ViewVC Help
View File | Revision Log | Show Annotations | View Changeset | Root Listing
root/OpenMD/trunk/src/parallel/ForceMatrixDecomposition.cpp
(Generate patch)

Comparing branches/development/src/parallel/ForceMatrixDecomposition.cpp (file contents):
Revision 1551 by gezelter, Thu Apr 28 18:38:21 2011 UTC vs.
Revision 1577 by gezelter, Wed Jun 8 20:26:56 2011 UTC

# Line 42 | Line 42
42   #include "math/SquareMatrix3.hpp"
43   #include "nonbonded/NonBondedInteraction.hpp"
44   #include "brains/SnapshotManager.hpp"
45 + #include "brains/PairList.hpp"
46  
47   using namespace std;
48   namespace OpenMD {
# Line 54 | Line 55 | namespace OpenMD {
55    void ForceMatrixDecomposition::distributeInitialData() {
56      snap_ = sman_->getCurrentSnapshot();
57      storageLayout_ = sman_->getStorageLayout();
58 < #ifdef IS_MPI    
59 <    int nLocal = snap_->getNumberOfAtoms();
59 <    int nGroups = snap_->getNumberOfCutoffGroups();
60 <    
61 <    AtomCommIntRow = new Communicator<Row,int>(nLocal);
62 <    AtomCommRealRow = new Communicator<Row,RealType>(nLocal);
63 <    AtomCommVectorRow = new Communicator<Row,Vector3d>(nLocal);
64 <    AtomCommMatrixRow = new Communicator<Row,Mat3x3d>(nLocal);
58 >    ff_ = info_->getForceField();
59 >    nLocal_ = snap_->getNumberOfAtoms();
60  
61 <    AtomCommIntColumn = new Communicator<Column,int>(nLocal);
62 <    AtomCommRealColumn = new Communicator<Column,RealType>(nLocal);
63 <    AtomCommVectorColumn = new Communicator<Column,Vector3d>(nLocal);
64 <    AtomCommMatrixColumn = new Communicator<Column,Mat3x3d>(nLocal);
61 >    nGroups_ = info_->getNLocalCutoffGroups();
62 >    // gather the information for atomtype IDs (atids):
63 >    identsLocal = info_->getIdentArray();
64 >    AtomLocalToGlobal = info_->getGlobalAtomIndices();
65 >    cgLocalToGlobal = info_->getGlobalGroupIndices();
66 >    vector<int> globalGroupMembership = info_->getGlobalGroupMembership();
67 >    vector<RealType> massFactorsLocal = info_->getMassFactors();
68 >    PairList excludes = info_->getExcludedInteractions();
69 >    PairList oneTwo = info_->getOneTwoInteractions();
70 >    PairList oneThree = info_->getOneThreeInteractions();
71 >    PairList oneFour = info_->getOneFourInteractions();
72  
73 <    cgCommIntRow = new Communicator<Row,int>(nGroups);
74 <    cgCommVectorRow = new Communicator<Row,Vector3d>(nGroups);
75 <    cgCommIntColumn = new Communicator<Column,int>(nGroups);
76 <    cgCommVectorColumn = new Communicator<Column,Vector3d>(nGroups);
73 > #ifdef IS_MPI
74 >
75 >    AtomCommIntRow = new Communicator<Row,int>(nLocal_);
76 >    AtomCommRealRow = new Communicator<Row,RealType>(nLocal_);
77 >    AtomCommVectorRow = new Communicator<Row,Vector3d>(nLocal_);
78 >    AtomCommMatrixRow = new Communicator<Row,Mat3x3d>(nLocal_);
79 >    AtomCommPotRow = new Communicator<Row,potVec>(nLocal_);
80  
81 <    int nAtomsInRow = AtomCommIntRow->getSize();
82 <    int nAtomsInCol = AtomCommIntColumn->getSize();
83 <    int nGroupsInRow = cgCommIntRow->getSize();
84 <    int nGroupsInCol = cgCommIntColumn->getSize();
81 >    AtomCommIntColumn = new Communicator<Column,int>(nLocal_);
82 >    AtomCommRealColumn = new Communicator<Column,RealType>(nLocal_);
83 >    AtomCommVectorColumn = new Communicator<Column,Vector3d>(nLocal_);
84 >    AtomCommMatrixColumn = new Communicator<Column,Mat3x3d>(nLocal_);
85 >    AtomCommPotColumn = new Communicator<Column,potVec>(nLocal_);
86  
87 +    cgCommIntRow = new Communicator<Row,int>(nGroups_);
88 +    cgCommVectorRow = new Communicator<Row,Vector3d>(nGroups_);
89 +    cgCommIntColumn = new Communicator<Column,int>(nGroups_);
90 +    cgCommVectorColumn = new Communicator<Column,Vector3d>(nGroups_);
91 +
92 +    nAtomsInRow_ = AtomCommIntRow->getSize();
93 +    nAtomsInCol_ = AtomCommIntColumn->getSize();
94 +    nGroupsInRow_ = cgCommIntRow->getSize();
95 +    nGroupsInCol_ = cgCommIntColumn->getSize();
96 +
97      // Modify the data storage objects with the correct layouts and sizes:
98 <    atomRowData.resize(nAtomsInRow);
98 >    atomRowData.resize(nAtomsInRow_);
99      atomRowData.setStorageLayout(storageLayout_);
100 <    atomColData.resize(nAtomsInCol);
100 >    atomColData.resize(nAtomsInCol_);
101      atomColData.setStorageLayout(storageLayout_);
102 <    cgRowData.resize(nGroupsInRow);
102 >    cgRowData.resize(nGroupsInRow_);
103      cgRowData.setStorageLayout(DataStorage::dslPosition);
104 <    cgColData.resize(nGroupsInCol);
104 >    cgColData.resize(nGroupsInCol_);
105      cgColData.setStorageLayout(DataStorage::dslPosition);
106 +        
107 +    identsRow.resize(nAtomsInRow_);
108 +    identsCol.resize(nAtomsInCol_);
109      
91 –    vector<vector<RealType> > pot_row(N_INTERACTION_FAMILIES,
92 –                                      vector<RealType> (nAtomsInRow, 0.0));
93 –    vector<vector<RealType> > pot_col(N_INTERACTION_FAMILIES,
94 –                                      vector<RealType> (nAtomsInCol, 0.0));
95 –
96 –
97 –    vector<RealType> pot_local(N_INTERACTION_FAMILIES, 0.0);
98 –    
99 –    // gather the information for atomtype IDs (atids):
100 –    vector<int> identsLocal = info_->getIdentArray();
101 –    identsRow.reserve(nAtomsInRow);
102 –    identsCol.reserve(nAtomsInCol);
103 –    
