All Classes Namespaces Files Functions Variables Typedefs Enumerations Enumerator Friends Macros Groups Pages
Xpetra_MapUtils.hpp
Go to the documentation of this file.
1 // @HEADER
2 //
3 // ***********************************************************************
4 //
5 // Xpetra: A linear algebra interface package
6 // Copyright 2012 Sandia Corporation
7 //
8 // Under the terms of Contract DE-AC04-94AL85000 with Sandia Corporation,
9 // the U.S. Government retains certain rights in this software.
10 //
11 // Redistribution and use in source and binary forms, with or without
12 // modification, are permitted provided that the following conditions are
13 // met:
14 //
15 // 1. Redistributions of source code must retain the above copyright
16 // notice, this list of conditions and the following disclaimer.
17 //
18 // 2. Redistributions in binary form must reproduce the above copyright
19 // notice, this list of conditions and the following disclaimer in the
20 // documentation and/or other materials provided with the distribution.
21 //
22 // 3. Neither the name of the Corporation nor the names of the
23 // contributors may be used to endorse or promote products derived from
24 // this software without specific prior written permission.
25 //
26 // THIS SOFTWARE IS PROVIDED BY SANDIA CORPORATION "AS IS" AND ANY
27 // EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
28 // IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
29 // PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL SANDIA CORPORATION OR THE
30 // CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
31 // EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
32 // PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
33 // PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
34 // LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
35 // NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
36 // SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
37 //
38 // Questions? Contact
39 // Jonathan Hu (jhu@sandia.gov)
40 // Andrey Prokopenko (aprokop@sandia.gov)
41 // Ray Tuminaro (rstumin@sandia.gov)
42 // Tobias Wiesner (tawiesn@sandia.gov)
43 //
44 // ***********************************************************************
45 //
46 // @HEADER
47 #ifndef PACKAGES_XPETRA_SUP_MAP_UTILS_HPP_
48 #define PACKAGES_XPETRA_SUP_MAP_UTILS_HPP_
49 
50 #include "Xpetra_ConfigDefs.hpp"
51 
52 #include "Xpetra_Exceptions.hpp"
53 #include "Xpetra_Map.hpp"
54 #include "Xpetra_MapFactory.hpp"
55 
56 namespace Xpetra {
57 
58 #ifndef DOXYGEN_SHOULD_SKIP_THIS
59  // forward declaration of BlockedMap, needed to prevent circular inclusions
60  template<class LO, class GO, class N> class BlockedMap;
61 #endif
62 
70 template <class LocalOrdinal,
71  class GlobalOrdinal,
73 class MapUtils {
74 #undef XPETRA_MAPUTILS_SHORT
76 
77 public:
78 
94 
95  // ToDo Resolve header issues to allow for using this routing in Xpetra::BlockedMap.
96 
97  // merge submaps to global map
98  std::vector<GlobalOrdinal> gids;
99  for(size_t tt = 0; tt<subMaps.size(); ++tt) {
101  Teuchos::ArrayView< const GlobalOrdinal > subMapGids = subMap->getNodeElementList();
102  gids.insert(gids.end(), subMapGids.begin(), subMapGids.end());
103  }
104 
105  const GlobalOrdinal INVALID = Teuchos::OrdinalTraits<Xpetra::global_size_t>::invalid();
106  //std::sort(gids.begin(), gids.end());
107  //gids.erase(std::unique(gids.begin(), gids.end()), gids.end());
108  Teuchos::ArrayView<GlobalOrdinal> gidsView(&gids[0], gids.size());
109  Teuchos::RCP<Xpetra::Map<LocalOrdinal,GlobalOrdinal,Node> > fullMap = Xpetra::MapFactory<LocalOrdinal,GlobalOrdinal,Node>::Build(subMaps[0]->lib(), INVALID, gidsView, subMaps[0]->getIndexBase(), subMaps[0]->getComm());
110  return fullMap;
111  }
112 
131  {
134  "Xpetra::MatrixUtils::shrinkMapGIDs: the non-overlapping map must not have more local ids than the overlapping map.")
135 
138  "Xpetra::MatrixUtils::shrinkMapGIDs: the maximum GIDs of the overlapping and non-overlapping maps must be the same.")
139 
140  RCP< const Teuchos::Comm<int> > comm = input.getComm();
141 
142  // we expect input to be the potentially overlapping map associated with nonOvlInput as the non-overlapping
143  // map with the same GIDs over all processors (e.g. column map and domain map). We use the nonOvlInput map
144  // to determine which GIDs are owned by which processor.
