src/ParaMEDMEM/ParaUMesh.cxx

   1 //
   2 // Copyright (C) 2020  CEA/DEN, EDF R&D
   3 //
   4 // This library is free software; you can redistribute it and/or
   5 // modify it under the terms of the GNU Lesser General Public
   6 // License as published by the Free Software Foundation; either
   7 // version 2.1 of the License, or (at your option) any later version.
   8 //
   9 // This library is distributed in the hope that it will be useful,
  10 // but WITHOUT ANY WARRANTY; without even the implied warranty of
  11 // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  12 // Lesser General Public License for more details.
  13 //
  14 // You should have received a copy of the GNU Lesser General Public
  15 // License along with this library; if not, write to the Free Software
  16 // Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307 USA
  17 //
  18 // See http://www.salome-platform.org/ or email : webmaster.salome@opencascade.com
  19 //
  20 // Author : Anthony Geay (EDF R&D)
  21
  22 #include "ParaUMesh.hxx"
  23 #include "ProcessorGroup.hxx"
  24 #include "MPIProcessorGroup.hxx"
  25 #include "Topology.hxx"
  26 #include "BlockTopology.hxx"
  27 #include "CommInterface.hxx"
  28 #include "MEDCouplingMemArray.hxx"
  29
  30 #include "mpi.h"
  31
  32 #include <fstream>
  33 #include <sstream>
  34 #include <numeric>
  35 #include <memory>
  36 #include <vector>
  37
  38 using namespace MEDCoupling;
  39
  40 ParaUMesh *ParaUMesh::New(MEDCouplingUMesh *mesh, DataArrayIdType *globalCellIds, DataArrayIdType *globalNodeIds)
  41 {
  42   return new ParaUMesh(mesh,globalCellIds,globalNodeIds);
  43 }
  44
  45 ParaUMesh::ParaUMesh(MEDCouplingUMesh *mesh, DataArrayIdType *globalCellIds, DataArrayIdType *globalNodeIds)
  46 {
  47   _mesh.takeRef(mesh);
  48   _cell_global.takeRef(globalCellIds);
  49   _node_global.takeRef(globalNodeIds);
  50   _mesh.checkNotNull();
  51   _cell_global.checkNotNull();
  52   _node_global.checkNotNull();
  53   _mesh->checkConsistencyLight();
  54   if(_mesh->getNumberOfNodes() != _node_global->getNumberOfTuples())
  55     throw INTERP_KERNEL::Exception("ParaUMesh constructor : mismatch between # nodes and len of global # nodes.");
  56   if(_mesh->getNumberOfCells() != _cell_global->getNumberOfTuples())
  57     throw INTERP_KERNEL::Exception("ParaUMesh constructor : mismatch between # cells and len of global # cells.");
  58 }
  59
  60 std::size_t ParaUMesh::getHeapMemorySizeWithoutChildren() const
  61 {
  62   return 0;
  63 }
  64
  65 std::vector<const BigMemoryObject *> ParaUMesh::getDirectChildrenWithNull() const
  66 {
  67   return {_mesh,_cell_global,_node_global};
  68 }
  69
  70 /*!
  71 * This method computes the cells part of distributed mesh lying on \a globalNodeIds nodes.
  72 * The input \a globalNodeIds are not supposed to reside on the current process.
