# | Line 1 | Line 1 | |
---|---|---|
1 | #include <iostream> | |
2 | < | #include <cstdlib> |
3 | < | #include <cmath> |
4 | < | |
2 | > | #include <stdlib.h> |
3 | > | #include <math.h> |
4 | > | #include "Rattle.hpp" |
5 | > | #include "Roll.hpp" |
6 | #ifdef IS_MPI | |
7 | #include "mpiSimulation.hpp" | |
8 | #include <unistd.h> | |
9 | #endif //is_mpi | |
10 | ||
11 | + | #ifdef PROFILE |
12 | + | #include "mdProfile.hpp" |
13 | + | #endif // profile |
14 | + | |
15 | #include "Integrator.hpp" | |
16 | #include "simError.h" | |
17 | ||
18 | ||
19 | < | Integrator::Integrator( SimInfo *theInfo, ForceFields* the_ff ){ |
20 | < | |
19 | > | template<typename T> Integrator<T>::Integrator(SimInfo* theInfo, |
20 | > | ForceFields* the_ff){ |
21 | info = theInfo; | |
22 | myFF = the_ff; | |
23 | isFirst = 1; | |
# | Line 21 | Line 26 | Integrator::Integrator( SimInfo *theInfo, ForceFields* | |
26 | nMols = info->n_mol; | |
27 | ||
28 | // give a little love back to the SimInfo object | |
24 | – | |
25 | – | if( info->the_integrator != NULL ) delete info->the_integrator; |
26 | – | info->the_integrator = this; |
29 | ||
30 | + | if (info->the_integrator != NULL){ |
31 | + | delete info->the_integrator; |
32 | + | } |
33 | + | |
34 | nAtoms = info->n_atoms; | |
35 | + | integrableObjects = info->integrableObjects; |
36 | ||
37 | < | // check for constraints |
37 | > | consFramework = new RattleFramework(info); |
38 | > | |
39 | > | if(consFramework == NULL){ |
40 | > | sprintf(painCave.errMsg, |
41 | > | "Integrator::Intergrator() Error: Memory allocation error for RattleFramework" ); |
42 | > | painCave.isFatal = 1; |
43 | > | simError(); |
44 | > | } |
45 | ||
46 | < | constrainedA = NULL; |
47 | < | constrainedB = NULL; |
46 | > | /* |
47 | > | // check for constraints |
48 | > | |
49 | > | constrainedA = NULL; |
50 | > | constrainedB = NULL; |
51 | constrainedDsqr = NULL; | |
52 | < | moving = NULL; |
53 | < | moved = NULL; |
54 | < | oldPos = NULL; |
55 | < | |
52 | > | moving = NULL; |
53 | > | moved = NULL; |
54 | > | oldPos = NULL; |
55 | > | |
56 | nConstrained = 0; | |
57 | ||
58 | checkConstraints(); | |
59 | + | */ |
60 | } | |
61 | ||
62 | < | Integrator::~Integrator() { |
63 | < | |
64 | < | if( nConstrained ){ |
62 | > | template<typename T> Integrator<T>::~Integrator(){ |
63 | > | if (consFramework != NULL) |
64 | > | delete consFramework; |
65 | > | /* |
66 | > | if (nConstrained){ |
67 | delete[] constrainedA; | |
68 | delete[] constrainedB; | |
69 | delete[] constrainedDsqr; | |
# | Line 51 | Line 71 | Integrator::~Integrator() { | |
71 | delete[] moved; | |
72 | delete[] oldPos; | |
73 | } | |
74 | < | |
74 | > | */ |
75 | } | |
76 | ||
77 | < | void Integrator::checkConstraints( void ){ |
78 | < | |
59 | < | |
77 | > | /* |
78 | > | template<typename T> void Integrator<T>::checkConstraints(void){ |
79 | isConstrained = 0; | |
80 | ||
81 | < | Constraint *temp_con; |
82 | < | Constraint *dummy_plug; |
81 | > | Constraint* temp_con; |
82 | > | Constraint* dummy_plug; |
83 | temp_con = new Constraint[info->n_SRI]; | |
84 | nConstrained = 0; | |
85 | int constrained = 0; | |
86 | < | |
86 | > | |
87 | SRI** theArray; | |