110      AtomCommIntRow->gather(identsLocal, identsRow);
111      AtomCommIntColumn->gather(identsLocal, identsCol);
112      
107 –    AtomLocalToGlobal = info_->getGlobalAtomIndices();
113      AtomCommIntRow->gather(AtomLocalToGlobal, AtomRowToGlobal);
114      AtomCommIntColumn->gather(AtomLocalToGlobal, AtomColToGlobal);
115      
111 –    cgLocalToGlobal = info_->getGlobalGroupIndices();
116      cgCommIntRow->gather(cgLocalToGlobal, cgRowToGlobal);
117      cgCommIntColumn->gather(cgLocalToGlobal, cgColToGlobal);
118  
119 <    // still need:
120 <    // topoDist
121 <    // exclude
119 >    AtomCommRealRow->gather(massFactorsLocal, massFactorsRow);
120 >    AtomCommRealColumn->gather(massFactorsLocal, massFactorsCol);
121 >
122 >    groupListRow_.clear();
123 >    groupListRow_.resize(nGroupsInRow_);
124 >    for (int i = 0; i < nGroupsInRow_; i++) {
125 >      int gid = cgRowToGlobal[i];
126 >      for (int j = 0; j < nAtomsInRow_; j++) {
127 >        int aid = AtomRowToGlobal[j];
128 >        if (globalGroupMembership[aid] == gid)
129 >          groupListRow_[i].push_back(j);
130 >      }      
131 >    }
132 >
133 >    groupListCol_.clear();
134 >    groupListCol_.resize(nGroupsInCol_);
135 >    for (int i = 0; i < nGroupsInCol_; i++) {
136 >      int gid = cgColToGlobal[i];
137 >      for (int j = 0; j < nAtomsInCol_; j++) {
138 >        int aid = AtomColToGlobal[j];
139 >        if (globalGroupMembership[aid] == gid)
140 >          groupListCol_[i].push_back(j);
141 >      }      
142 >    }
143 >
144 >    skipsForRowAtom.clear();
145 >    skipsForRowAtom.resize(nAtomsInRow_);
146 >    for (int i = 0; i < nAtomsInRow_; i++) {
147 >      int iglob = AtomRowToGlobal[i];
148 >      for (int j = 0; j < nAtomsInCol_; j++) {
149 >        int jglob = AtomColToGlobal[j];        
150 >        if (excludes.hasPair(iglob, jglob))
151 >          skipsForRowAtom[i].push_back(j);      
152 >      }      
153 >    }
154 >
155 >    toposForRowAtom.clear();
156 >    toposForRowAtom.resize(nAtomsInRow_);
157 >    for (int i = 0; i < nAtomsInRow_; i++) {
158 >      int iglob = AtomRowToGlobal[i];
159 >      int nTopos = 0;
160 >      for (int j = 0; j < nAtomsInCol_; j++) {
161 >        int jglob = AtomColToGlobal[j];        
162 >        if (oneTwo.hasPair(iglob, jglob)) {
163 >          toposForRowAtom[i].push_back(j);
164 >          topoDistRow[i][nTopos] = 1;
165 >          nTopos++;
166 >        }
167 >        if (oneThree.hasPair(iglob, jglob)) {
168 >          toposForRowAtom[i].push_back(j);
169 >          topoDistRow[i][nTopos] = 2;
170 >          nTopos++;
171 >        }
172 >        if (oneFour.hasPair(iglob, jglob)) {
173 >          toposForRowAtom[i].push_back(j);
174 >          topoDistRow[i][nTopos] = 3;
175 >          nTopos++;
176 >        }
177 >      }      
178 >    }
179 >
180   #endif
181 +    groupList_.clear();
182 +    groupList_.resize(nGroups_);
183 +    for (int i = 0; i < nGroups_; i++) {
184 +      int gid = cgLocalToGlobal[i];
185 +      for (int j = 0; j < nLocal_; j++) {
186 +        int aid = AtomLocalToGlobal[j];
187 +        if (globalGroupMembership[aid] == gid) {
188 +          groupList_[i].push_back(j);
189 +
190 +        }
191 +      }      
192 +    }
193 +
194 +    skipsForLocalAtom.clear();
195 +    skipsForLocalAtom.resize(nLocal_);
196 +
197 +    for (int i = 0; i < nLocal_; i++) {
198 +      int iglob = AtomLocalToGlobal[i];
199 +      for (int j = 0; j < nLocal_; j++) {
200 +        int jglob = AtomLocalToGlobal[j];        
201 +        if (excludes.hasPair(iglob, jglob))
202 +          skipsForLocalAtom[i].push_back(j);      
203 +      }      
204 +    }
205 +    toposForLocalAtom.clear();
206 +    toposForLocalAtom.resize(nLocal_);
207 +    for (int i = 0; i < nLocal_; i++) {
208 +      int iglob = AtomLocalToGlobal[i];
209 +      int nTopos = 0;
210 +      for (int j = 0; j < nLocal_; j++) {
211 +        int jglob = AtomLocalToGlobal[j];        
212 +        if (oneTwo.hasPair(iglob, jglob)) {
213 +          toposForLocalAtom[i].push_back(j);
214 +          topoDistLocal[i][nTopos] = 1;
215 +          nTopos++;
216 +        }
217 +        if (oneThree.hasPair(iglob, jglob)) {
218 +          toposForLocalAtom[i].push_back(j);
219 +          topoDistLocal[i][nTopos] = 2;
220 +          nTopos++;
221 +        }
222 +        if (oneFour.hasPair(iglob, jglob)) {
223 +          toposForLocalAtom[i].push_back(j);
224 +          topoDistLocal[i][nTopos] = 3;
225 +          nTopos++;
226 +        }
227 +      }      
228 +    }    
229 +
230    }
231 +  
232 +  void ForceMatrixDecomposition::createGtypeCutoffMap() {
233 +
234 +    RealType tol = 1e-6;
235 +    RealType rc;
236 +    int atid;
237 +    set<AtomType*> atypes = info_->getSimulatedAtomTypes();
238 +    vector<RealType> atypeCutoff;
239 +    atypeCutoff.resize( atypes.size() );
240 +
241 +    for (set<AtomType*>::iterator at = atypes.begin(); at != atypes.end(); ++at){
242 +      rc = interactionMan_->getSuggestedCutoffRadius(*at);
243 +      atid = (*at)->getIdent();
244 +      atypeCutoff[atid] = rc;
245 +    }
246 +
247 +    vector<RealType> gTypeCutoffs;
248 +
249 +    // first we do a single loop over the cutoff groups to find the
250 +    // largest cutoff for any atypes present in this group.
