38 |
|
* [3] Sun, Lin & Gezelter, J. Chem. Phys. 128, 24107 (2008). |
39 |
|
* [4] Vardeman & Gezelter, in progress (2009). |
40 |
|
*/ |
41 |
< |
#include "parallel/ForceDecomposition.hpp" |
42 |
< |
#include "parallel/Communicator.hpp" |
41 |
> |
#include "parallel/ForceMatrixDecomposition.hpp" |
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 { |
49 |
|
|
50 |
< |
void ForceDecomposition::distributeInitialData() { |
50 |
> |
ForceMatrixDecomposition::ForceMatrixDecomposition(SimInfo* info, InteractionManager* iMan) : ForceDecomposition(info, iMan) { |
51 |
> |
|
52 |
> |
// In a parallel computation, row and colum scans must visit all |
53 |
> |
// surrounding cells (not just the 14 upper triangular blocks that |
54 |
> |
// are used when the processor can see all pairs) |
55 |
|
#ifdef IS_MPI |
56 |
+ |
cellOffsets_.clear(); |
57 |
+ |
cellOffsets_.push_back( Vector3i(-1,-1,-1) ); |
58 |
+ |
cellOffsets_.push_back( Vector3i( 0,-1,-1) ); |
59 |
+ |
cellOffsets_.push_back( Vector3i( 1,-1,-1) ); |
60 |
+ |
cellOffsets_.push_back( Vector3i(-1, 0,-1) ); |
61 |
+ |
cellOffsets_.push_back( Vector3i( 0, 0,-1) ); |
62 |
+ |
cellOffsets_.push_back( Vector3i( 1, 0,-1) ); |
63 |
+ |
cellOffsets_.push_back( Vector3i(-1, 1,-1) ); |
64 |
+ |
cellOffsets_.push_back( Vector3i( 0, 1,-1) ); |
65 |
+ |
cellOffsets_.push_back( Vector3i( 1, 1,-1) ); |
66 |
+ |
cellOffsets_.push_back( Vector3i(-1,-1, 0) ); |
67 |
+ |
cellOffsets_.push_back( Vector3i( 0,-1, 0) ); |
68 |
+ |
cellOffsets_.push_back( Vector3i( 1,-1, 0) ); |
69 |
+ |
cellOffsets_.push_back( Vector3i(-1, 0, 0) ); |
70 |
+ |
cellOffsets_.push_back( Vector3i( 0, 0, 0) ); |
71 |
+ |
cellOffsets_.push_back( Vector3i( 1, 0, 0) ); |
72 |
+ |
cellOffsets_.push_back( Vector3i(-1, 1, 0) ); |
73 |
+ |
cellOffsets_.push_back( Vector3i( 0, 1, 0) ); |
74 |
+ |
cellOffsets_.push_back( Vector3i( 1, 1, 0) ); |
75 |
+ |
cellOffsets_.push_back( Vector3i(-1,-1, 1) ); |
76 |
+ |
cellOffsets_.push_back( Vector3i( 0,-1, 1) ); |
77 |
+ |
cellOffsets_.push_back( Vector3i( 1,-1, 1) ); |
78 |
+ |
cellOffsets_.push_back( Vector3i(-1, 0, 1) ); |
79 |
+ |
cellOffsets_.push_back( Vector3i( 0, 0, 1) ); |
80 |
+ |
cellOffsets_.push_back( Vector3i( 1, 0, 1) ); |
81 |
+ |
cellOffsets_.push_back( Vector3i(-1, 1, 1) ); |
82 |
+ |
cellOffsets_.push_back( Vector3i( 0, 1, 1) ); |
83 |
+ |
cellOffsets_.push_back( Vector3i( 1, 1, 1) ); |
84 |
+ |
#endif |
85 |
+ |
} |
86 |
|
|
50 |
– |
int nAtoms; |
51 |
– |
int nGroups; |
87 |
|
|
88 |
< |
AtomCommRealI = new Comm<I,RealType>(nAtoms); |
89 |
< |
AtomCommVectorI = new Comm<I,Vector3d>(nAtoms); |
90 |
< |
AtomCommMatrixI = new Comm<I,Mat3x3d>(nAtoms); |
88 |
> |
/** |
89 |
> |
* distributeInitialData is essentially a copy of the older fortran |
90 |
> |
* SimulationSetup |
91 |
> |
*/ |
92 |
> |
void ForceMatrixDecomposition::distributeInitialData() { |
93 |
> |
snap_ = sman_->getCurrentSnapshot(); |
94 |
> |
storageLayout_ = sman_->getStorageLayout(); |
95 |
> |
ff_ = info_->getForceField(); |
96 |
> |
nLocal_ = snap_->getNumberOfAtoms(); |
97 |
> |
|
98 |
> |
nGroups_ = info_->getNLocalCutoffGroups(); |
99 |
> |
// gather the information for atomtype IDs (atids): |
100 |
> |
idents = info_->getIdentArray(); |
101 |
> |
AtomLocalToGlobal = info_->getGlobalAtomIndices(); |
102 |
> |
cgLocalToGlobal = info_->getGlobalGroupIndices(); |
103 |
> |
vector<int> globalGroupMembership = info_->getGlobalGroupMembership(); |
104 |
|
|
105 |
< |
AtomCommRealJ = new Comm<J,RealType>(nAtoms); |
58 |
< |
AtomCommVectorJ = new Comm<J,Vector3d>(nAtoms); |
59 |
< |
AtomCommMatrixJ = new Comm<J,Mat3x3d>(nAtoms); |
105 |
> |
massFactors = info_->getMassFactors(); |
106 |
|
|
107 |
< |
cgCommVectorI = new Comm<I,Vector3d>(nGroups); |
108 |
< |
cgCommVectorJ = new Comm<J,Vector3d>(nGroups); |
109 |
< |
// more to come |
107 |
> |
PairList* excludes = info_->getExcludedInteractions(); |
108 |
> |
PairList* oneTwo = info_->getOneTwoInteractions(); |
109 |
> |
PairList* oneThree = info_->getOneThreeInteractions(); |
110 |
> |
PairList* oneFour = info_->getOneFourInteractions(); |
111 |
> |
|
112 |
> |
#ifdef IS_MPI |
113 |
> |
|
114 |
> |
MPI::Intracomm row = rowComm.getComm(); |
115 |
> |
MPI::Intracomm col = colComm.getComm(); |
116 |
> |
|
117 |
> |
AtomPlanIntRow = new Plan<int>(row, nLocal_); |
118 |
> |
AtomPlanRealRow = new Plan<RealType>(row, nLocal_); |
119 |
> |
AtomPlanVectorRow = new Plan<Vector3d>(row, nLocal_); |
120 |
> |
AtomPlanMatrixRow = new Plan<Mat3x3d>(row, nLocal_); |
121 |
> |
AtomPlanPotRow = new Plan<potVec>(row, nLocal_); |
122 |
> |
|
123 |
> |
AtomPlanIntColumn = new Plan<int>(col, nLocal_); |
124 |
> |
AtomPlanRealColumn = new Plan<RealType>(col, nLocal_); |
125 |
> |
AtomPlanVectorColumn = new Plan<Vector3d>(col, nLocal_); |
126 |
> |
AtomPlanMatrixColumn = new Plan<Mat3x3d>(col, nLocal_); |
127 |
> |
AtomPlanPotColumn = new Plan<potVec>(col, nLocal_); |
128 |
> |
|
129 |
> |
cgPlanIntRow = new Plan<int>(row, nGroups_); |
130 |
> |
cgPlanVectorRow = new Plan<Vector3d>(row, nGroups_); |
131 |
> |
cgPlanIntColumn = new Plan<int>(col, nGroups_); |
132 |
> |
cgPlanVectorColumn = new Plan<Vector3d>(col, nGroups_); |
133 |
> |
|
134 |
> |
nAtomsInRow_ = AtomPlanIntRow->getSize(); |
135 |
> |
nAtomsInCol_ = AtomPlanIntColumn->getSize(); |
136 |
> |
nGroupsInRow_ = cgPlanIntRow->getSize(); |
137 |
> |
nGroupsInCol_ = cgPlanIntColumn->getSize(); |
138 |
> |
|
139 |
> |
// Modify the data storage objects with the correct layouts and sizes: |
140 |
> |
atomRowData.resize(nAtomsInRow_); |
141 |
> |
atomRowData.setStorageLayout(storageLayout_); |
142 |
> |
atomColData.resize(nAtomsInCol_); |
143 |
> |
atomColData.setStorageLayout(storageLayout_); |
144 |
> |
cgRowData.resize(nGroupsInRow_); |
145 |
> |
cgRowData.setStorageLayout(DataStorage::dslPosition); |
146 |
> |
cgColData.resize(nGroupsInCol_); |
147 |
> |
cgColData.setStorageLayout(DataStorage::dslPosition); |
148 |
> |
|
149 |
> |
identsRow.resize(nAtomsInRow_); |
150 |
> |
identsCol.resize(nAtomsInCol_); |
151 |
> |
|
152 |
> |
AtomPlanIntRow->gather(idents, identsRow); |
153 |
> |
AtomPlanIntColumn->gather(idents, identsCol); |
154 |
> |
|
155 |
> |
// allocate memory for the parallel objects |
156 |
> |
atypesRow.resize(nAtomsInRow_); |