  73 */
  74 MCAuto<DataArrayIdType> ParaUMesh::getCellIdsLyingOnNodes(const DataArrayIdType *globalNodeIds, bool fullyIn) const
  75 {
  76   if(fullyIn)
  77     return this->getCellIdsLyingOnNodesTrue(globalNodeIds);
  78   else
  79     return this->getCellIdsLyingOnNodesFalse(globalNodeIds);
  80 }
  81
  82 MCAuto<DataArrayIdType> ParaUMesh::getCellIdsLyingOnNodesTrue(const DataArrayIdType *globalNodeIds) const
  83 {
  84   MPI_Comm comm(MPI_COMM_WORLD);
  85   CommInterface ci;
  86   int size;
  87   ci.commSize(comm,&size);
  88   std::unique_ptr<mcIdType[]> nbOfElems(new mcIdType[size]),nbOfElems2(new mcIdType[size]),nbOfElems3(new mcIdType[size]);
  89   mcIdType nbOfNodeIdsLoc(globalNodeIds->getNumberOfTuples());
  90   ci.allGather(&nbOfNodeIdsLoc,1,MPI_ID_TYPE,nbOfElems.get(),1,MPI_ID_TYPE,comm);
  91   //store for each proc the local nodeids intercepted by current proc
  92   std::vector< MCAuto<DataArrayIdType> > tabs(size);
  93   // loop to avoid to all procs to have all the nodes per proc
  94   for(int subDiv = 0 ; subDiv < size ; ++subDiv)
  95   {
  96     std::unique_ptr<mcIdType[]> nbOfElemsSp(CommInterface::SplitArrayOfLength(nbOfElems,size,subDiv,size));
  97     mcIdType nbOfNodeIdsSum(std::accumulate(nbOfElemsSp.get(),nbOfElemsSp.get()+size,0));
  98     std::unique_ptr<mcIdType[]> allGlobalNodeIds(new mcIdType[nbOfNodeIdsSum]);
  99     std::unique_ptr<int[]> nbOfElemsInt( CommInterface::ToIntArray<mcIdType>(nbOfElemsSp,size) );
 100     std::unique_ptr<int[]> offsetsIn( CommInterface::ComputeOffset(nbOfElemsInt,size) );
 101     mcIdType startGlobalNodeIds,endGlobalNodeIds;
 102     DataArray::GetSlice(0,globalNodeIds->getNumberOfTuples(),1,subDiv,size,startGlobalNodeIds,endGlobalNodeIds);
 103     ci.allGatherV(globalNodeIds->begin()+startGlobalNodeIds,endGlobalNodeIds-startGlobalNodeIds,MPI_ID_TYPE,allGlobalNodeIds.get(),nbOfElemsInt.get(),offsetsIn.get(),MPI_ID_TYPE,comm);
 104     mcIdType offset(0);
 105     for(int curRk = 0 ; curRk < size ; ++curRk)
 106     {
 107       MCAuto<DataArrayIdType> globalNodeIdsOfCurProc(DataArrayIdType::New());
 108       globalNodeIdsOfCurProc->useArray(allGlobalNodeIds.get()+offset,false,DeallocType::CPP_DEALLOC,nbOfElemsSp[curRk],1);
 109       offset += nbOfElemsSp[curRk];
 110       MCAuto<DataArrayIdType> globalNodeIdsCaptured(_node_global->buildIntersection(globalNodeIdsOfCurProc));
 111       MCAuto<DataArrayIdType> localNodeIdsToLocate(_node_global->findIdForEach(globalNodeIdsCaptured->begin(),globalNodeIdsCaptured->end()));
 112       if(tabs[curRk].isNull())
 113         tabs[curRk] = localNodeIdsToLocate;
 114       else
 115         tabs[curRk]->insertAtTheEnd(localNodeIdsToLocate->begin(),localNodeIdsToLocate->end());
 116     }
 117   }
 118
 119   for(int curRk = 0 ; curRk < size ; ++curRk)
 120   {
 121     MCAuto<DataArrayIdType> localNodeIds(tabs[curRk]);
 122     localNodeIds->sort();
 123     MCAuto<DataArrayIdType> localNodeIdsUnique(localNodeIds->buildUnique());
 124     MCAuto<DataArrayIdType> localCellCaptured(_mesh->getCellIdsLyingOnNodes(localNodeIdsUnique->begin(),localNodeIdsUnique->end(),true));
 125     MCAuto<DataArrayIdType> localCellCapturedGlob(_cell_global->selectByTupleIdSafe(localCellCaptured->begin(),localCellCaptured->end()));
 126     tabs[curRk] = localCellCapturedGlob;
 127   }
 128