145 
146  // calculate offset for new global Ids
147  std::vector<int> myGIDs(comm->getSize(),0);
148  std::vector<int> numGIDs(comm->getSize(),0);
149  myGIDs[comm->getRank()] = nonOvlInput.getNodeNumElements();
150  Teuchos::reduceAll(*comm,Teuchos::REDUCE_MAX,comm->getSize(),&myGIDs[0],&numGIDs[0]);
151  size_t gidOffset = 0;
152  for(int p = 0; p < comm->getRank(); p++) gidOffset += numGIDs[p];
153 
154  // we use nonOvlInput to assign the globally unique shrinked GIDs and communicate them to input.
155  std::map<const GlobalOrdinal, GlobalOrdinal> origGID2newGID;
156  for(size_t i = 0; i < nonOvlInput.getNodeNumElements(); i++)
157  {
158  origGID2newGID[nonOvlInput.getGlobalElement(i)] = Teuchos::as<GlobalOrdinal>(i) + Teuchos::as<GlobalOrdinal>(gidOffset);
159  }
160  // build an overlapping version of mySpecialMap
161  Teuchos::Array<GlobalOrdinal> ovlUnknownStatusGids;
162  Teuchos::Array<GlobalOrdinal> ovlFoundStatusGids;
163  // loop over global column map of A and find all GIDs where it is not sure, whether they are special or not
164  for(size_t i = 0; i<input.getNodeNumElements(); i++) {
165  GlobalOrdinal gcid = input.getGlobalElement(i);
166  if( nonOvlInput.isNodeGlobalElement(gcid) == false) {
167  ovlUnknownStatusGids.push_back(gcid);
168  }
169  }
170 
171  // Communicate the number of DOFs on each processor
172  std::vector<int> myUnknownDofGIDs(comm->getSize(),0);
173  std::vector<int> numUnknownDofGIDs(comm->getSize(),0);
174  myUnknownDofGIDs[comm->getRank()] = ovlUnknownStatusGids.size();
175  Teuchos::reduceAll(*comm,Teuchos::REDUCE_MAX,comm->getSize(),&myUnknownDofGIDs[0],&numUnknownDofGIDs[0]);
176 
177  // create array containing all DOF GIDs
178  size_t cntUnknownDofGIDs = 0;
179  for(int p = 0; p < comm->getSize(); p++) cntUnknownDofGIDs += numUnknownDofGIDs[p];
180  std::vector<GlobalOrdinal> lUnknownDofGIDs(cntUnknownDofGIDs,0); // local version to be filled
181  std::vector<GlobalOrdinal> gUnknownDofGIDs(cntUnknownDofGIDs,0); // global version after communication
182  // calculate the offset and fill chunk of memory with local data on each processor
183  size_t cntUnknownOffset = 0;
184  for(int p = 0; p < comm->getRank(); p++) cntUnknownOffset += numUnknownDofGIDs[p];
185  for(size_t k=0; k < Teuchos::as<size_t>(ovlUnknownStatusGids.size()); k++) {
186  lUnknownDofGIDs[k+cntUnknownOffset] = ovlUnknownStatusGids[k];
187  }
188  if(cntUnknownDofGIDs > 0) // only perform communication if there are unknown DOF GIDs
189  Teuchos::reduceAll(*comm,Teuchos::REDUCE_MAX,Teuchos::as<int>(cntUnknownDofGIDs),&lUnknownDofGIDs[0],&gUnknownDofGIDs[0]);
190  std::vector<GlobalOrdinal> lTranslatedDofGIDs(cntUnknownDofGIDs,0); // local version to be filled
191  std::vector<GlobalOrdinal> gTranslatedDofGIDs(cntUnknownDofGIDs,0); // global version after communication
192  // loop through all GIDs with unknown status
193  for(size_t k=0; k < gUnknownDofGIDs.size(); k++) {
194  GlobalOrdinal curgid = gUnknownDofGIDs[k];
195  if(nonOvlInput.isNodeGlobalElement(curgid)) {
196  lTranslatedDofGIDs[k] = origGID2newGID[curgid]; // curgid is in special map (on this processor)
197  }
198  }
199  if(cntUnknownDofGIDs > 0) // only perform communication if there are unknown DOF GIDs
200  Teuchos::reduceAll(*comm,Teuchos::REDUCE_MAX,Teuchos::as<int>(cntUnknownDofGIDs),&lTranslatedDofGIDs[0],&gTranslatedDofGIDs[0]);
201 
202  for(size_t k=0; k < Teuchos::as<size_t>(ovlUnknownStatusGids.size()); k++) {