251 + #ifdef IS_MPI
252 +    vector<RealType> groupCutoffRow(nGroupsInRow_, 0.0);
253 +    for (int cg1 = 0; cg1 < nGroupsInRow_; cg1++) {
254 +      vector<int> atomListRow = getAtomsInGroupRow(cg1);
255 +      for (vector<int>::iterator ia = atomListRow.begin();
256 +           ia != atomListRow.end(); ++ia) {            
257 +        int atom1 = (*ia);
258 +        atid = identsRow[atom1];
259 +        if (atypeCutoff[atid] > groupCutoffRow[cg1]) {
260 +          groupCutoffRow[cg1] = atypeCutoff[atid];
261 +        }
262 +      }
263 +
264 +      bool gTypeFound = false;
265 +      for (int gt = 0; gt < gTypeCutoffs.size(); gt++) {
266 +        if (abs(groupCutoffRow[cg1] - gTypeCutoffs[gt]) < tol) {
267 +          groupRowToGtype[cg1] = gt;
268 +          gTypeFound = true;
269 +        }
270 +      }
271 +      if (!gTypeFound) {
272 +        gTypeCutoffs.push_back( groupCutoffRow[cg1] );
273 +        groupRowToGtype[cg1] = gTypeCutoffs.size() - 1;
274 +      }
275 +      
276 +    }
277 +    vector<RealType> groupCutoffCol(nGroupsInCol_, 0.0);
278 +    for (int cg2 = 0; cg2 < nGroupsInCol_; cg2++) {
279 +      vector<int> atomListCol = getAtomsInGroupColumn(cg2);
280 +      for (vector<int>::iterator jb = atomListCol.begin();
281 +           jb != atomListCol.end(); ++jb) {            
282 +        int atom2 = (*jb);
283 +        atid = identsCol[atom2];
284 +        if (atypeCutoff[atid] > groupCutoffCol[cg2]) {
285 +          groupCutoffCol[cg2] = atypeCutoff[atid];
286 +        }
287 +      }
288 +      bool gTypeFound = false;
289 +      for (int gt = 0; gt < gTypeCutoffs.size(); gt++) {
290 +        if (abs(groupCutoffCol[cg2] - gTypeCutoffs[gt]) < tol) {
291 +          groupColToGtype[cg2] = gt;
292 +          gTypeFound = true;
293 +        }
294 +      }
295 +      if (!gTypeFound) {
296 +        gTypeCutoffs.push_back( groupCutoffCol[cg2] );
297 +        groupColToGtype[cg2] = gTypeCutoffs.size() - 1;
298 +      }
299 +    }
300 + #else
301 +    vector<RealType> groupCutoff(nGroups_, 0.0);
302 +    for (int cg1 = 0; cg1 < nGroups_; cg1++) {
303 +      groupCutoff[cg1] = 0.0;
304 +      vector<int> atomList = getAtomsInGroupRow(cg1);
305 +      for (vector<int>::iterator ia = atomList.begin();
306 +           ia != atomList.end(); ++ia) {            
307 +        int atom1 = (*ia);
308 +        atid = identsLocal[atom1];
309 +        if (atypeCutoff[atid] > groupCutoff[cg1]) {
310 +          groupCutoff[cg1] = atypeCutoff[atid];
311 +        }
312 +      }
313 +
314 +      bool gTypeFound = false;
315 +      for (int gt = 0; gt < gTypeCutoffs.size(); gt++) {
316 +        if (abs(groupCutoff[cg1] - gTypeCutoffs[gt]) < tol) {
317 +          groupToGtype[cg1] = gt;
318 +          gTypeFound = true;
319 +        }
320 +      }
321 +      if (!gTypeFound) {
322 +        gTypeCutoffs.push_back( groupCutoff[cg1] );
323 +        groupToGtype[cg1] = gTypeCutoffs.size() - 1;
324 +      }      
325 +    }
326 + #endif
327 +
328 +    // Now we find the maximum group cutoff value present in the simulation
329 +
330 +    vector<RealType>::iterator groupMaxLoc = max_element(gTypeCutoffs.begin(), gTypeCutoffs.end());
331 +    RealType groupMax = *groupMaxLoc;
332 +
333 + #ifdef IS_MPI
334 +    MPI::COMM_WORLD.Allreduce(&groupMax, &groupMax, 1, MPI::REALTYPE, MPI::MAX);
335 + #endif
336      
337 +    RealType tradRcut = groupMax;
338  
339 +    for (int i = 0; i < gTypeCutoffs.size();  i++) {
340 +      for (int j = 0; j < gTypeCutoffs.size();  j++) {
341 +        
342 +        RealType thisRcut;
343 +        switch(cutoffPolicy_) {
344 +        case TRADITIONAL:
345 +          thisRcut = tradRcut;
346 +        case MIX:
347 +          thisRcut = 0.5 * (gTypeCutoffs[i] + gTypeCutoffs[j]);
348 +        case MAX:
349 +          thisRcut = max(gTypeCutoffs[i], gTypeCutoffs[j]);
350 +        default:
351 +          sprintf(painCave.errMsg,
352 +                  "ForceMatrixDecomposition::createGtypeCutoffMap "
353 +                  "hit an unknown cutoff policy!\n");
354 +          painCave.severity = OPENMD_ERROR;
355 +          painCave.isFatal = 1;
356 +          simError();              
357 +        }
358  
359 +        pair<int,int> key = make_pair(i,j);
360 +        gTypeCutoffMap[key].first = thisRcut;
361 +
362 +        if (thisRcut > largestRcut_) largestRcut_ = thisRcut;