 129   for(int curRk = 0 ; curRk < size ; ++curRk)
 130   {
 131     tabs[curRk] = tabs[curRk]->buildUniqueNotSorted();
 132     nbOfElems3[curRk] = tabs[curRk]->getNumberOfTuples();
 133   }
 134   std::vector<const DataArrayIdType *> tabss(tabs.begin(),tabs.end());
 135   MCAuto<DataArrayIdType> cells(DataArrayIdType::Aggregate(tabss));
 136   ci.allToAll(nbOfElems3.get(),1,MPI_ID_TYPE,nbOfElems2.get(),1,MPI_ID_TYPE,comm);
 137   mcIdType nbOfCellIdsSum(std::accumulate(nbOfElems2.get(),nbOfElems2.get()+size,0));
 138   MCAuto<DataArrayIdType> cellIdsFromProcs(DataArrayIdType::New());
 139   cellIdsFromProcs->alloc(nbOfCellIdsSum,1);
 140   {
 141     std::unique_ptr<int[]> nbOfElemsInt( CommInterface::ToIntArray<mcIdType>(nbOfElems3,size) ),nbOfElemsOutInt( CommInterface::ToIntArray<mcIdType>(nbOfElems2,size) );
 142     std::unique_ptr<int[]> offsetsIn( CommInterface::ComputeOffset(nbOfElemsInt,size) ), offsetsOut( CommInterface::ComputeOffset(nbOfElemsOutInt,size) );
 143     ci.allToAllV(cells->begin(),nbOfElemsInt.get(),offsetsIn.get(),MPI_ID_TYPE,
 144                  cellIdsFromProcs->getPointer(),nbOfElemsOutInt.get(),offsetsOut.get(),MPI_ID_TYPE,comm);
 145   }
 146   cellIdsFromProcs->sort();
 147   return cellIdsFromProcs;
 148 }
 149
 150 MCAuto<DataArrayIdType> ParaUMesh::getCellIdsLyingOnNodesFalse(const DataArrayIdType *globalNodeIds) const
 151 {
 152   MPI_Comm comm(MPI_COMM_WORLD);
 153   CommInterface ci;
 154   int size;
 155   ci.commSize(comm,&size);
 156   std::unique_ptr<mcIdType[]> nbOfElems(new mcIdType[size]),nbOfElems2(new mcIdType[size]),nbOfElems3(new mcIdType[size]);
 157   mcIdType nbOfNodeIdsLoc(globalNodeIds->getNumberOfTuples());
 158   ci.allGather(&nbOfNodeIdsLoc,1,MPI_ID_TYPE,nbOfElems.get(),1,MPI_ID_TYPE,comm);
 159   std::vector< MCAuto<DataArrayIdType> > tabs(size);
 160   // loop to avoid to all procs to have all the nodes per proc
 161   for(int subDiv = 0 ; subDiv < size ; ++subDiv)
 162   {
 163     std::unique_ptr<mcIdType[]> nbOfElemsSp(CommInterface::SplitArrayOfLength(nbOfElems,size,subDiv,size));
 164     mcIdType nbOfNodeIdsSum(std::accumulate(nbOfElemsSp.get(),nbOfElemsSp.get()+size,0));
 165     std::unique_ptr<mcIdType[]> allGlobalNodeIds(new mcIdType[nbOfNodeIdsSum]);
 166     std::unique_ptr<int[]> nbOfElemsInt( CommInterface::ToIntArray<mcIdType>(nbOfElemsSp,size) );
 167     std::unique_ptr<int[]> offsetsIn( CommInterface::ComputeOffset(nbOfElemsInt,size) );
 168     mcIdType startGlobalNodeIds,endGlobalNodeIds;
 169     DataArray::GetSlice(0,globalNodeIds->getNumberOfTuples(),1,subDiv,size,startGlobalNodeIds,endGlobalNodeIds);
 170     ci.allGatherV(globalNodeIds->begin()+startGlobalNodeIds,endGlobalNodeIds-startGlobalNodeIds,MPI_ID_TYPE,allGlobalNodeIds.get(),nbOfElemsInt.get(),offsetsIn.get(),MPI_ID_TYPE,comm);
 171     mcIdType offset(0);
 172     for(int curRk = 0 ; curRk < size ; ++curRk)
 173     {
 174       MCAuto<DataArrayIdType> globalNodeIdsOfCurProc(DataArrayIdType::New());
 175       globalNodeIdsOfCurProc->useArray(allGlobalNodeIds.get()+offset,false,DeallocType::CPP_DEALLOC,nbOfElemsSp[curRk],1);
 176       offset += nbOfElemsSp[curRk];
 177       MCAuto<DataArrayIdType> globalNodeIdsCaptured(_node_global->buildIntersection(globalNodeIdsOfCurProc));
 178       MCAuto<DataArrayIdType> localNodeIdsToLocate(_node_global->findIdForEach(globalNodeIdsCaptured->begin(),globalNodeIdsCaptured->end()));