88 | < | for(int i = 0; i < nMols; i++){ |
89 | < | |
90 | < | theArray = (SRI**) molecules[i].getMyBonds(); |
91 | < | for(int j=0; j<molecules[i].getNBonds(); j++){ |
73 | < | |
88 | > | for (int i = 0; i < nMols; i++){ |
89 | > | |
90 | > | theArray = (SRI * *) molecules[i].getMyBonds(); |
91 | > | for (int j = 0; j < molecules[i].getNBonds(); j++){ |
92 | constrained = theArray[j]->is_constrained(); | |
93 | ||
94 | < | if(constrained){ |
94 | > | if (constrained){ |
95 | > | dummy_plug = theArray[j]->get_constraint(); |
96 | > | temp_con[nConstrained].set_a(dummy_plug->get_a()); |
97 | > | temp_con[nConstrained].set_b(dummy_plug->get_b()); |
98 | > | temp_con[nConstrained].set_dsqr(dummy_plug->get_dsqr()); |
99 | ||
100 | < | dummy_plug = theArray[j]->get_constraint(); |
101 | < | temp_con[nConstrained].set_a( dummy_plug->get_a() ); |
102 | < | temp_con[nConstrained].set_b( dummy_plug->get_b() ); |
81 | < | temp_con[nConstrained].set_dsqr( dummy_plug->get_dsqr() ); |
82 | < | |
83 | < | nConstrained++; |
84 | < | constrained = 0; |
85 | < | } |
100 | > | nConstrained++; |
101 | > | constrained = 0; |
102 | > | } |
103 | } | |
104 | ||
105 | < | theArray = (SRI**) molecules[i].getMyBends(); |
106 | < | for(int j=0; j<molecules[i].getNBends(); j++){ |
90 | < | |
105 | > | theArray = (SRI * *) molecules[i].getMyBends(); |
106 | > | for (int j = 0; j < molecules[i].getNBends(); j++){ |
107 | constrained = theArray[j]->is_constrained(); | |
108 | < | |
109 | < | if(constrained){ |
110 | < | |
111 | < | dummy_plug = theArray[j]->get_constraint(); |
112 | < | temp_con[nConstrained].set_a( dummy_plug->get_a() ); |
113 | < | temp_con[nConstrained].set_b( dummy_plug->get_b() ); |
114 | < | temp_con[nConstrained].set_dsqr( dummy_plug->get_dsqr() ); |
115 | < | |
116 | < | nConstrained++; |
101 | < | constrained = 0; |
108 | > | |
109 | > | if (constrained){ |
110 | > | dummy_plug = theArray[j]->get_constraint(); |
111 | > | temp_con[nConstrained].set_a(dummy_plug->get_a()); |
112 | > | temp_con[nConstrained].set_b(Dummy_plug->get_b()); |
113 | > | temp_con[nConstrained].set_dsqr(dummy_plug->get_dsqr()); |
114 | > | |
115 | > | nConstrained++; |
116 | > | constrained = 0; |
117 | } | |
118 | } | |
119 | ||
120 | < | theArray = (SRI**) molecules[i].getMyTorsions(); |
121 | < | for(int j=0; j<molecules[i].getNTorsions(); j++){ |
107 | < | |
120 | > | theArray = (SRI * *) molecules[i].getMyTorsions(); |
121 | > | for (int j = 0; j < molecules[i].getNTorsions(); j++){ |
122 | constrained = theArray[j]->is_constrained(); | |
123 | < | |
124 | < | if(constrained){ |
125 | < | |
126 | < | dummy_plug = theArray[j]->get_constraint(); |
127 | < | temp_con[nConstrained].set_a( dummy_plug->get_a() ); |
128 | < | temp_con[nConstrained].set_b( dummy_plug->get_b() ); |
129 | < | temp_con[nConstrained].set_dsqr( dummy_plug->get_dsqr() ); |
130 | < | |
131 | < | nConstrained++; |
118 | < | constrained = 0; |
123 | > | |
124 | > | if (constrained){ |
125 | > | dummy_plug = theArray[j]->get_constraint(); |
126 | > | temp_con[nConstrained].set_a(dummy_plug->get_a()); |
127 | > | temp_con[nConstrained].set_b(dummy_plug->get_b()); |
128 | > | temp_con[nConstrained].set_dsqr(dummy_plug->get_dsqr()); |