363 +
364 +        gTypeCutoffMap[key].second = thisRcut*thisRcut;
365 +        
366 +        gTypeCutoffMap[key].third = pow(thisRcut + skinThickness_, 2);
367 +
368 +        // sanity check
369 +        
370 +        if (userChoseCutoff_) {
371 +          if (abs(gTypeCutoffMap[key].first - userCutoff_) > 0.0001) {
372 +            sprintf(painCave.errMsg,
373 +                    "ForceMatrixDecomposition::createGtypeCutoffMap "
374 +                    "user-specified rCut does not match computed group Cutoff\n");
375 +            painCave.severity = OPENMD_ERROR;
376 +            painCave.isFatal = 1;
377 +            simError();            
378 +          }
379 +        }
380 +      }
381 +    }
382 +  }
383 +
384 +
385 +  groupCutoffs ForceMatrixDecomposition::getGroupCutoffs(int cg1, int cg2) {
386 +    int i, j;
387 +
388 + #ifdef IS_MPI
389 +    i = groupRowToGtype[cg1];
390 +    j = groupColToGtype[cg2];
391 + #else
392 +    i = groupToGtype[cg1];
393 +    j = groupToGtype[cg2];
394 + #endif
395 +    
396 +    return gTypeCutoffMap[make_pair(i,j)];
397 +  }
398 +
399 +
400 +  void ForceMatrixDecomposition::zeroWorkArrays() {
401 +
402 +    for (int j = 0; j < N_INTERACTION_FAMILIES; j++) {
403 +      longRangePot_[j] = 0.0;
404 +    }
405 +
406 + #ifdef IS_MPI
407 +    if (storageLayout_ & DataStorage::dslForce) {
408 +      fill(atomRowData.force.begin(), atomRowData.force.end(), V3Zero);
409 +      fill(atomColData.force.begin(), atomColData.force.end(), V3Zero);
410 +    }
411 +
412 +    if (storageLayout_ & DataStorage::dslTorque) {
413 +      fill(atomRowData.torque.begin(), atomRowData.torque.end(), V3Zero);
414 +      fill(atomColData.torque.begin(), atomColData.torque.end(), V3Zero);
415 +    }
416 +    
417 +    fill(pot_row.begin(), pot_row.end(),
418 +         Vector<RealType, N_INTERACTION_FAMILIES> (0.0));
419 +
420 +    fill(pot_col.begin(), pot_col.end(),
421 +         Vector<RealType, N_INTERACTION_FAMILIES> (0.0));
422 +    
423 +    pot_local = Vector<RealType, N_INTERACTION_FAMILIES>(0.0);
424 +
425 +    if (storageLayout_ & DataStorage::dslParticlePot) {    
426 +      fill(atomRowData.particlePot.begin(), atomRowData.particlePot.end(), 0.0);
427 +      fill(atomColData.particlePot.begin(), atomColData.particlePot.end(), 0.0);
428 +    }
429 +
430 +    if (storageLayout_ & DataStorage::dslDensity) {      
431 +      fill(atomRowData.density.begin(), atomRowData.density.end(), 0.0);
432 +      fill(atomColData.density.begin(), atomColData.density.end(), 0.0);
433 +    }
434 +
435 +    if (storageLayout_ & DataStorage::dslFunctional) {  
436 +      fill(atomRowData.functional.begin(), atomRowData.functional.end(), 0.0);
437 +      fill(atomColData.functional.begin(), atomColData.functional.end(), 0.0);
438 +    }
439 +
440 +    if (storageLayout_ & DataStorage::dslFunctionalDerivative) {      
441 +      fill(atomRowData.functionalDerivative.begin(),
442 +           atomRowData.functionalDerivative.end(), 0.0);
443 +      fill(atomColData.functionalDerivative.begin(),
444 +           atomColData.functionalDerivative.end(), 0.0);
445 +    }
446 +
447 + #else
448 +    
449 +    if (storageLayout_ & DataStorage::dslParticlePot) {      
450 +      fill(snap_->atomData.particlePot.begin(),
451 +           snap_->atomData.particlePot.end(), 0.0);
452 +    }
453 +    
454 +    if (storageLayout_ & DataStorage::dslDensity) {      
455 +      fill(snap_->atomData.density.begin(),
456 +           snap_->atomData.density.end(), 0.0);
457 +    }
458 +    if (storageLayout_ & DataStorage::dslFunctional) {
459 +      fill(snap_->atomData.functional.begin(),
460 +           snap_->atomData.functional.end(), 0.0);
461 +    }
462 +    if (storageLayout_ & DataStorage::dslFunctionalDerivative) {      
463 +      fill(snap_->atomData.functionalDerivative.begin(),
464 +           snap_->atomData.functionalDerivative.end(), 0.0);
465 +    }
466 + #endif
467 +    
468 +  }
469 +
470 +
471    void ForceMatrixDecomposition::distributeData()  {
472      snap_ = sman_->getCurrentSnapshot();
473      storageLayout_ = sman_->getStorageLayout();
# Line 155 | Line 503 | namespace OpenMD {
503   #endif      
504    }
505    
506 +  /* collects information obtained during the pre-pair loop onto local
507 +   * data structures.