157 |
> |
atypesCol.resize(nAtomsInCol_); |
158 |
> |
|
159 |
> |
for (int i = 0; i < nAtomsInRow_; i++) |
160 |
> |
atypesRow[i] = ff_->getAtomType(identsRow[i]); |
161 |
> |
for (int i = 0; i < nAtomsInCol_; i++) |
162 |
> |
atypesCol[i] = ff_->getAtomType(identsCol[i]); |
163 |
> |
|
164 |
> |
pot_row.resize(nAtomsInRow_); |
165 |
> |
pot_col.resize(nAtomsInCol_); |
166 |
> |
|
167 |
> |
AtomRowToGlobal.resize(nAtomsInRow_); |
168 |
> |
AtomColToGlobal.resize(nAtomsInCol_); |
169 |
> |
AtomPlanIntRow->gather(AtomLocalToGlobal, AtomRowToGlobal); |
170 |
> |
AtomPlanIntColumn->gather(AtomLocalToGlobal, AtomColToGlobal); |
171 |
> |
|
172 |
> |
cgRowToGlobal.resize(nGroupsInRow_); |
173 |
> |
cgColToGlobal.resize(nGroupsInCol_); |
174 |
> |
cgPlanIntRow->gather(cgLocalToGlobal, cgRowToGlobal); |
175 |
> |
cgPlanIntColumn->gather(cgLocalToGlobal, cgColToGlobal); |
176 |
> |
|
177 |
> |
massFactorsRow.resize(nAtomsInRow_); |
178 |
> |
massFactorsCol.resize(nAtomsInCol_); |
179 |
> |
AtomPlanRealRow->gather(massFactors, massFactorsRow); |
180 |
> |
AtomPlanRealColumn->gather(massFactors, massFactorsCol); |
181 |
> |
|
182 |
> |
groupListRow_.clear(); |
183 |
> |
groupListRow_.resize(nGroupsInRow_); |
184 |
> |
for (int i = 0; i < nGroupsInRow_; i++) { |
185 |
> |
int gid = cgRowToGlobal[i]; |
186 |
> |
for (int j = 0; j < nAtomsInRow_; j++) { |
187 |
> |
int aid = AtomRowToGlobal[j]; |
188 |
> |
if (globalGroupMembership[aid] == gid) |
189 |
> |
groupListRow_[i].push_back(j); |
190 |
> |
} |
191 |
> |
} |
192 |
> |
|
193 |
> |
groupListCol_.clear(); |
194 |
> |
groupListCol_.resize(nGroupsInCol_); |
195 |
> |
for (int i = 0; i < nGroupsInCol_; i++) { |
196 |
> |
int gid = cgColToGlobal[i]; |
197 |
> |
for (int j = 0; j < nAtomsInCol_; j++) { |
198 |
> |
int aid = AtomColToGlobal[j]; |
199 |
> |
if (globalGroupMembership[aid] == gid) |
200 |
> |
groupListCol_[i].push_back(j); |
201 |
> |
} |
202 |
> |
} |
203 |
> |
|
204 |
> |
excludesForAtom.clear(); |
205 |
> |
excludesForAtom.resize(nAtomsInRow_); |
206 |
> |
toposForAtom.clear(); |
207 |
> |
toposForAtom.resize(nAtomsInRow_); |
208 |
> |
topoDist.clear(); |
209 |
> |
topoDist.resize(nAtomsInRow_); |
210 |
> |
for (int i = 0; i < nAtomsInRow_; i++) { |
211 |
> |
int iglob = AtomRowToGlobal[i]; |
212 |
> |
|
213 |
> |
for (int j = 0; j < nAtomsInCol_; j++) { |
214 |
> |
int jglob = AtomColToGlobal[j]; |
215 |
> |
|
216 |
> |
if (excludes->hasPair(iglob, jglob)) |
217 |
> |
excludesForAtom[i].push_back(j); |
218 |
> |
|
219 |
> |
if (oneTwo->hasPair(iglob, jglob)) { |
220 |
> |
toposForAtom[i].push_back(j); |
221 |
> |
topoDist[i].push_back(1); |
222 |
> |
} else { |
223 |
> |
if (oneThree->hasPair(iglob, jglob)) { |
224 |
> |
toposForAtom[i].push_back(j); |
225 |
> |
topoDist[i].push_back(2); |
226 |
> |
} else { |
227 |
> |
if (oneFour->hasPair(iglob, jglob)) { |
228 |
> |
toposForAtom[i].push_back(j); |
229 |
> |
topoDist[i].push_back(3); |
230 |
> |
} |
231 |
> |
} |
232 |
> |
} |
233 |
> |
} |
234 |
> |
} |
235 |
> |
|
236 |
|
#endif |
237 |
+ |
|
238 |
+ |
// allocate memory for the parallel objects |
239 |
+ |
atypesLocal.resize(nLocal_); |
240 |
+ |
|
241 |
+ |
for (int i = 0; i < nLocal_; i++) |
242 |
+ |
atypesLocal[i] = ff_->getAtomType(idents[i]); |
243 |
+ |
|
244 |
+ |
groupList_.clear(); |
245 |
+ |
groupList_.resize(nGroups_); |
246 |
+ |
for (int i = 0; i < nGroups_; i++) { |
247 |
+ |
int gid = cgLocalToGlobal[i]; |
248 |
+ |
for (int j = 0; j < nLocal_; j++) { |
249 |
+ |
int aid = AtomLocalToGlobal[j]; |
250 |
+ |
if (globalGroupMembership[aid] == gid) { |
251 |
+ |
groupList_[i].push_back(j); |
252 |
+ |
} |
253 |
+ |
} |
254 |
+ |
} |
255 |
+ |
|
256 |
+ |
excludesForAtom.clear(); |
257 |
+ |
excludesForAtom.resize(nLocal_); |
258 |
+ |
toposForAtom.clear(); |
259 |
+ |
toposForAtom.resize(nLocal_); |
260 |
+ |
topoDist.clear(); |
261 |
+ |
topoDist.resize(nLocal_); |
262 |
+ |
|
263 |
+ |
for (int i = 0; i < nLocal_; i++) { |
264 |
+ |
int iglob = AtomLocalToGlobal[i]; |
265 |
+ |
|
266 |
+ |
for (int j = 0; j < nLocal_; j++) { |
267 |
+ |
int jglob = AtomLocalToGlobal[j]; |
268 |
+ |
|
269 |
+ |
if (excludes->hasPair(iglob, jglob)) |
270 |
+ |
excludesForAtom[i].push_back(j); |
271 |
+ |
|
272 |
+ |
if (oneTwo->hasPair(iglob, jglob)) { |
273 |
+ |
toposForAtom[i].push_back(j); |
274 |
+ |
topoDist[i].push_back(1); |
275 |
+ |
} else { |
276 |
+ |
if (oneThree->hasPair(iglob, jglob)) { |
277 |
+ |
toposForAtom[i].push_back(j); |
278 |
+ |
topoDist[i].push_back(2); |
279 |
+ |
} else { |
280 |
+ |
if (oneFour->hasPair(iglob, jglob)) { |
281 |
+ |
toposForAtom[i].push_back(j); |
282 |
+ |
topoDist[i].push_back(3); |
283 |
+ |
} |
284 |
+ |
} |
285 |
+ |
} |
286 |
+ |
} |
287 |
+ |
} |
288 |
+ |
|
289 |
+ |
createGtypeCutoffMap(); |
290 |
+ |
|
291 |
|
} |
292 |
+ |
|
293 |
+ |
void ForceMatrixDecomposition::createGtypeCutoffMap() { |
294 |
|
|
295 |
+ |
RealType tol = 1e-6; |
296 |
+ |
largestRcut_ = 0.0; |
297 |
+ |
RealType rc; |
298 |
+ |
int atid; |
299 |
+ |
set<AtomType*> atypes = info_->getSimulatedAtomTypes(); |
300 |
+ |
|
301 |
+ |
map<int, RealType> atypeCutoff; |
302 |
+ |
|
303 |
+ |
for (set<AtomType*>::iterator at = atypes.begin(); |
304 |
+ |
at != atypes.end(); ++at){ |
305 |
+ |
atid = (*at)->getIdent(); |
306 |
+ |
if (userChoseCutoff_) |
307 |
+ |
atypeCutoff[atid] = userCutoff_; |
308 |
+ |
else |
309 |
+ |
atypeCutoff[atid] = interactionMan_->getSuggestedCutoffRadius(*at); |
310 |
+ |
} |
311 |
+ |
|
312 |
+ |
vector<RealType> gTypeCutoffs; |
313 |
+ |
// first we do a single loop over the cutoff groups to find the |
314 |
+ |
// largest cutoff for any atypes present in this group. |
315 |
+ |
#ifdef IS_MPI |
316 |
+ |
vector<RealType> groupCutoffRow(nGroupsInRow_, 0.0); |
317 |
+ |
groupRowToGtype.resize(nGroupsInRow_); |
318 |
+ |
for (int cg1 = 0; cg1 < nGroupsInRow_; cg1++) { |
319 |
+ |
vector<int> atomListRow = getAtomsInGroupRow(cg1); |
320 |
+ |
for (vector<int>::iterator ia = atomListRow.begin(); |
321 |
+ |
ia != atomListRow.end(); ++ia) { |
322 |
+ |
int atom1 = (*ia); |
323 |
+ |
atid = identsRow[atom1]; |
324 |
+ |
if (atypeCutoff[atid] > groupCutoffRow[cg1]) { |
325 |
+ |
groupCutoffRow[cg1] = atypeCutoff[atid]; |