129 | > | |
130 | > | nConstrained++; |
131 | > | constrained = 0; |
132 | } | |
133 | } | |
134 | } | |
135 | ||
136 | < | if(nConstrained > 0){ |
137 | < | |
136 | > | |
137 | > | if (nConstrained > 0){ |
138 | isConstrained = 1; | |
139 | ||
140 | < | if(constrainedA != NULL ) delete[] constrainedA; |
141 | < | if(constrainedB != NULL ) delete[] constrainedB; |
142 | < | if(constrainedDsqr != NULL ) delete[] constrainedDsqr; |
140 | > | if (constrainedA != NULL) |
141 | > | delete[] constrainedA; |
142 | > | if (constrainedB != NULL) |
143 | > | delete[] constrainedB; |
144 | > | if (constrainedDsqr != NULL) |
145 | > | delete[] constrainedDsqr; |
146 | ||
147 | < | constrainedA = new int[nConstrained]; |
148 | < | constrainedB = new int[nConstrained]; |
147 | > | constrainedA = new int[nConstrained]; |
148 | > | constrainedB = new int[nConstrained]; |
149 | constrainedDsqr = new double[nConstrained]; | |
150 | < | |
151 | < | for( int i = 0; i < nConstrained; i++){ |
136 | < | |
150 | > | |
151 | > | for (int i = 0; i < nConstrained; i++){ |
152 | constrainedA[i] = temp_con[i].get_a(); | |
153 | constrainedB[i] = temp_con[i].get_b(); | |
154 | constrainedDsqr[i] = temp_con[i].get_dsqr(); | |
140 | – | |
155 | } | |
156 | ||
157 | < | |
158 | < | // save oldAtoms to check for lode balanceing later on. |
159 | < | |
157 | > | |
158 | > | // save oldAtoms to check for lode balancing later on. |
159 | > | |
160 | oldAtoms = nAtoms; | |
161 | < | |
161 | > | |
162 | moving = new int[nAtoms]; | |
163 | < | moved = new int[nAtoms]; |
163 | > | moved = new int[nAtoms]; |
164 | ||
165 | < | oldPos = new double[nAtoms*3]; |
165 | > | oldPos = new double[nAtoms * 3]; |
166 | } | |
167 | < | |
167 | > | |
168 | delete[] temp_con; | |
169 | } | |
170 | + | */ |
171 | ||
172 | + | template<typename T> void Integrator<T>::integrate(void){ |
173 | ||
174 | < | void Integrator::integrate( void ){ |
175 | < | |
176 | < | int i, j; // loop counters |
161 | < | |
162 | < | double runTime = info->run_time; |
163 | < | double sampleTime = info->sampleTime; |
164 | < | double statusTime = info->statusTime; |
174 | > | double runTime = info->run_time; |
175 | > | double sampleTime = info->sampleTime; |
176 | > | double statusTime = info->statusTime; |
177 | double thermalTime = info->thermalTime; | |
178 | + | double resetTime = info->resetTime; |
179 | ||
180 | + | double difference; |
181 | double currSample; | |
182 | double currThermal; | |
183 | double currStatus; | |
184 | + | double currReset; |
185 | ||
186 | int calcPot, calcStress; | |
172 | – | int isError; |
187 | ||
188 | < | tStats = new Thermo( info ); |
189 | < | statOut = new StatWriter( info ); |
190 | < | dumpOut = new DumpWriter( info ); |
188 | > | tStats = new Thermo(info); |
189 | > | statOut = new StatWriter(info); |
190 | > | dumpOut = new DumpWriter(info); |
191 | ||
192 | atoms = info->atoms; | |
179 | – | DirectionalAtom* dAtom; |
193 | ||
194 | dt = info->dt; | |
195 | dt2 = 0.5 * dt; | |
196 | ||
197 | + | readyCheck(); |
198 | + | |
199 | + | // remove center of mass drift velocity (in case we passed in a configuration |
200 | + | // that was drifting |