508 +   */
509    void ForceMatrixDecomposition::collectIntermediateData() {
510      snap_ = sman_->getCurrentSnapshot();
511      storageLayout_ = sman_->getStorageLayout();
# Line 166 | Line 517 | namespace OpenMD {
517                                 snap_->atomData.density);
518        
519        int n = snap_->atomData.density.size();
520 <      std::vector<RealType> rho_tmp(n, 0.0);
520 >      vector<RealType> rho_tmp(n, 0.0);
521        AtomCommRealColumn->scatter(atomColData.density, rho_tmp);
522        for (int i = 0; i < n; i++)
523          snap_->atomData.density[i] += rho_tmp[i];
524      }
525   #endif
526    }
527 <  
527 >
528 >  /*
529 >   * redistributes information obtained during the pre-pair loop out to
530 >   * row and column-indexed data structures
531 >   */
532    void ForceMatrixDecomposition::distributeIntermediateData() {
533      snap_ = sman_->getCurrentSnapshot();
534      storageLayout_ = sman_->getStorageLayout();
# Line 229 | Line 584 | namespace OpenMD {
584          snap_->atomData.torque[i] += trq_tmp[i];
585      }
586      
587 <    int nLocal = snap_->getNumberOfAtoms();
587 >    nLocal_ = snap_->getNumberOfAtoms();
588  
589 <    vector<vector<RealType> > pot_temp(N_INTERACTION_FAMILIES,
590 <                                       vector<RealType> (nLocal, 0.0));
589 >    vector<potVec> pot_temp(nLocal_,
590 >                            Vector<RealType, N_INTERACTION_FAMILIES> (0.0));
591 >
592 >    // scatter/gather pot_row into the members of my column
593 >          
594 >    AtomCommPotRow->scatter(pot_row, pot_temp);
595 >
596 >    for (int ii = 0;  ii < pot_temp.size(); ii++ )
597 >      pot_local += pot_temp[ii];
598      
599 <    for (int i = 0; i < N_INTERACTION_FAMILIES; i++) {
600 <      AtomCommRealRow->scatter(pot_row[i], pot_temp[i]);
601 <      for (int ii = 0;  ii < pot_temp[i].size(); ii++ ) {
602 <        pot_local[i] += pot_temp[i][ii];
603 <      }
604 <    }
599 >    fill(pot_temp.begin(), pot_temp.end(),
600 >         Vector<RealType, N_INTERACTION_FAMILIES> (0.0));
601 >      
602 >    AtomCommPotColumn->scatter(pot_col, pot_temp);    
603 >    
604 >    for (int ii = 0;  ii < pot_temp.size(); ii++ )
605 >      pot_local += pot_temp[ii];
606 >    
607   #endif
608    }
609  
610 +  int ForceMatrixDecomposition::getNAtomsInRow() {  
611 + #ifdef IS_MPI
612 +    return nAtomsInRow_;
613 + #else
614 +    return nLocal_;
615 + #endif
616 +  }
617 +
618 +  /**
619 +   * returns the list of atoms belonging to this group.  
620 +   */
621 +  vector<int> ForceMatrixDecomposition::getAtomsInGroupRow(int cg1){
622 + #ifdef IS_MPI
623 +    return groupListRow_[cg1];
624 + #else
625 +    return groupList_[cg1];
626 + #endif
627 +  }
628 +
629 +  vector<int> ForceMatrixDecomposition::getAtomsInGroupColumn(int cg2){
630 + #ifdef IS_MPI
631 +    return groupListCol_[cg2];
632 + #else
633 +    return groupList_[cg2];
634 + #endif
635 +  }
636    
637    Vector3d ForceMatrixDecomposition::getIntergroupVector(int cg1, int cg2){
638      Vector3d d;
# Line 284 | Line 674 | namespace OpenMD {
674      snap_->wrapVector(d);
675      return d;    
676    }
677 +
678 +  RealType ForceMatrixDecomposition::getMassFactorRow(int atom1) {
679 + #ifdef IS_MPI
680 +    return massFactorsRow[atom1];
681 + #else
682 +    return massFactorsLocal[atom1];
683 + #endif
684 +  }
685 +
686 +  RealType ForceMatrixDecomposition::getMassFactorColumn(int atom2) {
687 + #ifdef IS_MPI
688 +    return massFactorsCol[atom2];
689 + #else
690 +    return massFactorsLocal[atom2];
691 + #endif
692 +
693 +  }
694      
695    Vector3d ForceMatrixDecomposition::getInteratomicVector(int atom1, int atom2){
696      Vector3d d;
# Line 298 | Line 705 | namespace OpenMD {
705      return d;    
706    }
707  
708 +  vector<int> ForceMatrixDecomposition::getSkipsForRowAtom(int atom1) {
709 + #ifdef IS_MPI
710 +    return skipsForRowAtom[atom1];
711 + #else
712 +    return skipsForLocalAtom[atom1];
713 + #endif
714 +  }
715 +
716 +  /**
717 +   * There are a number of reasons to skip a pair or a
718 +   * particle. Mostly we do this to exclude atoms who are involved in
719 +   * short range interactions (bonds, bends, torsions), but we also
720 +   * need to exclude some overcounted interactions that result from
721 +   * the parallel decomposition.
