2),(2,0),(2,1),(2,2)
71 Here the pair (0,2) does not appear because the bounding box of fieldtemplateA of proc#2 does
72 not intersect that of fieldtemplate B on proc#0.
74 Stage performed by ParaMEDMEM::OverlapElementLocator::computeBoundingBoxes.
76 \subsection ParaMEDMEMOverlapDECAlgoStep2 Step 2 : Computation of local TODO list
78 Starting from the global interaction previously computed in \ref ParaMEDMEMOverlapDECAlgoStep1
79 "Step 1", each proc computes the TODO list per proc.
80 The following rules is chosen : a pair (x,y) can be treated by either proc \#x or proc \#y,
81 in order to reduce the amount of data transfert among
82 processors. The algorithm chosen for load balancing is the following : Each processor has
83 an empty \b local TODO list at the beginning. Then for each pair (k,m) in
84 \b global TODO list, if proc\#k has less temporary local list than proc\#m pair, (k,m) is added
85 to temparary local TODO list of proc\#k.
86 If proc\#m has less temporary local TODO list than proc\#k pair, (k,m) is added to temporary
87 local TODO list of proc\#m.
88 If proc\#k and proc\#m have the same amount of temporary local TODO list pair, (k,m) is added to
89 temporary local TODO list of proc\#k.
91 In the \ref ParaMEDMEMOverlapDECImgTest1 "example above" this computation leads to the following
95 - proc\#1 : (0,1),(1,0)
96 - proc\#2 : (1,2),(2,0),(2,1),(2,2)
98 The algorithm described here is not perfect for this use case, we hope to enhance it soon.
100 At this stage each proc knows precisely its \b local TODO list (with regard to interpolation).
101 The \b local TODO list of other procs than local
102 is kept for future computations.
104 \subsection ParaMEDMEMOverlapDECAlgoStep3 Step 3 : Matrix echange between procs
106 Knowing the \b local TODO list, the aim now is to exchange field-templates between procs.
107 Each proc computes knowing TODO list per
108 proc computed in \ref ParaMEDMEMOverlapDECAlgoStep2 "Step 2" the exchange TODO list :
110 In the \ref ParaMEDMEMOverlapDECImgTest1 "example above" the exchange TODO list gives the
113 Sending TODO list per proc :
115 - proc \#0 : Send fieldtemplate A to Proc\#1, Send fieldtemplate B to Proc\#1, Send fieldtemplate
117 - Proc \#1 : Send fieldtemplate A to Proc\#2, Send fieldtemplate B to Proc\#2
118 - Proc \#2 : No send.
120 Receiving TODO list per proc :
122 - proc \#0 : No receiving
123 - proc \#1 : receiving fieldtemplate A from Proc\#0, receiving fieldtemplate B from Proc\#0
124 - proc \#2 : receiving fieldtemplate B from Proc\#0, receiving fieldtemplate A from Proc\#1,
125 receiving fieldtemplate B from Proc\#1
127 To avoid as much as possible large volumes of transfers between procs, only relevant parts of
128 meshes are sent. In order for proc\#k to send fieldtemplate A to fieldtemplate B
129 of proc \#m., proc\#k computes the part of mesh A contained in the boundingbox B of proc\#m. It
130 implies that the corresponding cellIds or nodeIds of the
131 corresponding part are sent to proc \#m too.
133 Let's consider the couple (k,m) in the TODO list. This couple is treated by either k or m as
134 seen in \ref ParaMEDMEMOverlapDECAlgoStep2 "here in Step2".
136 As will be dealt in Step 6, for final matrix-vector computations, the resulting matrix of the
137 couple (k,m) whereever it is computed (proc \#k or proc \#m)
138 will be stored in \b proc\#m.
140 - If proc \#k is in charge (performs the matrix computation) for this couple (k,m), target ids
141 (cells or nodes) of the mesh in proc \#m are renumbered, because proc \#m has seelected a sub mesh
142 of the target mesh to avoid large amounts of data to transfer. In this case as proc \#m is ultimately
143 in charge of the matrix, proc \#k must keep preciously the
144 source ids needed to be sent to proc\#m. No problem will appear for matrix assembling in proc m
145 for source ids because no restriction was done.
146 Concerning source ids to be sent for the matrix-vector computation, proc k will know precisely
147 which source ids field values to send to proc \#m.