 179       MCAuto<DataArrayIdType> localCellCaptured(_mesh->getCellIdsLyingOnNodes(localNodeIdsToLocate->begin(),localNodeIdsToLocate->end(),false));
 180       MCAuto<DataArrayIdType> localCellCapturedGlob(_cell_global->selectByTupleIdSafe(localCellCaptured->begin(),localCellCaptured->end()));
 181       if(tabs[curRk].isNull())
 182         tabs[curRk] = localCellCapturedGlob;
 183       else
 184         tabs[curRk]->insertAtTheEnd(localCellCapturedGlob->begin(),localCellCapturedGlob->end());
 185     }
 186   }
 187   for(int curRk = 0 ; curRk < size ; ++curRk)
 188   {
 189     tabs[curRk] = tabs[curRk]->buildUniqueNotSorted();
 190     nbOfElems3[curRk] = tabs[curRk]->getNumberOfTuples();
 191   }
 192   std::vector<const DataArrayIdType *> tabss(tabs.begin(),tabs.end());
 193   MCAuto<DataArrayIdType> cells(DataArrayIdType::Aggregate(tabss));
 194   ci.allToAll(nbOfElems3.get(),1,MPI_ID_TYPE,nbOfElems2.get(),1,MPI_ID_TYPE,comm);
 195   mcIdType nbOfCellIdsSum(std::accumulate(nbOfElems2.get(),nbOfElems2.get()+size,0));
 196   MCAuto<DataArrayIdType> cellIdsFromProcs(DataArrayIdType::New());
 197   cellIdsFromProcs->alloc(nbOfCellIdsSum,1);
 198   {
 199     std::unique_ptr<int[]> nbOfElemsInt( CommInterface::ToIntArray<mcIdType>(nbOfElems3,size) ),nbOfElemsOutInt( CommInterface::ToIntArray<mcIdType>(nbOfElems2,size) );
 200     std::unique_ptr<int[]> offsetsIn( CommInterface::ComputeOffset(nbOfElemsInt,size) ), offsetsOut( CommInterface::ComputeOffset(nbOfElemsOutInt,size) );
 201     ci.allToAllV(cells->begin(),nbOfElemsInt.get(),offsetsIn.get(),MPI_ID_TYPE,
 202                  cellIdsFromProcs->getPointer(),nbOfElemsOutInt.get(),offsetsOut.get(),MPI_ID_TYPE,comm);
 203   }
 204   cellIdsFromProcs->sort();
 205   return cellIdsFromProcs;
 206 }
 207
 208 DataArrayIdType *ParaUMesh::redistributeCellField(const DataArrayIdType *globalCellIds, const DataArrayIdType *fieldValueToRed) const
 209 {
 210   return this->redistributeCellFieldT<mcIdType>(globalCellIds,fieldValueToRed);
 211 }
 212
 213 DataArrayDouble *ParaUMesh::redistributeCellField(const DataArrayIdType *globalCellIds, const DataArrayDouble *fieldValueToRed) const
 214 {
 215   return this->redistributeCellFieldT<double>(globalCellIds,fieldValueToRed);
 216 }
 217
 218 DataArrayIdType *ParaUMesh::redistributeNodeField(const DataArrayIdType *globalCellIds, const DataArrayIdType *fieldValueToRed) const
 219 {
 220   return this->redistributeNodeFieldT<mcIdType>(globalCellIds,fieldValueToRed);
 221 }
 222
 223 DataArrayDouble *ParaUMesh::redistributeNodeField(const DataArrayIdType *globalCellIds, const DataArrayDouble *fieldValueToRed) const
 224 {
 225   return this->redistributeNodeFieldT<double>(globalCellIds,fieldValueToRed);
 226 }
 227
 228 /*!
 229  * Return part of \a this mesh split over COMM_WORLD. Part is defined by global cell ids array \a globaCellIds.
 230  */
 231 ParaUMesh *ParaUMesh::redistributeCells(const DataArrayIdType *globalCellIds) const
 232 {
 233   MPI_Comm comm(MPI_COMM_WORLD);
 234   CommInterface ci;
 235   std::unique_ptr<mcIdType[]> allGlobalCellIds,allGlobalCellIdsIndex;
 236   int size(ci.allGatherArrays(comm,globalCellIds,allGlobalCellIds,allGlobalCellIdsIndex));
 237   // Prepare ParaUMesh parts to be sent : compute for each proc the contribution of current rank.