722 +   */
723 +  bool ForceMatrixDecomposition::skipAtomPair(int atom1, int atom2) {
724 +    int unique_id_1, unique_id_2;
725 +
726 + #ifdef IS_MPI
727 +    // in MPI, we have to look up the unique IDs for each atom
728 +    unique_id_1 = AtomRowToGlobal[atom1];
729 +    unique_id_2 = AtomColToGlobal[atom2];
730 +
731 +    // this situation should only arise in MPI simulations
732 +    if (unique_id_1 == unique_id_2) return true;
733 +    
734 +    // this prevents us from doing the pair on multiple processors
735 +    if (unique_id_1 < unique_id_2) {
736 +      if ((unique_id_1 + unique_id_2) % 2 == 0) return true;
737 +    } else {
738 +      if ((unique_id_1 + unique_id_2) % 2 == 1) return true;
739 +    }
740 + #else
741 +    // in the normal loop, the atom numbers are unique
742 +    unique_id_1 = atom1;
743 +    unique_id_2 = atom2;
744 + #endif
745 +    
746 + #ifdef IS_MPI
747 +    for (vector<int>::iterator i = skipsForRowAtom[atom1].begin();
748 +         i != skipsForRowAtom[atom1].end(); ++i) {
749 +      if ( (*i) == unique_id_2 ) return true;
750 +    }    
751 + #else
752 +    for (vector<int>::iterator i = skipsForLocalAtom[atom1].begin();
753 +         i != skipsForLocalAtom[atom1].end(); ++i) {
754 +      if ( (*i) == unique_id_2 ) return true;
755 +    }    
756 + #endif
757 +  }
758 +
759 +  int ForceMatrixDecomposition::getTopoDistance(int atom1, int atom2) {
760 +    
761 + #ifdef IS_MPI
762 +    for (int i = 0; i < toposForRowAtom[atom1].size(); i++) {
763 +      if ( toposForRowAtom[atom1][i] == atom2 ) return topoDistRow[atom1][i];
764 +    }
765 + #else
766 +    for (int i = 0; i < toposForLocalAtom[atom1].size(); i++) {
767 +      if ( toposForLocalAtom[atom1][i] == atom2 ) return topoDistLocal[atom1][i];
768 +    }
769 + #endif
770 +
771 +    // zero is default for unconnected (i.e. normal) pair interactions
772 +    return 0;
773 +  }
774 +
775    void ForceMatrixDecomposition::addForceToAtomRow(int atom1, Vector3d fg){
776   #ifdef IS_MPI
777      atomRowData.force[atom1] += fg;
# Line 312 | Line 786 | namespace OpenMD {
786   #else
787      snap_->atomData.force[atom2] += fg;
788   #endif
315 –
789    }
790  
791      // filling interaction blocks with pointers
792    InteractionData ForceMatrixDecomposition::fillInteractionData(int atom1, int atom2) {    
320 –
793      InteractionData idat;
794 +
795   #ifdef IS_MPI
796 +    
797 +    idat.atypes = make_pair( ff_->getAtomType(identsRow[atom1]),
798 +                             ff_->getAtomType(identsCol[atom2]) );
799 +
800 +    
801      if (storageLayout_ & DataStorage::dslAmat) {
802 <      idat.A1 = atomRowData.aMat[atom1];
803 <      idat.A2 = atomColData.aMat[atom2];
802 >      idat.A1 = &(atomRowData.aMat[atom1]);
803 >      idat.A2 = &(atomColData.aMat[atom2]);
804      }
805 +    
806 +    if (storageLayout_ & DataStorage::dslElectroFrame) {
807 +      idat.eFrame1 = &(atomRowData.electroFrame[atom1]);
808 +      idat.eFrame2 = &(atomColData.electroFrame[atom2]);
809 +    }
810  
811 +    if (storageLayout_ & DataStorage::dslTorque) {
812 +      idat.t1 = &(atomRowData.torque[atom1]);
813 +      idat.t2 = &(atomColData.torque[atom2]);
814 +    }
815 +
816 +    if (storageLayout_ & DataStorage::dslDensity) {
817 +      idat.rho1 = &(atomRowData.density[atom1]);
818 +      idat.rho2 = &(atomColData.density[atom2]);
819 +    }
820 +
821 +    if (storageLayout_ & DataStorage::dslFunctional) {
822 +      idat.frho1 = &(atomRowData.functional[atom1]);
823 +      idat.frho2 = &(atomColData.functional[atom2]);
824 +    }
825 +
826 +    if (storageLayout_ & DataStorage::dslFunctionalDerivative) {
827 +      idat.dfrho1 = &(atomRowData.functionalDerivative[atom1]);
828 +      idat.dfrho2 = &(atomColData.functionalDerivative[atom2]);
829 +    }
830 +
831 +    if (storageLayout_ & DataStorage::dslParticlePot) {
832 +      idat.particlePot1 = &(atomRowData.particlePot[atom1]);
833 +      idat.particlePot2 = &(atomColData.particlePot[atom2]);
834 +    }
835 +
836 + #else
837 +
838 +    idat.atypes = make_pair( ff_->getAtomType(identsLocal[atom1]),
839 +                             ff_->getAtomType(identsLocal[atom2]) );
840 +
841 +    if (storageLayout_ & DataStorage::dslAmat) {
842 +      idat.A1 = &(snap_->atomData.aMat[atom1]);
843 +      idat.A2 = &(snap_->atomData.aMat[atom2]);
844 +    }
845 +
846      if (storageLayout_ & DataStorage::dslElectroFrame) {
847 <      idat.eFrame1 = atomRowData.electroFrame[atom1];
848 <      idat.eFrame2 = atomColData.electroFrame[atom2];
847 >      idat.eFrame1 = &(snap_->atomData.electroFrame[atom1]);
848 >      idat.eFrame2 = &(snap_->atomData.electroFrame[atom2]);
849      }
850  
851      if (storageLayout_ & DataStorage::dslTorque) {
852 <      idat.t1 = atomRowData.torque[atom1];
853 <      idat.t2 = atomColData.torque[atom2];
852 >      idat.t1 = &(snap_->atomData.torque[atom1]);
853 >      idat.t2 = &(snap_->atomData.torque[atom2]);
854      }
855  
856      if (storageLayout_ & DataStorage::dslDensity) {
857 <      idat.rho1 = atomRowData.density[atom1];
858 <      idat.rho2 = atomColData.density[atom2];
857 >      idat.rho1 = &(snap_->atomData.density[atom1]);