326 |
+ |
} |
327 |
+ |
} |
328 |
|
|
329 |
< |
|
330 |
< |
void ForceDecomposition::distributeData() { |
329 |
> |
bool gTypeFound = false; |
330 |
> |
for (int gt = 0; gt < gTypeCutoffs.size(); gt++) { |
331 |
> |
if (abs(groupCutoffRow[cg1] - gTypeCutoffs[gt]) < tol) { |
332 |
> |
groupRowToGtype[cg1] = gt; |
333 |
> |
gTypeFound = true; |
334 |
> |
} |
335 |
> |
} |
336 |
> |
if (!gTypeFound) { |
337 |
> |
gTypeCutoffs.push_back( groupCutoffRow[cg1] ); |
338 |
> |
groupRowToGtype[cg1] = gTypeCutoffs.size() - 1; |
339 |
> |
} |
340 |
> |
|
341 |
> |
} |
342 |
> |
vector<RealType> groupCutoffCol(nGroupsInCol_, 0.0); |
343 |
> |
groupColToGtype.resize(nGroupsInCol_); |
344 |
> |
for (int cg2 = 0; cg2 < nGroupsInCol_; cg2++) { |
345 |
> |
vector<int> atomListCol = getAtomsInGroupColumn(cg2); |
346 |
> |
for (vector<int>::iterator jb = atomListCol.begin(); |
347 |
> |
jb != atomListCol.end(); ++jb) { |
348 |
> |
int atom2 = (*jb); |
349 |
> |
atid = identsCol[atom2]; |
350 |
> |
if (atypeCutoff[atid] > groupCutoffCol[cg2]) { |
351 |
> |
groupCutoffCol[cg2] = atypeCutoff[atid]; |
352 |
> |
} |
353 |
> |
} |
354 |
> |
bool gTypeFound = false; |
355 |
> |
for (int gt = 0; gt < gTypeCutoffs.size(); gt++) { |
356 |
> |
if (abs(groupCutoffCol[cg2] - gTypeCutoffs[gt]) < tol) { |
357 |
> |
groupColToGtype[cg2] = gt; |
358 |
> |
gTypeFound = true; |
359 |
> |
} |
360 |
> |
} |
361 |
> |
if (!gTypeFound) { |
362 |
> |
gTypeCutoffs.push_back( groupCutoffCol[cg2] ); |
363 |
> |
groupColToGtype[cg2] = gTypeCutoffs.size() - 1; |
364 |
> |
} |
365 |
> |
} |
366 |
> |
#else |
367 |
> |
|
368 |
> |
vector<RealType> groupCutoff(nGroups_, 0.0); |
369 |
> |
groupToGtype.resize(nGroups_); |
370 |
> |
for (int cg1 = 0; cg1 < nGroups_; cg1++) { |
371 |
> |
groupCutoff[cg1] = 0.0; |
372 |
> |
vector<int> atomList = getAtomsInGroupRow(cg1); |
373 |
> |
for (vector<int>::iterator ia = atomList.begin(); |
374 |
> |
ia != atomList.end(); ++ia) { |
375 |
> |
int atom1 = (*ia); |
376 |
> |
atid = idents[atom1]; |
377 |
> |
if (atypeCutoff[atid] > groupCutoff[cg1]) |
378 |
> |
groupCutoff[cg1] = atypeCutoff[atid]; |
379 |
> |
} |
380 |
> |
|
381 |
> |
bool gTypeFound = false; |
382 |
> |
for (int gt = 0; gt < gTypeCutoffs.size(); gt++) { |
383 |
> |
if (abs(groupCutoff[cg1] - gTypeCutoffs[gt]) < tol) { |
384 |
> |
groupToGtype[cg1] = gt; |
385 |
> |
gTypeFound = true; |
386 |
> |
} |
387 |
> |
} |
388 |
> |
if (!gTypeFound) { |
389 |
> |
gTypeCutoffs.push_back( groupCutoff[cg1] ); |
390 |
> |
groupToGtype[cg1] = gTypeCutoffs.size() - 1; |
391 |
> |
} |
392 |
> |
} |
393 |
> |
#endif |
394 |
> |
|
395 |
> |
// Now we find the maximum group cutoff value present in the simulation |
396 |
> |
|
397 |
> |
RealType groupMax = *max_element(gTypeCutoffs.begin(), |
398 |
> |
gTypeCutoffs.end()); |
399 |
> |
|
400 |
|
#ifdef IS_MPI |
401 |
< |
Snapshot* snap = sman_->getCurrentSnapshot(); |
401 |
> |
MPI::COMM_WORLD.Allreduce(&groupMax, &groupMax, 1, MPI::REALTYPE, |
402 |
> |
MPI::MAX); |
403 |
> |
#endif |
404 |
> |
|
405 |
> |
RealType tradRcut = groupMax; |
406 |
|
|
407 |
+ |
for (int i = 0; i < gTypeCutoffs.size(); i++) { |
408 |
+ |
for (int j = 0; j < gTypeCutoffs.size(); j++) { |
409 |
+ |
RealType thisRcut; |
410 |
+ |
switch(cutoffPolicy_) { |
411 |
+ |
case TRADITIONAL: |
412 |
+ |
thisRcut = tradRcut; |
413 |
+ |
break; |
414 |
+ |
case MIX: |
415 |
+ |
thisRcut = 0.5 * (gTypeCutoffs[i] + gTypeCutoffs[j]); |
416 |
+ |
break; |
417 |
+ |
case MAX: |
418 |
+ |
thisRcut = max(gTypeCutoffs[i], gTypeCutoffs[j]); |
419 |
+ |
break; |
420 |
+ |
default: |
421 |
+ |
sprintf(painCave.errMsg, |
422 |
+ |
"ForceMatrixDecomposition::createGtypeCutoffMap " |
423 |
+ |
"hit an unknown cutoff policy!\n"); |
424 |
+ |
painCave.severity = OPENMD_ERROR; |
425 |
+ |
painCave.isFatal = 1; |
426 |
+ |
simError(); |
427 |
+ |
break; |
428 |
+ |
} |
429 |
+ |
|
430 |
+ |
pair<int,int> key = make_pair(i,j); |
431 |
+ |
gTypeCutoffMap[key].first = thisRcut; |
432 |
+ |
if (thisRcut > largestRcut_) largestRcut_ = thisRcut; |
433 |
+ |
gTypeCutoffMap[key].second = thisRcut*thisRcut; |
434 |
+ |
gTypeCutoffMap[key].third = pow(thisRcut + skinThickness_, 2); |
435 |
+ |
// sanity check |
436 |
+ |
|
437 |
+ |
if (userChoseCutoff_) { |
438 |
+ |
if (abs(gTypeCutoffMap[key].first - userCutoff_) > 0.0001) { |
439 |
+ |
sprintf(painCave.errMsg, |
440 |
+ |
"ForceMatrixDecomposition::createGtypeCutoffMap " |
441 |
+ |
"user-specified rCut (%lf) does not match computed group Cutoff\n", userCutoff_); |
442 |
+ |
painCave.severity = OPENMD_ERROR; |
443 |
+ |
painCave.isFatal = 1; |
444 |
+ |
simError(); |
445 |
+ |
} |
446 |
+ |
} |
447 |
+ |
} |
448 |
+ |
} |
449 |
+ |
} |
450 |
+ |
|
451 |
+ |
|
452 |
+ |
groupCutoffs ForceMatrixDecomposition::getGroupCutoffs(int cg1, int cg2) { |
453 |
+ |
int i, j; |
454 |
+ |
#ifdef IS_MPI |
455 |
+ |
i = groupRowToGtype[cg1]; |
456 |
+ |
j = groupColToGtype[cg2]; |
457 |
+ |
#else |
458 |
+ |
i = groupToGtype[cg1]; |
459 |
+ |
j = groupToGtype[cg2]; |
460 |
+ |
#endif |
461 |
+ |
return gTypeCutoffMap[make_pair(i,j)]; |
462 |
+ |
} |
463 |
+ |
|
464 |
+ |
int ForceMatrixDecomposition::getTopologicalDistance(int atom1, int atom2) { |
465 |
+ |
for (int j = 0; j < toposForAtom[atom1].size(); j++) { |
466 |
+ |
if (toposForAtom[atom1][j] == atom2) |
467 |
+ |
return topoDist[atom1][j]; |
468 |
+ |
} |
469 |
+ |
return 0; |
470 |
+ |
} |
471 |
+ |
|
472 |
+ |
void ForceMatrixDecomposition::zeroWorkArrays() { |
473 |
+ |
pairwisePot = 0.0; |
474 |
+ |
embeddingPot = 0.0; |
475 |
+ |
|
476 |
+ |
#ifdef IS_MPI |
477 |
+ |
if (storageLayout_ & DataStorage::dslForce) { |
478 |
+ |
fill(atomRowData.force.begin(), atomRowData.force.end(), V3Zero); |
479 |
+ |
fill(atomColData.force.begin(), atomColData.force.end(), V3Zero); |
480 |
+ |
} |
481 |
+ |
|
482 |
+ |
if (storageLayout_ & DataStorage::dslTorque) { |
483 |
+ |
fill(atomRowData.torque.begin(), atomRowData.torque.end(), V3Zero); |
484 |
+ |
fill(atomColData.torque.begin(), atomColData.torque.end(), V3Zero); |
485 |
+ |
} |
486 |
+ |
|
487 |
+ |
fill(pot_row.begin(), pot_row.end(), |
488 |
+ |
Vector<RealType, N_INTERACTION_FAMILIES> (0.0)); |
489 |
+ |
|
490 |
+ |
fill(pot_col.begin(), pot_col.end(), |
491 |
+ |