148 This is embodied by OverlapMapping::keepTracksOfTargetIds in proc m.
150 - If proc \#m is in charge (performs matrix computation) for this couple (k,m), source ids (cells
151 or nodes) of the mesh in proc \#k are renumbered, because proc \#k has selected a sub mesh of the
152 source mesh to avoid large amounts of data to transfer. In this case as proc \#k is ultimately
153 in charge of the matrix, proc \#m receives the source ids
154 from remote proc \#k, and thus the matrix is directly correct, no need for renumbering as
155 in \ref ParaMEDMEMOverlapDECAlgoStep5 "Step 5". However proc \#k must
156 keep track of the ids sent to proc \#m for te matrix-vector computation.
157 This is incarnated by OverlapMapping::keepTracksOfSourceIds in proc k.
159 This step is performed in ParaMEDMEM::OverlapElementLocator::exchangeMeshes method.
161 \subsection ParaMEDMEMOverlapDECAlgoStep4 Step 4 : Computation of the interpolation matrix
163 After mesh exchange in \ref ParaMEDMEMOverlapDECAlgoStep3 "Step3" each processor has all the
164 required information to treat its \b local TODO list computed in
165 \ref ParaMEDMEMOverlapDECAlgoStep2 "Step2". This step is potentially CPU costly, which is why
166 the \b local TODO list per proc is expected to
167 be as well balanced as possible.
169 The interpolation is performed as the \ref ParaMEDMEM::MEDCouplingRemapper "remapper" does.
171 This operation is performed by OverlapInterpolationMatrix::addContribution method.
173 \subsection ParaMEDMEMOverlapDECAlgoStep5 Step 5 : Global matrix construction.
175 After having performed the TODO list at the end of \ref ParaMEDMEMOverlapDECAlgoStep4 "Step4"
176 we need to assemble the final matrix.
178 The final aim is to have a distributed matrix \f$ M_k \f$ on each proc\#k. In order to reduce
179 data exchange during the matrix product process,
180 \f$ M_k \f$ is built using sizeof(Proc group) \c std::vector< \c std::map<int,double> \c >.
182 For a proc\#k, it is necessary to fetch info of all matrices built in
183 \ref ParaMEDMEMOverlapDECAlgoStep4 "Step4" where the first element in pair (i,j)
186 After this step, the matrix repartition is the following after a call to
187 ParaMEDMEM::OverlapMapping::prepare :
189 - proc\#0 : (0,0),(1,0),(2,0)
190 - proc\#1 : (0,1),(2,1)
191 - proc\#2 : (1,2),(2,2)
193 Tuple (2,1) computed on proc 2 is stored in proc 1 after execution of the function
194 "prepare". This is an example of item 0 in \ref ParaMEDMEMOverlapDECAlgoStep2 "Step2".
195 Tuple (0,1) computed on proc 1 is stored in proc 1 too. This is an example of item 1 in \ref ParaMEDMEMOverlapDECAlgoStep2 "Step2".