201 | + | tStats->removeCOMdrift(); |
202 | + | |
203 | + | // initialize the retraints if necessary |
204 | + | if (info->useSolidThermInt && !info->useLiquidThermInt) { |
205 | + | myFF->initRestraints(); |
206 | + | } |
207 | + | |
208 | // initialize the forces before the first step | |
209 | ||
210 | < | myFF->doForces(1,1); |
210 | > | calcForce(1, 1); |
211 | > | |
212 | > | //execute constraint algorithm to make sure at the very beginning the system is constrained |
213 | > | consFramework->doPreConstraint(); |
214 | > | consFramework->doConstrainA(); |
215 | > | calcForce(1, 1); |
216 | > | consFramework->doConstrainB(); |
217 | ||
218 | < | if( info->setTemp ){ |
219 | < | |
190 | < | tStats->velocitize(); |
218 | > | if (info->setTemp){ |
219 | > | thermalize(); |
220 | } | |
221 | < | |
221 | > | |
222 | calcPot = 0; | |
223 | calcStress = 0; | |
224 | < | currSample = sampleTime; |
225 | < | currThermal = thermalTime; |
226 | < | currStatus = statusTime; |
224 | > | currSample = sampleTime + info->getTime(); |
225 | > | currThermal = thermalTime+ info->getTime(); |
226 | > | currStatus = statusTime + info->getTime(); |
227 | > | currReset = resetTime + info->getTime(); |
228 | ||
229 | < | dumpOut->writeDump( info->getTime() ); |
230 | < | statOut->writeStat( info->getTime() ); |
229 | > | dumpOut->writeDump(info->getTime()); |
230 | > | statOut->writeStat(info->getTime()); |
231 | ||
202 | – | readyCheck(); |
232 | ||
233 | #ifdef IS_MPI | |
234 | < | strcpy( checkPointMsg, |
206 | < | "The integrator is ready to go." ); |
234 | > | strcpy(checkPointMsg, "The integrator is ready to go."); |
235 | MPIcheckPoint(); | |
236 | #endif // is_mpi | |
237 | ||
238 | < | while( info->getTime() < runTime ){ |
239 | < | |
240 | < | if( (info->getTime()+dt) >= currStatus ){ |
238 | > | while (info->getTime() < runTime && !stopIntegrator()){ |
239 | > | difference = info->getTime() + dt - currStatus; |
240 | > | if (difference > 0 || fabs(difference) < 1e-4 ){ |
241 | calcPot = 1; | |
242 | calcStress = 1; | |
243 | } | |
244 | ||
245 | < | integrateStep( calcPot, calcStress ); |
246 | < | |
245 | > | #ifdef PROFILE |
246 | > | startProfile( pro1 ); |
247 | > | #endif |
248 | > | |
249 | > | integrateStep(calcPot, calcStress); |
250 | > | |
251 | > | #ifdef PROFILE |
252 | > | endProfile( pro1 ); |
253 | > | |
254 | > | startProfile( pro2 ); |
255 | > | #endif // profile |
256 | > | |
257 | info->incrTime(dt); | |
258 | ||
259 | < | if( info->setTemp ){ |
260 | < | if( info->getTime() >= currThermal ){ |
261 | < | tStats->velocitize(); |
262 | < | currThermal += thermalTime; |
259 | > | if (info->setTemp){ |
260 | > | if (info->getTime() >= currThermal){ |
261 | > | thermalize(); |
262 | > | currThermal += thermalTime; |
263 | } | |
264 | } | |
265 | ||
266 | < | if( info->getTime() >= currSample ){ |
267 | < | dumpOut->writeDump( info->getTime() ); |
266 | > | if (info->getTime() >= currSample){ |
267 | > | dumpOut->writeDump(info->getTime()); |
268 | currSample += sampleTime; | |
269 | } | |
270 | ||
271 | < | if( info->getTime() >= currStatus ){ |
272 | < | statOut->writeStat( info->getTime() ); |
273 | < | calcPot = 0; |