Vector<RealType, N_INTERACTION_FAMILIES> (0.0)); |
492 |
+ |
|
493 |
+ |
if (storageLayout_ & DataStorage::dslParticlePot) { |
494 |
+ |
fill(atomRowData.particlePot.begin(), atomRowData.particlePot.end(), |
495 |
+ |
0.0); |
496 |
+ |
fill(atomColData.particlePot.begin(), atomColData.particlePot.end(), |
497 |
+ |
0.0); |
498 |
+ |
} |
499 |
+ |
|
500 |
+ |
if (storageLayout_ & DataStorage::dslDensity) { |
501 |
+ |
fill(atomRowData.density.begin(), atomRowData.density.end(), 0.0); |
502 |
+ |
fill(atomColData.density.begin(), atomColData.density.end(), 0.0); |
503 |
+ |
} |
504 |
+ |
|
505 |
+ |
if (storageLayout_ & DataStorage::dslFunctional) { |
506 |
+ |
fill(atomRowData.functional.begin(), atomRowData.functional.end(), |
507 |
+ |
0.0); |
508 |
+ |
fill(atomColData.functional.begin(), atomColData.functional.end(), |
509 |
+ |
0.0); |
510 |
+ |
} |
511 |
+ |
|
512 |
+ |
if (storageLayout_ & DataStorage::dslFunctionalDerivative) { |
513 |
+ |
fill(atomRowData.functionalDerivative.begin(), |
514 |
+ |
atomRowData.functionalDerivative.end(), 0.0); |
515 |
+ |
fill(atomColData.functionalDerivative.begin(), |
516 |
+ |
atomColData.functionalDerivative.end(), 0.0); |
517 |
+ |
} |
518 |
+ |
|
519 |
+ |
if (storageLayout_ & DataStorage::dslSkippedCharge) { |
520 |
+ |
fill(atomRowData.skippedCharge.begin(), |
521 |
+ |
atomRowData.skippedCharge.end(), 0.0); |
522 |
+ |
fill(atomColData.skippedCharge.begin(), |
523 |
+ |
atomColData.skippedCharge.end(), 0.0); |
524 |
+ |
} |
525 |
+ |
|
526 |
+ |
#endif |
527 |
+ |
// even in parallel, we need to zero out the local arrays: |
528 |
+ |
|
529 |
+ |
if (storageLayout_ & DataStorage::dslParticlePot) { |
530 |
+ |
fill(snap_->atomData.particlePot.begin(), |
531 |
+ |
snap_->atomData.particlePot.end(), 0.0); |
532 |
+ |
} |
533 |
+ |
|
534 |
+ |
if (storageLayout_ & DataStorage::dslDensity) { |
535 |
+ |
fill(snap_->atomData.density.begin(), |
536 |
+ |
snap_->atomData.density.end(), 0.0); |
537 |
+ |
} |
538 |
+ |
if (storageLayout_ & DataStorage::dslFunctional) { |
539 |
+ |
fill(snap_->atomData.functional.begin(), |
540 |
+ |
snap_->atomData.functional.end(), 0.0); |
541 |
+ |
} |
542 |
+ |
if (storageLayout_ & DataStorage::dslFunctionalDerivative) { |
543 |
+ |
fill(snap_->atomData.functionalDerivative.begin(), |
544 |
+ |
snap_->atomData.functionalDerivative.end(), 0.0); |
545 |
+ |
} |
546 |
+ |
if (storageLayout_ & DataStorage::dslSkippedCharge) { |
547 |
+ |
fill(snap_->atomData.skippedCharge.begin(), |
548 |
+ |
snap_->atomData.skippedCharge.end(), 0.0); |
549 |
+ |
} |
550 |
+ |
|
551 |
+ |
} |
552 |
+ |
|
553 |
+ |
|
554 |
+ |
void ForceMatrixDecomposition::distributeData() { |
555 |
+ |
snap_ = sman_->getCurrentSnapshot(); |
556 |
+ |
storageLayout_ = sman_->getStorageLayout(); |
557 |
+ |
#ifdef IS_MPI |
558 |
+ |
|
559 |
|
// gather up the atomic positions |
560 |
< |
AtomCommVectorI->gather(snap->atomData.position, |
561 |
< |
snap->atomIData.position); |
562 |
< |
AtomCommVectorJ->gather(snap->atomData.position, |
563 |
< |
snap->atomJData.position); |
560 |
> |
AtomPlanVectorRow->gather(snap_->atomData.position, |
561 |
> |
atomRowData.position); |
562 |
> |
AtomPlanVectorColumn->gather(snap_->atomData.position, |
563 |
> |
atomColData.position); |
564 |
|
|
565 |
|
// gather up the cutoff group positions |
566 |
< |
cgCommVectorI->gather(snap->cgData.position, |
567 |
< |
snap->cgIData.position); |
568 |
< |
cgCommVectorJ->gather(snap->cgData.position, |
569 |
< |
snap->cgJData.position); |
566 |
> |
|
567 |
> |
cgPlanVectorRow->gather(snap_->cgData.position, |
568 |
> |
cgRowData.position); |
569 |
> |
|
570 |
> |
cgPlanVectorColumn->gather(snap_->cgData.position, |
571 |
> |
cgColData.position); |
572 |
> |
|
573 |
|
|
574 |
|
// if needed, gather the atomic rotation matrices |
575 |
< |
if (snap->atomData.getStorageLayout() & DataStorage::dslAmat) { |
576 |
< |
AtomCommMatrixI->gather(snap->atomData.aMat, |
577 |
< |
snap->atomIData.aMat); |
578 |
< |
AtomCommMatrixJ->gather(snap->atomData.aMat, |
579 |
< |
snap->atomJData.aMat); |
575 |
> |
if (storageLayout_ & DataStorage::dslAmat) { |
576 |
> |
AtomPlanMatrixRow->gather(snap_->atomData.aMat, |
577 |
> |
atomRowData.aMat); |
578 |
> |
AtomPlanMatrixColumn->gather(snap_->atomData.aMat, |
579 |
> |
atomColData.aMat); |
580 |
|
} |
581 |
|
|
582 |
|
// if needed, gather the atomic eletrostatic frames |
583 |
< |
if (snap->atomData.getStorageLayout() & DataStorage::dslElectroFrame) { |
584 |
< |
AtomCommMatrixI->gather(snap->atomData.electroFrame, |
585 |
< |
snap->atomIData.electroFrame); |
586 |
< |
AtomCommMatrixJ->gather(snap->atomData.electroFrame, |
587 |
< |
snap->atomJData.electroFrame); |
583 |
> |
if (storageLayout_ & DataStorage::dslElectroFrame) { |
584 |
> |
AtomPlanMatrixRow->gather(snap_->atomData.electroFrame, |
585 |
> |
atomRowData.electroFrame); |
586 |
> |
AtomPlanMatrixColumn->gather(snap_->atomData.electroFrame, |
587 |
> |
atomColData.electroFrame); |
588 |
|
} |
589 |
+ |
|
590 |
|
#endif |
591 |
|
} |
592 |
|
|
593 |
< |
void ForceDecomposition::collectIntermediateData() { |
593 |
> |
/* collects information obtained during the pre-pair loop onto local |
594 |
> |
* data structures. |
595 |
> |
*/ |
596 |
> |
void ForceMatrixDecomposition::collectIntermediateData() { |
597 |
> |
snap_ = sman_->getCurrentSnapshot(); |
598 |
> |
storageLayout_ = sman_->getStorageLayout(); |
599 |
|
#ifdef IS_MPI |
105 |
– |
Snapshot* snap = sman_->getCurrentSnapshot(); |
106 |
– |
// gather up the atomic positions |
600 |
|
|
601 |
< |
if (snap->atomData.getStorageLayout() & DataStorage::dslDensity) { |
602 |
< |
AtomCommRealI->scatter(snap->atomIData.density, |
603 |
< |
snap->atomData.density); |
604 |
< |
std::vector<RealType> rho_tmp; |
605 |
< |
int n = snap->getNumberOfAtoms(); |
606 |
< |
rho_tmp.reserve( n ); |
607 |
< |
AtomCommRealJ->scatter(snap->atomJData.density, rho_tmp); |
601 |
> |
if (storageLayout_ & DataStorage::dslDensity) { |
602 |
> |
|
603 |
> |
AtomPlanRealRow->scatter(atomRowData.density, |
604 |
> |
snap_->atomData.density); |
605 |
> |
|
606 |
> |
int n = snap_->atomData.density.size(); |
607 |
> |
vector<RealType> rho_tmp(n, 0.0); |
608 |
> |
AtomPlanRealColumn->scatter(atomColData.density, rho_tmp); |