203  origGID2newGID[ovlUnknownStatusGids[k]] = gTranslatedDofGIDs[k+cntUnknownOffset];
204  }
205  Teuchos::Array<GlobalOrdinal> ovlDomainMapArray;
206  for(size_t i = 0; i<input.getNodeNumElements(); i++) {
207  GlobalOrdinal gcid = input.getGlobalElement(i);
208  ovlDomainMapArray.push_back(origGID2newGID[gcid]);
209  }
212  (nonOvlInput.lib(),Teuchos::OrdinalTraits<GlobalOrdinal>::invalid(),ovlDomainMapArray(),0,comm);
213  return ovlDomainMap;
214  }
215 
235  const Xpetra::Map<LocalOrdinal, GlobalOrdinal, Node>& nonOvlReferenceInput) {
236  //TEUCHOS_TEST_FOR_EXCEPTION(nonOvlInput.getNodeNumElements() > input.getNodeNumElements(), Xpetra::Exceptions::Incompatible, "Xpetra::MatrixUtils::transformThyra2XpetraGIDs: the non-overlapping map must not have more local ids than the overlapping map.");
237  TEUCHOS_TEST_FOR_EXCEPTION(nonOvlInput.getNodeNumElements() != nonOvlReferenceInput.getNodeNumElements(), Xpetra::Exceptions::Incompatible, "Xpetra::MatrixUtils::transformThyra2XpetraGIDs: the number of local Xpetra reference GIDs and local Thyra GIDs of the non-overlapping maps must be the same!");
238  //TEUCHOS_TEST_FOR_EXCEPTION(nonOvlInput.getMaxAllGlobalIndex() != input.getMaxAllGlobalIndex(), Xpetra::Exceptions::Incompatible, "Xpetra::MatrixUtils::transformThyra2XpetraGIDs: the maximum GIDs of the overlapping and non-overlapping maps must be the same. nonOvlInput.getMaxAllGlobalIndex() = " << nonOvlInput.getMaxAllGlobalIndex() << " ovlInput.getMaxAllGlobalIndex() = " << input.getMaxAllGlobalIndex());
239 
240  RCP< const Teuchos::Comm<int> > comm = input.getComm();
241 
242  // fill translation map as far as possible
243  std::map<const GlobalOrdinal, GlobalOrdinal> thyra2xpetraGID;
244  for(size_t i = 0; i < nonOvlInput.getNodeNumElements(); i++) {
245  thyra2xpetraGID[nonOvlInput.getGlobalElement(i)] =
246  nonOvlReferenceInput.getGlobalElement(i);
247  }
248 
249  // find all GIDs of the overlapping Thyra map which are not owned by this proc
250  Teuchos::Array<GlobalOrdinal> ovlUnknownStatusGids;
251  // loop over global column map of A and find all GIDs where it is not sure, whether they are special or not
252  for(size_t i = 0; i<input.getNodeNumElements(); i++) {
253  GlobalOrdinal gcid = input.getGlobalElement(i);
254  if( nonOvlInput.isNodeGlobalElement(gcid) == false) {
255  ovlUnknownStatusGids.push_back(gcid);
256  }
257  }
258 
259  // Communicate the number of DOFs on each processor
260  std::vector<int> myUnknownDofGIDs(comm->getSize(),0);
261  std::vector<int> numUnknownDofGIDs(comm->getSize(),0);
262  myUnknownDofGIDs[comm->getRank()] = ovlUnknownStatusGids.size();
263  Teuchos::reduceAll(*comm,Teuchos::REDUCE_MAX,comm->getSize(),&myUnknownDofGIDs[0],&numUnknownDofGIDs[0]);
264 
265  // create array containing all DOF GIDs
266  size_t cntUnknownDofGIDs = 0;
267  for(int p = 0; p < comm->getSize(); p++) cntUnknownDofGIDs += numUnknownDofGIDs[p];
268  std::vector<GlobalOrdinal> lUnknownDofGIDs(cntUnknownDofGIDs,0); // local version to be filled
269  std::vector<GlobalOrdinal> gUnknownDofGIDs(cntUnknownDofGIDs,0); // global version after communication
270  // calculate the offset and fill chunk of memory with local data on each processor
271  size_t cntUnknownOffset = 0;
272  for(int p = 0; p < comm->getRank(); p++) cntUnknownOffset += numUnknownDofGIDs[p];
273  for(size_t k=0; k < Teuchos::as<size_t>(ovlUnknownStatusGids.size()); k++) {
274  lUnknownDofGIDs[k+cntUnknownOffset] = ovlUnknownStatusGids[k];
275  }