271 | > | if (info->getTime() >= currStatus){ |
272 | > | statOut->writeStat(info->getTime()); |
273 | > | calcPot = 0; |
274 | calcStress = 0; | |
275 | currStatus += statusTime; | |
276 | < | } |
276 | > | } |
277 | ||
278 | + | if (info->resetIntegrator){ |
279 | + | if (info->getTime() >= currReset){ |
280 | + | this->resetIntegrator(); |
281 | + | currReset += resetTime; |
282 | + | } |
283 | + | } |
284 | + | |
285 | + | #ifdef PROFILE |
286 | + | endProfile( pro2 ); |
287 | + | #endif //profile |
288 | + | |
289 | #ifdef IS_MPI | |
290 | < | strcpy( checkPointMsg, |
242 | < | "successfully took a time step." ); |
290 | > | strcpy(checkPointMsg, "successfully took a time step."); |
291 | MPIcheckPoint(); | |
292 | #endif // is_mpi | |
245 | – | |
293 | } | |
294 | ||
295 | < | dumpOut->writeFinal(info->getTime()); |
295 | > | // dump out a file containing the omega values for the final configuration |
296 | > | if (info->useSolidThermInt && !info->useLiquidThermInt) |
297 | > | myFF->dumpzAngle(); |
298 | > | |
299 | ||
300 | delete dumpOut; | |
301 | delete statOut; | |
302 | } | |
303 | ||
304 | < | void Integrator::integrateStep( int calcPot, int calcStress ){ |
304 | > | template<typename T> void Integrator<T>::integrateStep(int calcPot, |
305 | > | int calcStress){ |
306 | > | // Position full step, and velocity half step |
307 | ||
308 | + | #ifdef PROFILE |
309 | + | startProfile(pro3); |
310 | + | #endif //profile |
311 | ||
312 | < | |
313 | < | // Position full step, and velocity half step |
312 | > | //save old state (position, velocity etc) |
313 | > | consFramework->doPreConstraint(); |
314 | ||
315 | < | preMove(); |
315 | > | #ifdef PROFILE |
316 | > | endProfile(pro3); |
317 | > | |
318 | > | startProfile(pro4); |
319 | > | #endif // profile |
320 | > | |
321 | moveA(); | |
262 | – | if( nConstrained ) constrainA(); |
322 | ||
323 | + | #ifdef PROFILE |
324 | + | endProfile(pro4); |
325 | ||
326 | + | startProfile(pro5); |
327 | + | #endif//profile |
328 | + | |
329 | + | |
330 | #ifdef IS_MPI | |
331 | < | strcpy( checkPointMsg, "Succesful moveA\n" ); |
331 | > | strcpy(checkPointMsg, "Succesful moveA\n"); |
332 | MPIcheckPoint(); | |
333 | #endif // is_mpi | |
269 | – | |
334 | ||
335 | // calc forces | |
336 | + | calcForce(calcPot, calcStress); |
337 | ||
273 | – | myFF->doForces(calcPot,calcStress); |
274 | – | |
338 | #ifdef IS_MPI | |
339 | < | strcpy( checkPointMsg, "Succesful doForces\n" ); |
339 | > | strcpy(checkPointMsg, "Succesful doForces\n"); |
340 | MPIcheckPoint(); | |
341 | #endif // is_mpi | |
279 | – | |
342 | ||
343 | + | #ifdef PROFILE |
344 | + | endProfile( pro5 ); |
345 | + | |
346 | + | startProfile( pro6 ); |
347 | + | #endif //profile |
348 | + | |
349 | // finish the velocity half step | |
350 | < | |
350 | > | |
351 | moveB(); | |
352 | < | if( nConstrained ) constrainB(); |
353 | < | |
352 | > | |
353 | > | #ifdef PROFILE |
354 | > | endProfile(pro6); |
355 | > | #endif // profile |
356 | > | |
357 | #ifdef IS_MPI | |
358 | < | strcpy( checkPointMsg, "Succesful moveB\n" ); |
358 | > | strcpy(checkPointMsg, "Succesful moveB\n"); |
359 | MPIcheckPoint(); | |
360 | #endif // is_mpi | |
290 | – | |
291 | – | |