609 |
|
for (int i = 0; i < n; i++) |
610 |
< |
snap->atomData.density[i] += rho_tmp[i]; |
610 |
> |
snap_->atomData.density[i] += rho_tmp[i]; |
611 |
|
} |
612 |
|
#endif |
613 |
|
} |
614 |
< |
|
615 |
< |
void ForceDecomposition::distributeIntermediateData() { |
614 |
> |
|
615 |
> |
/* |
616 |
> |
* redistributes information obtained during the pre-pair loop out to |
617 |
> |
* row and column-indexed data structures |
618 |
> |
*/ |
619 |
> |
void ForceMatrixDecomposition::distributeIntermediateData() { |
620 |
> |
snap_ = sman_->getCurrentSnapshot(); |
621 |
> |
storageLayout_ = sman_->getStorageLayout(); |
622 |
|
#ifdef IS_MPI |
623 |
< |
Snapshot* snap = sman_->getCurrentSnapshot(); |
624 |
< |
if (snap->atomData.getStorageLayout() & DataStorage::dslFunctional) { |
625 |
< |
AtomCommRealI->gather(snap->atomData.functional, |
626 |
< |
snap->atomIData.functional); |
627 |
< |
AtomCommRealJ->gather(snap->atomData.functional, |
128 |
< |
snap->atomJData.functional); |
623 |
> |
if (storageLayout_ & DataStorage::dslFunctional) { |
624 |
> |
AtomPlanRealRow->gather(snap_->atomData.functional, |
625 |
> |
atomRowData.functional); |
626 |
> |
AtomPlanRealColumn->gather(snap_->atomData.functional, |
627 |
> |
atomColData.functional); |
628 |
|
} |
629 |
|
|
630 |
< |
if (snap->atomData.getStorageLayout() & DataStorage::dslFunctionalDerivative) { |
631 |
< |
AtomCommRealI->gather(snap->atomData.functionalDerivative, |
632 |
< |
snap->atomIData.functionalDerivative); |
633 |
< |
AtomCommRealJ->gather(snap->atomData.functionalDerivative, |
634 |
< |
snap->atomJData.functionalDerivative); |
630 |
> |
if (storageLayout_ & DataStorage::dslFunctionalDerivative) { |
631 |
> |
AtomPlanRealRow->gather(snap_->atomData.functionalDerivative, |
632 |
> |
atomRowData.functionalDerivative); |
633 |
> |
AtomPlanRealColumn->gather(snap_->atomData.functionalDerivative, |
634 |
> |
atomColData.functionalDerivative); |
635 |
|
} |
636 |
|
#endif |
637 |
|
} |
638 |
|
|
639 |
|
|
640 |
< |
void ForceDecomposition::collectData() { |
640 |
> |
void ForceMatrixDecomposition::collectData() { |
641 |
> |
snap_ = sman_->getCurrentSnapshot(); |
642 |
> |
storageLayout_ = sman_->getStorageLayout(); |
643 |
> |
#ifdef IS_MPI |
644 |
> |
int n = snap_->atomData.force.size(); |
645 |
> |
vector<Vector3d> frc_tmp(n, V3Zero); |
646 |
> |
|
647 |
> |
AtomPlanVectorRow->scatter(atomRowData.force, frc_tmp); |
648 |
> |
for (int i = 0; i < n; i++) { |
649 |
> |
snap_->atomData.force[i] += frc_tmp[i]; |
650 |
> |
frc_tmp[i] = 0.0; |
651 |
> |
} |
652 |
> |
|
653 |
> |
AtomPlanVectorColumn->scatter(atomColData.force, frc_tmp); |
654 |
> |
for (int i = 0; i < n; i++) { |
655 |
> |
snap_->atomData.force[i] += frc_tmp[i]; |
656 |
> |
} |
657 |
> |
|
658 |
> |
if (storageLayout_ & DataStorage::dslTorque) { |
659 |
> |
|
660 |
> |
int nt = snap_->atomData.torque.size(); |
661 |
> |
vector<Vector3d> trq_tmp(nt, V3Zero); |
662 |
> |
|
663 |
> |
AtomPlanVectorRow->scatter(atomRowData.torque, trq_tmp); |
664 |
> |
for (int i = 0; i < nt; i++) { |
665 |
> |
snap_->atomData.torque[i] += trq_tmp[i]; |
666 |
> |
trq_tmp[i] = 0.0; |
667 |
> |
} |
668 |
> |
|
669 |
> |
AtomPlanVectorColumn->scatter(atomColData.torque, trq_tmp); |
670 |
> |
for (int i = 0; i < nt; i++) |
671 |
> |
snap_->atomData.torque[i] += trq_tmp[i]; |
672 |
> |
} |
673 |
> |
|
674 |
> |
if (storageLayout_ & DataStorage::dslSkippedCharge) { |
675 |
> |
|
676 |
> |
int ns = snap_->atomData.skippedCharge.size(); |
677 |
> |
vector<RealType> skch_tmp(ns, 0.0); |
678 |
> |
|
679 |
> |
AtomPlanRealRow->scatter(atomRowData.skippedCharge, skch_tmp); |
680 |
> |
for (int i = 0; i < ns; i++) { |
681 |
> |
snap_->atomData.skippedCharge[i] += skch_tmp[i]; |
682 |
> |
skch_tmp[i] = 0.0; |
683 |
> |
} |
684 |
> |
|
685 |
> |
AtomPlanRealColumn->scatter(atomColData.skippedCharge, skch_tmp); |
686 |
> |
for (int i = 0; i < ns; i++) |
687 |
> |
snap_->atomData.skippedCharge[i] += skch_tmp[i]; |
688 |
> |
} |
689 |
> |
|
690 |
> |
nLocal_ = snap_->getNumberOfAtoms(); |
691 |
> |
|
692 |
> |
vector<potVec> pot_temp(nLocal_, |
693 |
> |
Vector<RealType, N_INTERACTION_FAMILIES> (0.0)); |
694 |
> |
|
695 |
> |
// scatter/gather pot_row into the members of my column |
696 |
> |
|
697 |
> |
AtomPlanPotRow->scatter(pot_row, pot_temp); |
698 |
> |
|
699 |
> |
for (int ii = 0; ii < pot_temp.size(); ii++ ) |
700 |
> |
pairwisePot += pot_temp[ii]; |
701 |
> |
|
702 |
> |
fill(pot_temp.begin(), pot_temp.end(), |
703 |
> |
Vector<RealType, N_INTERACTION_FAMILIES> (0.0)); |
704 |
> |
|
705 |
> |
AtomPlanPotColumn->scatter(pot_col, pot_temp); |
706 |
> |
|
707 |
> |
for (int ii = 0; ii < pot_temp.size(); ii++ ) |
708 |
> |
pairwisePot += pot_temp[ii]; |
709 |
> |
|
710 |
> |
for (int ii = 0; ii < N_INTERACTION_FAMILIES; ii++) { |
711 |
> |
RealType ploc1 = pairwisePot[ii]; |
712 |
> |
RealType ploc2 = 0.0; |
713 |
> |
MPI::COMM_WORLD.Allreduce(&ploc1, &ploc2, 1, MPI::REALTYPE, MPI::SUM); |
714 |
> |
pairwisePot[ii] = ploc2; |
715 |
> |
} |
716 |
> |
|
717 |
> |
#endif |
718 |
> |
|
719 |
> |
} |
720 |
> |
|
721 |
> |
int ForceMatrixDecomposition::getNAtomsInRow() { |
722 |
|
#ifdef IS_MPI |
723 |
+ |
return nAtomsInRow_; |
724 |
+ |
#else |
725 |
+ |
return nLocal_; |
726 |
+ |
#endif |
727 |
+ |
} |
728 |
+ |
|
729 |
+ |
/** |
730 |
+ |
* returns the list of atoms belonging to this group. |
731 |
+ |
*/ |
732 |
+ |
vector<int> ForceMatrixDecomposition::getAtomsInGroupRow(int cg1){ |
733 |
+ |
#ifdef IS_MPI |
734 |
+ |
return groupListRow_[cg1]; |
735 |
+ |
#else |
736 |
+ |
return groupList_[cg1]; |
737 |
+ |
#endif |
738 |
+ |
} |
739 |
+ |
|
740 |
+ |
vector<int> ForceMatrixDecomposition::getAtomsInGroupColumn(int cg2){ |
741 |
+ |
#ifdef IS_MPI |
742 |
+ |
return groupListCol_[cg2]; |
743 |
+ |
#else |
744 |
+ |
return groupList_[cg2]; |
745 |
+ |
#endif |
746 |
+ |
} |
747 |
+ |
|
748 |
+ |
Vector3d ForceMatrixDecomposition::getIntergroupVector(int cg1, int cg2){ |
749 |
+ |
Vector3d d; |
750 |
+ |
|
751 |
+ |
#ifdef IS_MPI |
752 |
+ |
d = cgColData.position[cg2] - cgRowData.position[cg1]; |
753 |
+ |
#else |
754 |
+ |
d = snap_->cgData.position[cg2] - snap_->cgData.position[cg1]; |
755 |
+ |
#endif |
756 |
+ |
|
757 |
+ |