 238   std::vector< MCAuto<DataArrayIdType> > globalCellIdsToBeSent(size),globalNodeIdsToBeSent(size);
 239   std::vector< MCAuto<MEDCouplingUMesh> > meshPartsToBeSent(size);
 240   for(int curRk = 0 ; curRk < size ; ++curRk)
 241   {
 242     mcIdType offset(allGlobalCellIdsIndex[curRk]);
 243     MCAuto<DataArrayIdType> globalCellIdsOfCurProc(DataArrayIdType::New());
 244     globalCellIdsOfCurProc->useArray(allGlobalCellIds.get()+offset,false,DeallocType::CPP_DEALLOC,allGlobalCellIdsIndex[curRk+1]-offset,1);
 245     // the key call is here : compute for rank curRk the cells to be sent
 246     MCAuto<DataArrayIdType> globalCellIdsCaptured(_cell_global->buildIntersection(globalCellIdsOfCurProc));// OK for the global cellIds
 247     MCAuto<DataArrayIdType> localCellIdsCaptured(_cell_global->findIdForEach(globalCellIdsCaptured->begin(),globalCellIdsCaptured->end()));
 248     MCAuto<MEDCouplingUMesh> meshPart(_mesh->buildPartOfMySelf(localCellIdsCaptured->begin(),localCellIdsCaptured->end(),true));
 249     MCAuto<DataArrayIdType> o2n(meshPart->zipCoordsTraducer());// OK for the mesh
 250     MCAuto<DataArrayIdType> n2o(o2n->invertArrayO2N2N2O(meshPart->getNumberOfNodes()));
 251     MCAuto<DataArrayIdType> globalNodeIdsPart(_node_global->selectByTupleIdSafe(n2o->begin(),n2o->end())); // OK for the global nodeIds
 252     meshPartsToBeSent[curRk] = meshPart;
 253     globalCellIdsToBeSent[curRk] = globalCellIdsCaptured;
 254     globalNodeIdsToBeSent[curRk] = globalNodeIdsPart;
 255   }
 256   // Receive
 257   std::vector< MCAuto<DataArrayIdType> > globalCellIdsReceived,globalNodeIdsReceived;
 258   ci.allToAllArrays(comm,globalCellIdsToBeSent,globalCellIdsReceived);
 259   ci.allToAllArrays(comm,globalNodeIdsToBeSent,globalNodeIdsReceived);
 260   //now exchange the 3 arrays for the umesh : connectivity, connectivityindex and coordinates
 261   std::vector<const MEDCouplingUMesh *> meshPartsToBeSent2(FromVecAutoToVecOfConst<MEDCouplingUMesh>(meshPartsToBeSent));
 262   //connectivityindex
 263   std::vector< MCAuto<DataArrayIdType> > connectivityIndexReceived,connectivityReceived;
 264   {
 265     std::vector<const DataArrayIdType *> connectivityIndexToBeSent(UMeshConnectivityIndexIterator(0,&meshPartsToBeSent2),UMeshConnectivityIndexIterator(meshPartsToBeSent2.size(),&meshPartsToBeSent2));
 266     ci.allToAllArrays(comm,FromVecConstToVecAuto<DataArrayIdType>(connectivityIndexToBeSent),connectivityIndexReceived);
 267   }
 268   //connectivity
 269   {
 270     std::vector<const DataArrayIdType *> connectivityToBeSent(UMeshConnectivityIterator(0,&meshPartsToBeSent2),UMeshConnectivityIterator(meshPartsToBeSent2.size(),&meshPartsToBeSent2));
 271     ci.allToAllArrays(comm,FromVecConstToVecAuto<DataArrayIdType>(connectivityToBeSent),connectivityReceived);
 272   }
 273   //coordinates
 274   MCAuto<DataArrayDouble> coords;
 275   {
 276     std::vector<const DataArrayDouble *> coordsToBeSent(UMeshCoordsIterator(0,&meshPartsToBeSent2),UMeshCoordsIterator(meshPartsToBeSent2.size(),&meshPartsToBeSent2));
 277     ci.allToAllArrays(comm,FromVecConstToVecAuto<DataArrayDouble>(coordsToBeSent),coords);
 278   }
 279   /////// Sort it all !
 280   // firstly deal with nodes.