858 >      idat.rho2 = &(snap_->atomData.density[atom2]);
859      }
860  
861 +    if (storageLayout_ & DataStorage::dslFunctional) {
862 +      idat.frho1 = &(snap_->atomData.functional[atom1]);
863 +      idat.frho2 = &(snap_->atomData.functional[atom2]);
864 +    }
865 +
866      if (storageLayout_ & DataStorage::dslFunctionalDerivative) {
867 <      idat.dfrho1 = atomRowData.functionalDerivative[atom1];
868 <      idat.dfrho2 = atomColData.functionalDerivative[atom2];
867 >      idat.dfrho1 = &(snap_->atomData.functionalDerivative[atom1]);
868 >      idat.dfrho2 = &(snap_->atomData.functionalDerivative[atom2]);
869      }
870 +
871 +    if (storageLayout_ & DataStorage::dslParticlePot) {
872 +      idat.particlePot1 = &(snap_->atomData.particlePot[atom1]);
873 +      idat.particlePot2 = &(snap_->atomData.particlePot[atom2]);
874 +    }
875 +
876   #endif
877 +    return idat;
878 +  }
879 +
880 +  
881 +  void ForceMatrixDecomposition::unpackInteractionData(InteractionData idat, int atom1, int atom2) {    
882 + #ifdef IS_MPI
883 +    pot_row[atom1] += 0.5 *  *(idat.pot);
884 +    pot_col[atom2] += 0.5 *  *(idat.pot);
885 +
886 +    atomRowData.force[atom1] += *(idat.f1);
887 +    atomColData.force[atom2] -= *(idat.f1);
888 + #else
889 +    longRangePot_ += *(idat.pot);
890      
891 +    snap_->atomData.force[atom1] += *(idat.f1);
892 +    snap_->atomData.force[atom2] -= *(idat.f1);
893 + #endif
894 +
895    }
896 +
897 +
898    InteractionData ForceMatrixDecomposition::fillSkipData(int atom1, int atom2){
899 +
900 +    InteractionData idat;
901 + #ifdef IS_MPI
902 +    idat.atypes = make_pair( ff_->getAtomType(identsRow[atom1]),
903 +                             ff_->getAtomType(identsCol[atom2]) );
904 +
905 +    if (storageLayout_ & DataStorage::dslElectroFrame) {
906 +      idat.eFrame1 = &(atomRowData.electroFrame[atom1]);
907 +      idat.eFrame2 = &(atomColData.electroFrame[atom2]);
908 +    }
909 +    if (storageLayout_ & DataStorage::dslTorque) {
910 +      idat.t1 = &(atomRowData.torque[atom1]);
911 +      idat.t2 = &(atomColData.torque[atom2]);
912 +    }
913 + #else
914 +    idat.atypes = make_pair( ff_->getAtomType(identsLocal[atom1]),
915 +                             ff_->getAtomType(identsLocal[atom2]) );
916 +
917 +    if (storageLayout_ & DataStorage::dslElectroFrame) {
918 +      idat.eFrame1 = &(snap_->atomData.electroFrame[atom1]);
919 +      idat.eFrame2 = &(snap_->atomData.electroFrame[atom2]);
920 +    }
921 +    if (storageLayout_ & DataStorage::dslTorque) {
922 +      idat.t1 = &(snap_->atomData.torque[atom1]);
923 +      idat.t2 = &(snap_->atomData.torque[atom2]);
924 +    }
925 + #endif    
926    }
352 –  SelfData ForceMatrixDecomposition::fillSelfData(int atom1) {
353 –  }
927  
928 <  
928 >  /*
929 >   * buildNeighborList
930 >   *
931 >   * first element of pair is row-indexed CutoffGroup
932 >   * second element of pair is column-indexed CutoffGroup
933 >   */
934 >  vector<pair<int, int> > ForceMatrixDecomposition::buildNeighborList() {
935 >      
936 >    vector<pair<int, int> > neighborList;
937 >    groupCutoffs cuts;
938 > #ifdef IS_MPI
939 >    cellListRow_.clear();
940 >    cellListCol_.clear();
941 > #else
942 >    cellList_.clear();
943 > #endif
944 >
945 >    RealType rList_ = (largestRcut_ + skinThickness_);
946 >    RealType rl2 = rList_ * rList_;
947 >    Snapshot* snap_ = sman_->getCurrentSnapshot();
948 >    Mat3x3d Hmat = snap_->getHmat();
949 >    Vector3d Hx = Hmat.getColumn(0);
950 >    Vector3d Hy = Hmat.getColumn(1);
951 >    Vector3d Hz = Hmat.getColumn(2);
952 >
953 >    nCells_.x() = (int) ( Hx.length() )/ rList_;
954 >    nCells_.y() = (int) ( Hy.length() )/ rList_;
955 >    nCells_.z() = (int) ( Hz.length() )/ rList_;
956 >
957 >    Mat3x3d invHmat = snap_->getInvHmat();
958 >    Vector3d rs, scaled, dr;
959 >    Vector3i whichCell;
960 >    int cellIndex;
961 >
962 > #ifdef IS_MPI
963 >    for (int i = 0; i < nGroupsInRow_; i++) {
964 >      rs = cgRowData.position[i];
965 >      // scaled positions relative to the box vectors
966 >      scaled = invHmat * rs;
967 >      // wrap the vector back into the unit box by subtracting integer box
968 >      // numbers
969 >      for (int j = 0; j < 3; j++)
970 >        scaled[j] -= roundMe(scaled[j]);
971 >    
972 >      // find xyz-indices of cell that cutoffGroup is in.
973 >      whichCell.x() = nCells_.x() * scaled.x();
974 >      whichCell.y() = nCells_.y() * scaled.y();
975 >      whichCell.z() = nCells_.z() * scaled.z();
976 >
977 >      // find single index of this cell:
978 >      cellIndex = Vlinear(whichCell, nCells_);
979 >      // add this cutoff group to the list of groups in this cell;
980 >      cellListRow_[cellIndex].push_back(i);
981 >    }
982 >
983 >    for (int i = 0; i < nGroupsInCol_; i++) {
984 >      rs = cgColData.position[i];
985 >      // scaled positions relative to the box vectors
986 >      scaled = invHmat * rs;
987 >      // wrap the vector back into the unit box by subtracting integer box
988 >      // numbers
989 >      for (int j = 0; j < 3; j++)
990 >        scaled[j] -= roundMe(scaled[j]);
991 >
992 >      // find xyz-indices of cell that cutoffGroup is in.