snap_->wrapVector(d); |
758 |
+ |
return d; |
759 |
+ |
} |
760 |
+ |
|
761 |
+ |
|
762 |
+ |
Vector3d ForceMatrixDecomposition::getAtomToGroupVectorRow(int atom1, int cg1){ |
763 |
+ |
|
764 |
+ |
Vector3d d; |
765 |
+ |
|
766 |
+ |
#ifdef IS_MPI |
767 |
+ |
d = cgRowData.position[cg1] - atomRowData.position[atom1]; |
768 |
+ |
#else |
769 |
+ |
d = snap_->cgData.position[cg1] - snap_->atomData.position[atom1]; |
770 |
+ |
#endif |
771 |
+ |
|
772 |
+ |
snap_->wrapVector(d); |
773 |
+ |
return d; |
774 |
+ |
} |
775 |
+ |
|
776 |
+ |
Vector3d ForceMatrixDecomposition::getAtomToGroupVectorColumn(int atom2, int cg2){ |
777 |
+ |
Vector3d d; |
778 |
+ |
|
779 |
+ |
#ifdef IS_MPI |
780 |
+ |
d = cgColData.position[cg2] - atomColData.position[atom2]; |
781 |
+ |
#else |
782 |
+ |
d = snap_->cgData.position[cg2] - snap_->atomData.position[atom2]; |
783 |
+ |
#endif |
784 |
+ |
|
785 |
+ |
snap_->wrapVector(d); |
786 |
+ |
return d; |
787 |
+ |
} |
788 |
+ |
|
789 |
+ |
RealType ForceMatrixDecomposition::getMassFactorRow(int atom1) { |
790 |
+ |
#ifdef IS_MPI |
791 |
+ |
return massFactorsRow[atom1]; |
792 |
+ |
#else |
793 |
+ |
return massFactors[atom1]; |
794 |
+ |
#endif |
795 |
+ |
} |
796 |
+ |
|
797 |
+ |
RealType ForceMatrixDecomposition::getMassFactorColumn(int atom2) { |
798 |
+ |
#ifdef IS_MPI |
799 |
+ |
return massFactorsCol[atom2]; |
800 |
+ |
#else |
801 |
+ |
return massFactors[atom2]; |
802 |
+ |
#endif |
803 |
+ |
|
804 |
+ |
} |
805 |
+ |
|
806 |
+ |
Vector3d ForceMatrixDecomposition::getInteratomicVector(int atom1, int atom2){ |
807 |
+ |
Vector3d d; |
808 |
+ |
|
809 |
+ |
#ifdef IS_MPI |
810 |
+ |
d = atomColData.position[atom2] - atomRowData.position[atom1]; |
811 |
+ |
#else |
812 |
+ |
d = snap_->atomData.position[atom2] - snap_->atomData.position[atom1]; |
813 |
+ |
#endif |
814 |
+ |
|
815 |
+ |
snap_->wrapVector(d); |
816 |
+ |
return d; |
817 |
+ |
} |
818 |
+ |
|
819 |
+ |
vector<int> ForceMatrixDecomposition::getExcludesForAtom(int atom1) { |
820 |
+ |
return excludesForAtom[atom1]; |
821 |
+ |
} |
822 |
+ |
|
823 |
+ |
/** |
824 |
+ |
* We need to exclude some overcounted interactions that result from |
825 |
+ |
* the parallel decomposition. |
826 |
+ |
*/ |
827 |
+ |
bool ForceMatrixDecomposition::skipAtomPair(int atom1, int atom2) { |
828 |
+ |
int unique_id_1, unique_id_2; |
829 |
+ |
|
830 |
+ |
#ifdef IS_MPI |
831 |
+ |
// in MPI, we have to look up the unique IDs for each atom |
832 |
+ |
unique_id_1 = AtomRowToGlobal[atom1]; |
833 |
+ |
unique_id_2 = AtomColToGlobal[atom2]; |
834 |
+ |
|
835 |
+ |
// this situation should only arise in MPI simulations |
836 |
+ |
if (unique_id_1 == unique_id_2) return true; |
837 |
+ |
|
838 |
+ |
// this prevents us from doing the pair on multiple processors |
839 |
+ |
if (unique_id_1 < unique_id_2) { |
840 |
+ |
if ((unique_id_1 + unique_id_2) % 2 == 0) return true; |
841 |
+ |
} else { |
842 |
+ |
if ((unique_id_1 + unique_id_2) % 2 == 1) return true; |
843 |
+ |
} |
844 |
+ |
#endif |
845 |
+ |
return false; |
846 |
+ |
} |
847 |
+ |
|
848 |
+ |
/** |
849 |
+ |
* We need to handle the interactions for atoms who are involved in |
850 |
+ |
* the same rigid body as well as some short range interactions |
851 |
+ |
* (bonds, bends, torsions) differently from other interactions. |
852 |
+ |
* We'll still visit the pairwise routines, but with a flag that |
853 |
+ |
* tells those routines to exclude the pair from direct long range |
854 |
+ |
* interactions. Some indirect interactions (notably reaction |
855 |
+ |
* field) must still be handled for these pairs. |
856 |
+ |
*/ |
857 |
+ |
bool ForceMatrixDecomposition::excludeAtomPair(int atom1, int atom2) { |
858 |
+ |
int unique_id_2; |
859 |
+ |
#ifdef IS_MPI |
860 |
+ |
// in MPI, we have to look up the unique IDs for the row atom. |
861 |
+ |
unique_id_2 = AtomColToGlobal[atom2]; |
862 |
+ |
#else |
863 |
+ |
// in the normal loop, the atom numbers are unique |
864 |
+ |
unique_id_2 = atom2; |
865 |
+ |
#endif |
866 |
+ |
|
867 |
+ |
for (vector<int>::iterator i = excludesForAtom[atom1].begin(); |
868 |
+ |
i != excludesForAtom[atom1].end(); ++i) { |
869 |
+ |
if ( (*i) == unique_id_2 ) return true; |
870 |
+ |
} |
871 |
+ |
|
872 |
+ |
return false; |
873 |
+ |
} |
874 |
+ |
|
875 |
+ |
|
876 |
+ |
void ForceMatrixDecomposition::addForceToAtomRow(int atom1, Vector3d fg){ |
877 |
+ |
#ifdef IS_MPI |
878 |
+ |
atomRowData.force[atom1] += fg; |
879 |
+ |
#else |
880 |
+ |
snap_->atomData.force[atom1] += fg; |
881 |
+ |
#endif |
882 |
+ |
} |
883 |
+ |
|
884 |
+ |
void ForceMatrixDecomposition::addForceToAtomColumn(int atom2, Vector3d fg){ |
885 |
+ |
#ifdef IS_MPI |
886 |
+ |
atomColData.force[atom2] += fg; |
887 |
+ |
#else |
888 |
+ |
snap_->atomData.force[atom2] += fg; |
889 |
+ |
#endif |
890 |
+ |
} |
891 |
+ |
|
892 |
+ |
// filling interaction blocks with pointers |
893 |
+ |
void ForceMatrixDecomposition::fillInteractionData(InteractionData &idat, |
894 |
+ |
int atom1, int atom2) { |
895 |
+ |
|
896 |
+ |
idat.excluded = excludeAtomPair(atom1, atom2); |
897 |
+ |
|
898 |
+ |
#ifdef IS_MPI |
899 |
+ |
idat.atypes = make_pair( atypesRow[atom1], atypesCol[atom2]); |
900 |
+ |
//idat.atypes = make_pair( ff_->getAtomType(identsRow[atom1]), |
901 |
+ |
// ff_->getAtomType(identsCol[atom2]) ); |
902 |
+ |
|
903 |
+ |
if (storageLayout_ & DataStorage::dslAmat) { |
904 |
+ |
idat.A1 = &(atomRowData.aMat[atom1]); |
905 |
+ |
idat.A2 = &(atomColData.aMat[atom2]); |
906 |
+ |
} |
907 |
+ |
|
908 |
+ |
if (storageLayout_ & DataStorage::dslElectroFrame) { |
909 |
+ |
idat.eFrame1 = &(atomRowData.electroFrame[atom1]); |
910 |
+ |
idat.eFrame2 = &(atomColData.electroFrame[atom2]); |
911 |
+ |
} |
912 |
+ |
|
913 |
+ |
if (storageLayout_ & DataStorage::dslTorque) { |
914 |
+ |
idat.t1 = &(atomRowData.torque[atom1]); |
915 |
+ |
idat.t2 = &(atomColData.torque[atom2]); |
916 |
+ |
} |
917 |
+ |
|
918 |
+ |
if (storageLayout_ & DataStorage::dslDensity) { |
919 |
+ |
idat.rho1 = &(atomRowData.density[atom1]); |
920 |
+ |
idat.rho2 = &(atomColData.density[atom2]); |
921 |
+ |
} |
922 |
+ |
|
923 |
+ |
if (storageLayout_ & DataStorage::dslFunctional) { |
924 |
+ |
idat.frho1 = &(atomRowData.functional[atom1]); |
925 |
+ |