276  if(cntUnknownDofGIDs > 0) // only perform communication if there are unknown DOF GIDs
277  Teuchos::reduceAll(*comm,Teuchos::REDUCE_MAX,Teuchos::as<int>(cntUnknownDofGIDs),&lUnknownDofGIDs[0],&gUnknownDofGIDs[0]);
278  std::vector<GlobalOrdinal> lTranslatedDofGIDs(cntUnknownDofGIDs,0); // local version to be filled
279  std::vector<GlobalOrdinal> gTranslatedDofGIDs(cntUnknownDofGIDs,0); // global version after communication
280  // loop through all GIDs with unknown status
281  for(size_t k=0; k < gUnknownDofGIDs.size(); k++) {
282  GlobalOrdinal curgid = gUnknownDofGIDs[k];
283  if(nonOvlInput.isNodeGlobalElement(curgid)) {
284  lTranslatedDofGIDs[k] = thyra2xpetraGID[curgid];
285  }
286  }
287  if(cntUnknownDofGIDs > 0) // only perform communication if there are unknown DOF GIDs
288  Teuchos::reduceAll(*comm,Teuchos::REDUCE_MAX,Teuchos::as<int>(cntUnknownDofGIDs),&lTranslatedDofGIDs[0],&gTranslatedDofGIDs[0]);
289 
290  for(size_t k=0; k < Teuchos::as<size_t>(ovlUnknownStatusGids.size()); k++) {
291  thyra2xpetraGID[ovlUnknownStatusGids[k]] = gTranslatedDofGIDs[k+cntUnknownOffset];
292  }
293  Teuchos::Array<GlobalOrdinal> ovlDomainMapArray;
294  for(size_t i = 0; i<input.getNodeNumElements(); i++) {
295  GlobalOrdinal gcid = input.getGlobalElement(i);
296  ovlDomainMapArray.push_back(thyra2xpetraGID[gcid]);
297  }
300  (nonOvlInput.lib(),Teuchos::OrdinalTraits<GlobalOrdinal>::invalid(),ovlDomainMapArray(),0,comm);
301 
302  TEUCHOS_TEST_FOR_EXCEPTION(input.getNodeNumElements() != ovlDomainMap->getNodeNumElements(), Xpetra::Exceptions::Incompatible, "Xpetra::MatrixUtils::transformThyra2XpetraGIDs: the number of local Thyra reference GIDs (overlapping) and local Xpetra GIDs (overlapping) must be the same!");
303  //TEUCHOS_TEST_FOR_EXCEPTION(nonOvlReferenceInput.getMaxAllGlobalIndex() != ovlDomainMap->getMaxAllGlobalIndex(), Xpetra::Exceptions::Incompatible, "Xpetra::MatrixUtils::transformThyra2XpetraGIDs: the maximum GIDs of the overlapping and non-overlapping Xpetra maps must be the same.");
304 
305  return ovlDomainMap;
306  }
307 
315  const Xpetra::Map<LocalOrdinal, GlobalOrdinal, Node>& input, GlobalOrdinal offset) {
316 
318  RCP< const Teuchos::Comm<int> > comm = input.getComm();
319 
320  RCP<const Xpetra::Map<LocalOrdinal,GlobalOrdinal,Node> > rcpInput = Teuchos::rcpFromRef(input);
321 
322  // check whether input map is a BlockedMap or a standard Map
324  if(rcpBlockedInput.is_null() == true) {
325 
326  // create a new map with Xpetra GIDs (may start not from GID = 0)
327  std::vector<GlobalOrdinal> gids;
328  for(LocalOrdinal l = 0; l < Teuchos::as<LocalOrdinal>(rcpInput->getNodeNumElements()); ++l) {
329  GlobalOrdinal gid = rcpInput->getGlobalElement(l) + offset;
330  gids.push_back(gid);
331  }
332  Teuchos::ArrayView<GO> gidsView(&gids[0], gids.size());
333  RCP<Map> fullMap = MapFactory::Build(rcpInput->lib(), INVALID, gidsView, rcpInput->getIndexBase(), comm);
334  return fullMap;
335  }
336 
337  // SPECIAL CASE: input is a blocked map
338  // we have to recursively call this routine to get a BlockedMap containing (unique) Xpetra style GIDs
339 
340  size_t numMaps = rcpBlockedInput->getNumMaps();
341 
342  // first calucale GID offsets in submaps
343  // we need that for generating Xpetra GIDs
344  std::vector<GlobalOrdinal> gidOffsets(numMaps,0);
345  for(size_t i = 1; i < numMaps; ++i) {
346  gidOffsets[i] = rcpBlockedInput->getMap(i-1,true)->getMaxAllGlobalIndex() + gidOffsets[i-1] + 1;
347  }
348 