197 In the end ParaMEDMEM::OverlapMapping::_proc_ids_to_send_vector_st will contain :
203 In the end ParaMEDMEM::OverlapMapping::_proc_ids_to_recv_vector_st will contain :
209 The method in charge to perform this is : ParaMEDMEM::OverlapMapping::prepare.
211 OverlapDEC::OverlapDEC(const std::set<int>& procIds, const MPI_Comm& world_comm):
212 _load_balancing_algo(1),
213 _own_group(true),_interpolation_matrix(0), _locator(0),
214 _source_field(0),_own_source_field(false),
215 _target_field(0),_own_target_field(false),
216 _default_field_value(0.0),
219 ParaMEDMEM::CommInterface comm;
220 int *ranks_world=new int[procIds.size()]; // ranks of sources and targets in world_comm
221 std::copy(procIds.begin(),procIds.end(),ranks_world);
222 MPI_Group group,world_group;
223 comm.commGroup(world_comm,&world_group);
224 comm.groupIncl(world_group,procIds.size(),ranks_world,&group);
225 delete [] ranks_world;
226 comm.commCreate(world_comm,group,&_comm);
227 comm.groupFree(&group);
228 comm.groupFree(&world_group);
229 if(_comm==MPI_COMM_NULL)
234 std::set<int> idsUnion;
235 for(std::size_t i=0;i<procIds.size();i++)
237 _group=new MPIProcessorGroup(comm,idsUnion,_comm);
240 OverlapDEC::~OverlapDEC()
244 if(_own_source_field)
245 delete _source_field;
246 if(_own_target_field)
247 delete _target_field;
248 delete _interpolation_matrix;
250 if (_comm != MPI_COMM_NULL)
252 ParaMEDMEM::CommInterface comm;
253 comm.commFree(&_comm);
257 void OverlapDEC::sendRecvData(bool way)
265 void OverlapDEC::sendData()
267 _interpolation_matrix->multiply(_default_field_value);
270 void OverlapDEC::recvData()
272 throw INTERP_KERNEL::Exception("Not implemented yet !!!!");
273 //_interpolation_matrix->transposeMultiply();
276 void OverlapDEC::synchronize()
280 // Check number of components of field on both side (for now allowing void field/mesh on one proc is not allowed)
281 if (!_source_field || !_source_field->getField())
282 throw INTERP_KERNEL::Exception("OverlapDEC::synchronize(): currently, having a void source field on a proc is not allowed!");
283 if (!_target_field || !_target_field->getField())
284 throw INTERP_KERNEL::Exception("OverlapDEC::synchronize(): currently, having a void target field on a proc is not allowed!");
285 if (_target_field->getField()->getNumberOfComponents() != _source_field->getField()->getNumberOfComponents())
286 throw INTERP_KERNEL::Exception("OverlapDEC::synchronize(): source and target field have different number of components!");
287 delete _interpolation_matrix;
288 _locator = new OverlapElementLocator(_source_field,_target_field,*_group, getBoundingBoxAdjustmentAbs(), _load_balancing_algo);
289 _interpolation_matrix=new OverlapInterpolationMatrix(_source_field,_target_field,*_group,*this,*this, *_locator);
290 _locator->copyOptions(*this);
291 _locator->exchangeMeshes(*_interpolation_matrix);
292 std::vector< std::pair<int,int> > jobs=_locator->getToDoList();
293 std::string srcMeth=_locator->getSourceMethod();
294 std::string trgMeth=_locator->getTargetMethod();
295 for(std::vector< std::pair<int,int> >::const_iterator it=jobs.begin();it!=jobs.end();it++)
297 const MEDCouplingPointSet *src=_locator->getSourceMesh((*it).first);
298 const DataArrayInt *srcIds=_locator->getSourceIds((*it).first);
299 const MEDCouplingPointSet *trg=_locator->getTargetMesh((*it).second);
300 const DataArrayInt *trgIds=_locator->getTargetIds((*it).second);
301 _interpolation_matrix->computeLocalIntersection(src,srcIds,srcMeth,(*it).first,trg,trgIds,trgMeth,(*it).second);
303 _interpolation_matrix->prepare(_locator->getProcsToSendFieldData());
304 _interpolation_matrix->computeSurfacesAndDeno();
307 void OverlapDEC::attachSourceLocalField(ParaFIELD *field, bool ownPt)
311 if(_own_source_field)
312 delete _source_field;
314 _own_source_field=ownPt;
317 void OverlapDEC::attachTargetLocalField(ParaFIELD *field, bool ownPt)
321 if(_own_target_field)
322 delete _target_field;
324 _own_target_field=ownPt;
327 void OverlapDEC::attachSourceLocalField(MEDCouplingFieldDouble *field)
332 ParaMESH *paramesh = new ParaMESH(static_cast<MEDCouplingPointSet *>(const_cast<MEDCouplingMesh *>(field->getMesh())),
333 *_group,field->getMesh()->getName());
334 ParaFIELD *tmpField=new ParaFIELD(field, paramesh, *_group);
335 tmpField->setOwnSupport(true);
336 attachSourceLocalField(tmpField,true);
339 void OverlapDEC::attachTargetLocalField(MEDCouplingFieldDouble *field)
344 ParaMESH *paramesh = new ParaMESH(static_cast<MEDCouplingPointSet *>(const_cast<MEDCouplingMesh *>(field->getMesh())),
345 *_group,field->getMesh()->getName());
346 ParaFIELD *tmpField=new ParaFIELD(field, paramesh, *_group);
347 tmpField->setOwnSupport(true);
348 attachTargetLocalField(tmpField,true);
351 bool OverlapDEC::isInGroup() const
355 return _group->containsMyRank();