idat.frho2 = &(atomColData.functional[atom2]); |
926 |
+ |
} |
927 |
+ |
|
928 |
+ |
if (storageLayout_ & DataStorage::dslFunctionalDerivative) { |
929 |
+ |
idat.dfrho1 = &(atomRowData.functionalDerivative[atom1]); |
930 |
+ |
idat.dfrho2 = &(atomColData.functionalDerivative[atom2]); |
931 |
+ |
} |
932 |
+ |
|
933 |
+ |
if (storageLayout_ & DataStorage::dslParticlePot) { |
934 |
+ |
idat.particlePot1 = &(atomRowData.particlePot[atom1]); |
935 |
+ |
idat.particlePot2 = &(atomColData.particlePot[atom2]); |
936 |
+ |
} |
937 |
+ |
|
938 |
+ |
if (storageLayout_ & DataStorage::dslSkippedCharge) { |
939 |
+ |
idat.skippedCharge1 = &(atomRowData.skippedCharge[atom1]); |
940 |
+ |
idat.skippedCharge2 = &(atomColData.skippedCharge[atom2]); |
941 |
+ |
} |
942 |
+ |
|
943 |
+ |
#else |
944 |
+ |
|
945 |
+ |
idat.atypes = make_pair( atypesLocal[atom1], atypesLocal[atom2]); |
946 |
+ |
//idat.atypes = make_pair( ff_->getAtomType(idents[atom1]), |
947 |
+ |
// ff_->getAtomType(idents[atom2]) ); |
948 |
+ |
|
949 |
+ |
if (storageLayout_ & DataStorage::dslAmat) { |
950 |
+ |
idat.A1 = &(snap_->atomData.aMat[atom1]); |
951 |
+ |
idat.A2 = &(snap_->atomData.aMat[atom2]); |
952 |
+ |
} |
953 |
+ |
|
954 |
+ |
if (storageLayout_ & DataStorage::dslElectroFrame) { |
955 |
+ |
idat.eFrame1 = &(snap_->atomData.electroFrame[atom1]); |
956 |
+ |
idat.eFrame2 = &(snap_->atomData.electroFrame[atom2]); |
957 |
+ |
} |
958 |
+ |
|
959 |
+ |
if (storageLayout_ & DataStorage::dslTorque) { |
960 |
+ |
idat.t1 = &(snap_->atomData.torque[atom1]); |
961 |
+ |
idat.t2 = &(snap_->atomData.torque[atom2]); |
962 |
+ |
} |
963 |
+ |
|
964 |
+ |
if (storageLayout_ & DataStorage::dslDensity) { |
965 |
+ |
idat.rho1 = &(snap_->atomData.density[atom1]); |
966 |
+ |
idat.rho2 = &(snap_->atomData.density[atom2]); |
967 |
+ |
} |
968 |
+ |
|
969 |
+ |
if (storageLayout_ & DataStorage::dslFunctional) { |
970 |
+ |
idat.frho1 = &(snap_->atomData.functional[atom1]); |
971 |
+ |
idat.frho2 = &(snap_->atomData.functional[atom2]); |
972 |
+ |
} |
973 |
+ |
|
974 |
+ |
if (storageLayout_ & DataStorage::dslFunctionalDerivative) { |
975 |
+ |
idat.dfrho1 = &(snap_->atomData.functionalDerivative[atom1]); |
976 |
+ |
idat.dfrho2 = &(snap_->atomData.functionalDerivative[atom2]); |
977 |
+ |
} |
978 |
+ |
|
979 |
+ |
if (storageLayout_ & DataStorage::dslParticlePot) { |
980 |
+ |
idat.particlePot1 = &(snap_->atomData.particlePot[atom1]); |
981 |
+ |
idat.particlePot2 = &(snap_->atomData.particlePot[atom2]); |
982 |
+ |
} |
983 |
+ |
|
984 |
+ |
if (storageLayout_ & DataStorage::dslSkippedCharge) { |
985 |
+ |
idat.skippedCharge1 = &(snap_->atomData.skippedCharge[atom1]); |
986 |
+ |
idat.skippedCharge2 = &(snap_->atomData.skippedCharge[atom2]); |
987 |
+ |
} |
988 |
|
#endif |
989 |
|
} |
990 |
+ |
|
991 |
|
|
992 |
+ |
void ForceMatrixDecomposition::unpackInteractionData(InteractionData &idat, int atom1, int atom2) { |
993 |
+ |
#ifdef IS_MPI |
994 |
+ |
pot_row[atom1] += 0.5 * *(idat.pot); |
995 |
+ |
pot_col[atom2] += 0.5 * *(idat.pot); |
996 |
+ |
|
997 |
+ |
atomRowData.force[atom1] += *(idat.f1); |
998 |
+ |
atomColData.force[atom2] -= *(idat.f1); |
999 |
+ |
#else |
1000 |
+ |
pairwisePot += *(idat.pot); |
1001 |
+ |
|
1002 |
+ |
snap_->atomData.force[atom1] += *(idat.f1); |
1003 |
+ |
snap_->atomData.force[atom2] -= *(idat.f1); |
1004 |
+ |
#endif |
1005 |
+ |
|
1006 |
+ |
} |
1007 |
+ |
|
1008 |
+ |
/* |
1009 |
+ |
* buildNeighborList |
1010 |
+ |
* |
1011 |
+ |
* first element of pair is row-indexed CutoffGroup |
1012 |
+ |
* second element of pair is column-indexed CutoffGroup |
1013 |
+ |
*/ |
1014 |
+ |
vector<pair<int, int> > ForceMatrixDecomposition::buildNeighborList() { |
1015 |
+ |
|
1016 |
+ |
vector<pair<int, int> > neighborList; |
1017 |
+ |
groupCutoffs cuts; |
1018 |
+ |
bool doAllPairs = false; |
1019 |
+ |
|
1020 |
+ |
#ifdef IS_MPI |
1021 |
+ |
cellListRow_.clear(); |
1022 |
+ |
cellListCol_.clear(); |
1023 |
+ |
#else |
1024 |
+ |
cellList_.clear(); |
1025 |
+ |
#endif |
1026 |
+ |
|
1027 |
+ |
RealType rList_ = (largestRcut_ + skinThickness_); |
1028 |
+ |
RealType rl2 = rList_ * rList_; |
1029 |
+ |
Snapshot* snap_ = sman_->getCurrentSnapshot(); |
1030 |
+ |
Mat3x3d Hmat = snap_->getHmat(); |
1031 |
+ |
Vector3d Hx = Hmat.getColumn(0); |
1032 |
+ |
Vector3d Hy = Hmat.getColumn(1); |
1033 |
+ |
Vector3d Hz = Hmat.getColumn(2); |
1034 |
+ |
|
1035 |
+ |
nCells_.x() = (int) ( Hx.length() )/ rList_; |
1036 |
+ |
nCells_.y() = (int) ( Hy.length() )/ rList_; |
1037 |
+ |
nCells_.z() = (int) ( Hz.length() )/ rList_; |
1038 |
+ |
|
1039 |
+ |
// handle small boxes where the cell offsets can end up repeating cells |
1040 |
+ |
|
1041 |
+ |
if (nCells_.x() < 3) doAllPairs = true; |
1042 |
+ |
if (nCells_.y() < 3) doAllPairs = true; |
1043 |
+ |
if (nCells_.z() < 3) doAllPairs = true; |
1044 |
+ |
|
1045 |
+ |
Mat3x3d invHmat = snap_->getInvHmat(); |
1046 |
+ |
Vector3d rs, scaled, dr; |
1047 |
+ |
Vector3i whichCell; |
1048 |
+ |
int cellIndex; |
1049 |
+ |
int nCtot = nCells_.x() * nCells_.y() * nCells_.z(); |
1050 |
+ |
|
1051 |
+ |
#ifdef IS_MPI |
1052 |
+ |
cellListRow_.resize(nCtot); |
1053 |
+ |
cellListCol_.resize(nCtot); |
1054 |
+ |
#else |
1055 |
+ |
cellList_.resize(nCtot); |
1056 |
+ |
#endif |
1057 |
+ |
|
1058 |
+ |
if (!doAllPairs) { |
1059 |
+ |
#ifdef IS_MPI |
1060 |
+ |
|
1061 |
+ |
for (int i = 0; i < nGroupsInRow_; i++) { |
1062 |
+ |
rs = cgRowData.position[i]; |
1063 |
+ |
|
1064 |
+ |
// scaled positions relative to the box vectors |
1065 |
+ |
scaled = invHmat * rs; |
1066 |
+ |
|
1067 |
+ |
// wrap the vector back into the unit box by subtracting integer box |
1068 |
+ |
// numbers |
1069 |
+ |
for (int j = 0; j < 3; j++) { |
1070 |
+ |
scaled[j] -= roundMe(scaled[j]); |
1071 |
+ |
scaled[j] += 0.5; |
1072 |
+ |
} |
1073 |
+ |
|
1074 |
+ |
// find xyz-indices of cell that cutoffGroup is in. |
1075 |
+ |
whichCell.x() = nCells_.x() * scaled.x(); |
1076 |
+ |
whichCell.y() = nCells_.y() * scaled.y(); |
1077 |
+ |
whichCell.z() = nCells_.z() * scaled.z(); |
1078 |
+ |
|
1079 |
+ |
// find single index of this cell: |
1080 |
+ |
cellIndex = Vlinear(whichCell, nCells_); |
1081 |
+ |
|
1082 |
+ |
// add this cutoff group to the list of groups in this cell; |