349  std::vector<RCP<const Map> > mapsXpetra(rcpBlockedInput->getNumMaps(), Teuchos::null);
350  std::vector<RCP<const Map> > mapsThyra (rcpBlockedInput->getNumMaps(), Teuchos::null);
351  for (size_t b = 0; b < rcpBlockedInput->getNumMaps(); ++b){
352  // extract sub map with Thyra style gids
353  // this can be an underlying Map or BlockedMap object
354  RCP<const Map> subMapThyra = rcpBlockedInput->getMap(b,true);
355  RCP<const Map> subMapXpetra = MapUtils::transformThyra2XpetraGIDs(*subMapThyra, gidOffsets[b] + offset); // recursive call
356  mapsXpetra[b] = subMapXpetra; // map can be of type Map or BlockedMap
357  mapsThyra[b] = subMapThyra; // map can be of type Map or BlockedMap
358  }
359 
361  return resultMap;
362  }
363 
364 };
365 
366 } // end namespace Xpetra
367 
368 #define XPETRA_MAPUTILS_SHORT
369 
370 #endif // PACKAGES_XPETRA_SUP_MAP_UTILS_HPP_
virtual Teuchos::RCP< const Teuchos::Comm< int > > getComm() const =0
Get this Map&#39;s Comm object.
GlobalOrdinal GO
iterator begin() const
#define TEUCHOS_TEST_FOR_EXCEPTION(throw_exception_test, Exception, msg)
virtual size_t getNodeNumElements() const =0
The number of elements belonging to the calling process.
static Teuchos::RCP< const Xpetra::Map< LocalOrdinal, GlobalOrdinal, Node > > concatenateMaps(const std::vector< Teuchos::RCP< const Xpetra::Map< LocalOrdinal, GlobalOrdinal, Node > > > &subMaps)
Helper function to concatenate several maps.
virtual bool isNodeGlobalElement(GlobalOrdinal globalIndex) const =0
Whether the given global index is valid for this Map on this process.
TEUCHOS_DEPRECATED RCP< T > rcp(T *p, Dealloc_T dealloc, bool owns_mem)
static Teuchos::RCP< Map< LocalOrdinal, GlobalOrdinal, Node > > Build(UnderlyingLib lib, global_size_t numGlobalElements, GlobalOrdinal indexBase, const Teuchos::RCP< const Teuchos::Comm< int >> &comm, LocalGlobal lg=Xpetra::GloballyDistributed)
Map constructor with Xpetra-defined contiguous uniform distribution.
static Teuchos::RCP< Xpetra::Map< LocalOrdinal, GlobalOrdinal, Node > > transformThyra2XpetraGIDs(const Xpetra::Map< LocalOrdinal, GlobalOrdinal, Node > &input, const Xpetra::Map< LocalOrdinal, GlobalOrdinal, Node > &nonOvlInput, const Xpetra::Map< LocalOrdinal, GlobalOrdinal, Node > &nonOvlReferenceInput)
replace set of global ids by new global ids
TEUCHOS_DEPRECATED void reduceAll(const Comm< Ordinal > &comm, const EReductionType reductType, const Packet &send, Packet *globalReduct)
virtual GlobalOrdinal getMaxAllGlobalIndex() const =0
The maximum global index over all processes in the communicator.
virtual GlobalOrdinal getGlobalElement(LocalOrdinal localIndex) const =0
The global index corresponding to the given local index.
static Teuchos::RCP< Xpetra::Map< LocalOrdinal, GlobalOrdinal, Node > > shrinkMapGIDs(const Xpetra::Map< LocalOrdinal, GlobalOrdinal, Node > &input, const Xpetra::Map< LocalOrdinal, GlobalOrdinal, Node > &nonOvlInput)
Helper function to shrink the GIDs and generate a standard map whith GIDs starting at 0...
void push_back(const value_type &x)
iterator end() const
size_type size() const
Exception throws to report incompatible objects (like maps).
static Teuchos::RCP< const Xpetra::Map< LocalOrdinal, GlobalOrdinal, Node > > transformThyra2XpetraGIDs(const Xpetra::Map< LocalOrdinal, GlobalOrdinal, Node > &input, GlobalOrdinal offset)
replace set of global ids by new global ids
virtual UnderlyingLib lib() const =0
Get the library used by this object (Tpetra or Epetra?)
bool is_null() const