 281   MCAuto<DataArrayIdType> aggregatedNodeIds( DataArrayIdType::Aggregate(FromVecAutoToVecOfConst<DataArrayIdType>(globalNodeIdsReceived)) );
 282   MCAuto<DataArrayIdType> aggregatedNodeIdsSorted(aggregatedNodeIds->copySorted());
 283   MCAuto<DataArrayIdType> nodeIdsIntoAggregatedIds(DataArrayIdType::FindPermutationFromFirstToSecondDuplicate(aggregatedNodeIdsSorted,aggregatedNodeIds));
 284   MCAuto<DataArrayIdType> idxOfSameNodeIds(aggregatedNodeIdsSorted->indexOfSameConsecutiveValueGroups());
 285   MCAuto<DataArrayIdType> n2o_nodes(nodeIdsIntoAggregatedIds->selectByTupleIdSafe(idxOfSameNodeIds->begin(),idxOfSameNodeIds->end()-1));//new == new ordering so that global node ids are sorted . old == coarse ordering implied by the aggregation
 286   MCAuto<DataArrayIdType> finalGlobalNodeIds(aggregatedNodeIdsSorted->selectByTupleIdSafe(idxOfSameNodeIds->begin(),idxOfSameNodeIds->end()-1));
 287   MCAuto<DataArrayDouble> finalCoords(coords->selectByTupleIdSafe(n2o_nodes->begin(),n2o_nodes->end()));
 288   finalCoords->copyStringInfoFrom(*_mesh->getCoords());
 289   // secondly renumbering of node ids in connectivityReceived
 290   for(int curRk = 0 ; curRk < size ; ++curRk)
 291   {
 292     auto current(globalNodeIdsReceived[curRk]);
 293     MCAuto<DataArrayIdType> aa(finalGlobalNodeIds->findIdForEach(current->begin(),current->end()));
 294     // work on connectivityReceived[curRk] with transformWithIndArr but do not forget type of cells that should be excluded !
 295     auto connectivityToModify(connectivityReceived[curRk]);
 296     auto connectivityIndex(connectivityIndexReceived[curRk]);
 297     MCAuto<DataArrayIdType> types(connectivityToModify->selectByTupleIdSafe(connectivityIndex->begin(),connectivityIndex->end()-1));
 298     connectivityToModify->setPartOfValuesSimple3(0,connectivityIndex->begin(),connectivityIndex->end()-1,0,1,1);
 299     connectivityToModify->transformWithIndArr(aa->begin(),aa->end());
 300     connectivityToModify->setPartOfValues3(types,connectivityIndex->begin(),connectivityIndex->end()-1,0,1,1,true);
 301   }
 302   // thirdly renumber cells
 303   MCAuto<DataArrayIdType> aggregatedCellIds( DataArrayIdType::Aggregate(FromVecAutoToVecOfConst<DataArrayIdType>(globalCellIdsReceived)) );
 304   MCAuto<DataArrayIdType> aggregatedCellIdsSorted(aggregatedCellIds->copySorted());
 305   MCAuto<DataArrayIdType> idsIntoAggregatedIds(DataArrayIdType::FindPermutationFromFirstToSecondDuplicate(aggregatedCellIdsSorted,aggregatedCellIds));
 306   MCAuto<DataArrayIdType> cellIdsOfSameNodeIds(aggregatedCellIdsSorted->indexOfSameConsecutiveValueGroups());
 307   MCAuto<DataArrayIdType> n2o_cells(idsIntoAggregatedIds->selectByTupleIdSafe(cellIdsOfSameNodeIds->begin(),cellIdsOfSameNodeIds->end()-1));//new == new ordering so that global cell ids are sorted . old == coarse ordering implied by the aggregation
 308   // TODO : check coordsReceived==globalCellIds
 309   MCAuto<DataArrayIdType> connSorted,indicesSorted;
 310   {
 311     MCAuto<DataArrayIdType> conn(DataArrayIdType::Aggregate(FromVecAutoToVecOfConst<DataArrayIdType>(connectivityReceived)));
 312     MCAuto<DataArrayIdType> connIndex(DataArrayIdType::AggregateIndexes(FromVecAutoToVecOfConst<DataArrayIdType>(connectivityIndexReceived)));
 313     {
 314       DataArrayIdType *indicesSortedTmp(nullptr),*valuesSortedTmp(nullptr);
 315       DataArrayIdType::ExtractFromIndexedArrays(n2o_cells->begin(),n2o_cells->end(),conn,connIndex,valuesSortedTmp,indicesSortedTmp);
 316       indicesSorted = indicesSortedTmp; connSorted=valuesSortedTmp;
 317     }
 318   }
 319   // finalize all
 320   MCAuto<MEDCouplingUMesh> mesh(MEDCouplingUMesh::New(_mesh->getName(),_mesh->getMeshDimension()));
 321   mesh->setConnectivity(connSorted,indicesSorted,true);
 322   mesh->setCoords(finalCoords);
 323   mesh->setDescription(_mesh->getDescription());
 324   MCAuto<ParaUMesh> ret(ParaUMesh::New(mesh,aggregatedCellIdsSorted,finalGlobalNodeIds));
 325   return ret.retn();
 326 }