1083 |
+ |
cellListRow_[cellIndex].push_back(i); |
1084 |
+ |
} |
1085 |
+ |
for (int i = 0; i < nGroupsInCol_; i++) { |
1086 |
+ |
rs = cgColData.position[i]; |
1087 |
+ |
|
1088 |
+ |
// scaled positions relative to the box vectors |
1089 |
+ |
scaled = invHmat * rs; |
1090 |
+ |
|
1091 |
+ |
// wrap the vector back into the unit box by subtracting integer box |
1092 |
+ |
// numbers |
1093 |
+ |
for (int j = 0; j < 3; j++) { |
1094 |
+ |
scaled[j] -= roundMe(scaled[j]); |
1095 |
+ |
scaled[j] += 0.5; |
1096 |
+ |
} |
1097 |
+ |
|
1098 |
+ |
// find xyz-indices of cell that cutoffGroup is in. |
1099 |
+ |
whichCell.x() = nCells_.x() * scaled.x(); |
1100 |
+ |
whichCell.y() = nCells_.y() * scaled.y(); |
1101 |
+ |
whichCell.z() = nCells_.z() * scaled.z(); |
1102 |
+ |
|
1103 |
+ |
// find single index of this cell: |
1104 |
+ |
cellIndex = Vlinear(whichCell, nCells_); |
1105 |
+ |
|
1106 |
+ |
// add this cutoff group to the list of groups in this cell; |
1107 |
+ |
cellListCol_[cellIndex].push_back(i); |
1108 |
+ |
} |
1109 |
+ |
|
1110 |
+ |
#else |
1111 |
+ |
for (int i = 0; i < nGroups_; i++) { |
1112 |
+ |
rs = snap_->cgData.position[i]; |
1113 |
+ |
|
1114 |
+ |
// scaled positions relative to the box vectors |
1115 |
+ |
scaled = invHmat * rs; |
1116 |
+ |
|
1117 |
+ |
// wrap the vector back into the unit box by subtracting integer box |
1118 |
+ |
// numbers |
1119 |
+ |
for (int j = 0; j < 3; j++) { |
1120 |
+ |
scaled[j] -= roundMe(scaled[j]); |
1121 |
+ |
scaled[j] += 0.5; |
1122 |
+ |
} |
1123 |
+ |
|
1124 |
+ |
// find xyz-indices of cell that cutoffGroup is in. |
1125 |
+ |
whichCell.x() = nCells_.x() * scaled.x(); |
1126 |
+ |
whichCell.y() = nCells_.y() * scaled.y(); |
1127 |
+ |
whichCell.z() = nCells_.z() * scaled.z(); |
1128 |
+ |
|
1129 |
+ |
// find single index of this cell: |
1130 |
+ |
cellIndex = Vlinear(whichCell, nCells_); |
1131 |
+ |
|
1132 |
+ |
// add this cutoff group to the list of groups in this cell; |
1133 |
+ |
cellList_[cellIndex].push_back(i); |
1134 |
+ |
} |
1135 |
+ |
|
1136 |
+ |
#endif |
1137 |
+ |
|
1138 |
+ |
for (int m1z = 0; m1z < nCells_.z(); m1z++) { |
1139 |
+ |
for (int m1y = 0; m1y < nCells_.y(); m1y++) { |
1140 |
+ |
for (int m1x = 0; m1x < nCells_.x(); m1x++) { |
1141 |
+ |
Vector3i m1v(m1x, m1y, m1z); |
1142 |
+ |
int m1 = Vlinear(m1v, nCells_); |
1143 |
+ |
|
1144 |
+ |
for (vector<Vector3i>::iterator os = cellOffsets_.begin(); |
1145 |
+ |
os != cellOffsets_.end(); ++os) { |
1146 |
+ |
|
1147 |
+ |
Vector3i m2v = m1v + (*os); |
1148 |
+ |
|
1149 |
+ |
|
1150 |
+ |
if (m2v.x() >= nCells_.x()) { |
1151 |
+ |
m2v.x() = 0; |
1152 |
+ |
} else if (m2v.x() < 0) { |
1153 |
+ |
m2v.x() = nCells_.x() - 1; |
1154 |
+ |
} |
1155 |
+ |
|
1156 |
+ |
if (m2v.y() >= nCells_.y()) { |
1157 |
+ |
m2v.y() = 0; |
1158 |
+ |
} else if (m2v.y() < 0) { |
1159 |
+ |
m2v.y() = nCells_.y() - 1; |
1160 |
+ |
} |
1161 |
+ |
|
1162 |
+ |
if (m2v.z() >= nCells_.z()) { |
1163 |
+ |
m2v.z() = 0; |
1164 |
+ |
} else if (m2v.z() < 0) { |
1165 |
+ |
m2v.z() = nCells_.z() - 1; |
1166 |
+ |
} |
1167 |
+ |
|
1168 |
+ |
int m2 = Vlinear (m2v, nCells_); |
1169 |
+ |
|
1170 |
+ |
#ifdef IS_MPI |
1171 |
+ |
for (vector<int>::iterator j1 = cellListRow_[m1].begin(); |
1172 |
+ |
j1 != cellListRow_[m1].end(); ++j1) { |
1173 |
+ |
for (vector<int>::iterator j2 = cellListCol_[m2].begin(); |
1174 |
+ |
j2 != cellListCol_[m2].end(); ++j2) { |
1175 |
+ |
|
1176 |
+ |
// In parallel, we need to visit *all* pairs of row |
1177 |
+ |
// & column indicies and will divide labor in the |
1178 |
+ |
// force evaluation later. |
1179 |
+ |
dr = cgColData.position[(*j2)] - cgRowData.position[(*j1)]; |
1180 |
+ |
snap_->wrapVector(dr); |
1181 |
+ |
cuts = getGroupCutoffs( (*j1), (*j2) ); |
1182 |
+ |
if (dr.lengthSquare() < cuts.third) { |
1183 |
+ |
neighborList.push_back(make_pair((*j1), (*j2))); |
1184 |
+ |
} |
1185 |
+ |
} |
1186 |
+ |
} |
1187 |
+ |
#else |
1188 |
+ |
|
1189 |
+ |
for (vector<int>::iterator j1 = cellList_[m1].begin(); |
1190 |
+ |
j1 != cellList_[m1].end(); ++j1) { |
1191 |
+ |
for (vector<int>::iterator j2 = cellList_[m2].begin(); |
1192 |
+ |
j2 != cellList_[m2].end(); ++j2) { |
1193 |
+ |
|
1194 |
+ |
// Always do this if we're in different cells or if |
1195 |
+ |
// we're in the same cell and the global index of the |
1196 |
+ |
// j2 cutoff group is less than the j1 cutoff group |
1197 |
+ |
|
1198 |
+ |
if (m2 != m1 || (*j2) < (*j1)) { |
1199 |
+ |
dr = snap_->cgData.position[(*j2)] - snap_->cgData.position[(*j1)]; |
1200 |
+ |
snap_->wrapVector(dr); |
1201 |
+ |
cuts = getGroupCutoffs( (*j1), (*j2) ); |
1202 |
+ |
if (dr.lengthSquare() < cuts.third) { |
1203 |
+ |
neighborList.push_back(make_pair((*j1), (*j2))); |
1204 |
+ |
} |
1205 |
+ |
} |
1206 |
+ |
} |
1207 |
+ |
} |
1208 |
+ |
#endif |
1209 |
+ |
} |
1210 |
+ |
} |
1211 |
+ |
} |
1212 |
+ |
} |
1213 |
+ |
} else { |
1214 |
+ |
// branch to do all cutoff group pairs |
1215 |
+ |
#ifdef IS_MPI |
1216 |
+ |
for (int j1 = 0; j1 < nGroupsInRow_; j1++) { |
1217 |
+ |
for (int j2 = 0; j2 < nGroupsInCol_; j2++) { |
1218 |
+ |
dr = cgColData.position[j2] - cgRowData.position[j1]; |
1219 |
+ |
snap_->wrapVector(dr); |
1220 |
+ |
cuts = getGroupCutoffs( j1, j2 ); |
1221 |
+ |
if (dr.lengthSquare() < cuts.third) { |
1222 |
+ |
neighborList.push_back(make_pair(j1, j2)); |
1223 |
+ |
} |
1224 |
+ |
} |
1225 |
+ |
} |
1226 |
+ |
#else |
1227 |
+ |
for (int j1 = 0; j1 < nGroups_ - 1; j1++) { |
1228 |
+ |
for (int j2 = j1 + 1; j2 < nGroups_; j2++) { |
1229 |
+ |
dr = snap_->cgData.position[j2] - snap_->cgData.position[j1]; |
1230 |
+ |
snap_->wrapVector(dr); |
1231 |
+ |
cuts = getGroupCutoffs( j1, j2 ); |
1232 |
+ |
if (dr.lengthSquare() < cuts.third) { |
1233 |
+ |
neighborList.push_back(make_pair(j1, j2)); |
1234 |
+ |
} |
1235 |
+ |
} |
1236 |
+ |
} |
1237 |
+ |
#endif |
1238 |
+ |
} |
1239 |
+ |
|
1240 |
+ |
// save the local cutoff group positions for the check that is |
1241 |
+ |
// done on each loop: |
1242 |
+ |
saved_CG_positions_.clear(); |
1243 |
+ |
for (int i = 0; i < nGroups_; i++) |
1244 |
+ |
saved_CG_positions_.push_back(snap_->cgData.position[i]); |
1245 |
+ |
|
1246 |
+ |
return neighborList; |
1247 |
+ |
} |
1248 |
|
} //end namespace OpenMD |