993 >      whichCell.x() = nCells_.x() * scaled.x();
994 >      whichCell.y() = nCells_.y() * scaled.y();
995 >      whichCell.z() = nCells_.z() * scaled.z();
996 >
997 >      // find single index of this cell:
998 >      cellIndex = Vlinear(whichCell, nCells_);
999 >      // add this cutoff group to the list of groups in this cell;
1000 >      cellListCol_[cellIndex].push_back(i);
1001 >    }
1002 > #else
1003 >    for (int i = 0; i < nGroups_; i++) {
1004 >      rs = snap_->cgData.position[i];
1005 >      // scaled positions relative to the box vectors
1006 >      scaled = invHmat * rs;
1007 >      // wrap the vector back into the unit box by subtracting integer box
1008 >      // numbers
1009 >      for (int j = 0; j < 3; j++)
1010 >        scaled[j] -= roundMe(scaled[j]);
1011 >
1012 >      // find xyz-indices of cell that cutoffGroup is in.
1013 >      whichCell.x() = nCells_.x() * scaled.x();
1014 >      whichCell.y() = nCells_.y() * scaled.y();
1015 >      whichCell.z() = nCells_.z() * scaled.z();
1016 >
1017 >      // find single index of this cell:
1018 >      cellIndex = Vlinear(whichCell, nCells_);
1019 >      // add this cutoff group to the list of groups in this cell;
1020 >      cellList_[cellIndex].push_back(i);
1021 >    }
1022 > #endif
1023 >
1024 >    for (int m1z = 0; m1z < nCells_.z(); m1z++) {
1025 >      for (int m1y = 0; m1y < nCells_.y(); m1y++) {
1026 >        for (int m1x = 0; m1x < nCells_.x(); m1x++) {
1027 >          Vector3i m1v(m1x, m1y, m1z);
1028 >          int m1 = Vlinear(m1v, nCells_);
1029 >
1030 >          for (vector<Vector3i>::iterator os = cellOffsets_.begin();
1031 >               os != cellOffsets_.end(); ++os) {
1032 >            
1033 >            Vector3i m2v = m1v + (*os);
1034 >            
1035 >            if (m2v.x() >= nCells_.x()) {
1036 >              m2v.x() = 0;          
1037 >            } else if (m2v.x() < 0) {
1038 >              m2v.x() = nCells_.x() - 1;
1039 >            }
1040 >            
1041 >            if (m2v.y() >= nCells_.y()) {
1042 >              m2v.y() = 0;          
1043 >            } else if (m2v.y() < 0) {
1044 >              m2v.y() = nCells_.y() - 1;
1045 >            }
1046 >            
1047 >            if (m2v.z() >= nCells_.z()) {
1048 >              m2v.z() = 0;          
1049 >            } else if (m2v.z() < 0) {
1050 >              m2v.z() = nCells_.z() - 1;
1051 >            }
1052 >            
1053 >            int m2 = Vlinear (m2v, nCells_);
1054 >
1055 > #ifdef IS_MPI
1056 >            for (vector<int>::iterator j1 = cellListRow_[m1].begin();
1057 >                 j1 != cellListRow_[m1].end(); ++j1) {
1058 >              for (vector<int>::iterator j2 = cellListCol_[m2].begin();
1059 >                   j2 != cellListCol_[m2].end(); ++j2) {
1060 >                              
1061 >                // Always do this if we're in different cells or if
1062 >                // we're in the same cell and the global index of the
1063 >                // j2 cutoff group is less than the j1 cutoff group
1064 >
1065 >                if (m2 != m1 || cgColToGlobal[(*j2)] < cgRowToGlobal[(*j1)]) {
1066 >                  dr = cgColData.position[(*j2)] - cgRowData.position[(*j1)];
1067 >                  snap_->wrapVector(dr);
1068 >                  cuts = getGroupCutoffs( (*j1), (*j2) );
1069 >                  if (dr.lengthSquare() < cuts.third) {
1070 >                    neighborList.push_back(make_pair((*j1), (*j2)));
1071 >                  }
1072 >                }
1073 >              }
1074 >            }
1075 > #else
1076 >            for (vector<int>::iterator j1 = cellList_[m1].begin();
1077 >                 j1 != cellList_[m1].end(); ++j1) {
1078 >              for (vector<int>::iterator j2 = cellList_[m2].begin();
1079 >                   j2 != cellList_[m2].end(); ++j2) {
1080 >                              
1081 >                // Always do this if we're in different cells or if
1082 >                // we're in the same cell and the global index of the
1083 >                // j2 cutoff group is less than the j1 cutoff group
1084 >
1085 >                if (m2 != m1 || (*j2) < (*j1)) {
1086 >                  dr = snap_->cgData.position[(*j2)] - snap_->cgData.position[(*j1)];
1087 >                  snap_->wrapVector(dr);
1088 >                  cuts = getGroupCutoffs( (*j1), (*j2) );
1089 >                  if (dr.lengthSquare() < cuts.third) {
1090 >                    neighborList.push_back(make_pair((*j1), (*j2)));
1091 >                  }
1092 >                }
1093 >              }
1094 >            }
1095 > #endif
1096 >          }
1097 >        }
1098 >      }
1099 >    }
1100 >
1101 >    // save the local cutoff group positions for the check that is
1102 >    // done on each loop:
1103 >    saved_CG_positions_.clear();
1104 >    for (int i = 0; i < nGroups_; i++)
1105 >      saved_CG_positions_.push_back(snap_->cgData.position[i]);
1106 >
1107 >    return neighborList;
1108 >  }
1109   } //end namespace OpenMD

Diff Legend

– Removed lines
+ Added lines
< Changed lines
> Changed lines