361 | } | |
362 | ||
363 | ||
364 | < | void Integrator::moveA( void ){ |
365 | < | |
297 | < | int i, j; |
364 | > | template<typename T> void Integrator<T>::moveA(void){ |
365 | > | size_t i, j; |
366 | DirectionalAtom* dAtom; | |
367 | double Tb[3], ji[3]; | |
300 | – | double A[3][3], I[3][3]; |
301 | – | double angle; |
368 | double vel[3], pos[3], frc[3]; | |
369 | double mass; | |
370 | + | double omega; |
371 | + | |
372 | + | for (i = 0; i < integrableObjects.size() ; i++){ |
373 | + | integrableObjects[i]->getVel(vel); |
374 | + | integrableObjects[i]->getPos(pos); |
375 | + | integrableObjects[i]->getFrc(frc); |
376 | + | |
377 | + | mass = integrableObjects[i]->getMass(); |
378 | ||
379 | < | for( i=0; i<nAtoms; i++ ){ |
306 | < | |
307 | < | atoms[i]->getVel( vel ); |
308 | < | atoms[i]->getPos( pos ); |
309 | < | atoms[i]->getFrc( frc ); |
310 | < | |
311 | < | mass = atoms[i]->getMass(); |
312 | < | |
313 | < | for (j=0; j < 3; j++) { |
379 | > | for (j = 0; j < 3; j++){ |
380 | // velocity half step | |
381 | < | vel[j] += ( dt2 * frc[j] / mass ) * eConvert; |
381 | > | vel[j] += (dt2 * frc[j] / mass) * eConvert; |
382 | // position whole step | |
383 | pos[j] += dt * vel[j]; | |
384 | } | |
385 | ||
386 | < | atoms[i]->setVel( vel ); |
387 | < | atoms[i]->setPos( pos ); |
386 | > | integrableObjects[i]->setVel(vel); |
387 | > | integrableObjects[i]->setPos(pos); |
388 | ||
389 | < | if( atoms[i]->isDirectional() ){ |
389 | > | if (integrableObjects[i]->isDirectional()){ |
390 | ||
325 | – | dAtom = (DirectionalAtom *)atoms[i]; |
326 | – | |
391 | // get and convert the torque to body frame | |
328 | – | |
329 | – | dAtom->getTrq( Tb ); |
330 | – | dAtom->lab2Body( Tb ); |
392 | ||
393 | + | integrableObjects[i]->getTrq(Tb); |
394 | + | integrableObjects[i]->lab2Body(Tb); |
395 | + | |
396 | // get the angular momentum, and propagate a half step | |
397 | ||
398 | < | dAtom->getJ( ji ); |
398 | > | integrableObjects[i]->getJ(ji); |
399 | ||
400 | < | for (j=0; j < 3; j++) |
400 | > | for (j = 0; j < 3; j++) |
401 | ji[j] += (dt2 * Tb[j]) * eConvert; | |
338 | – | |
339 | – | // use the angular velocities to propagate the rotation matrix a |
340 | – | // full time step |
402 | ||
403 | < | dAtom->getA(A); |
343 | < | dAtom->getI(I); |
344 | < | |
345 | < | // rotate about the x-axis |
346 | < | angle = dt2 * ji[0] / I[0][0]; |
347 | < | this->rotate( 1, 2, angle, ji, A ); |
403 | > | this->rotationPropagation( integrableObjects[i], ji ); |
404 | ||
405 | < | // rotate about the y-axis |
406 | < | angle = dt2 * ji[1] / I[1][1]; |
351 | < | this->rotate( 2, 0, angle, ji, A ); |
352 | < | |
353 | < | // rotate about the z-axis |
354 | < | angle = dt * ji[2] / I[2][2]; |
355 | < | this->rotate( 0, 1, angle, ji, A); |
356 | < | |
357 | < | // rotate about the y-axis |
358 | < | angle = dt2 * ji[1] / I[1][1]; |
359 | < | this->rotate( 2, 0, angle, ji, A ); |
360 | < | |
361 | < | // rotate about the x-axis |
362 | < | angle = dt2 * ji[0] / I[0][0]; |
363 | < | this->rotate( 1, 2, angle, ji, A ); |
364 | < | |
365 | < | |
366 | < | dAtom->setJ( ji ); |
367 | < | dAtom->setA( A ); |
368 | < | |
369 | < | } |
405 | > | integrableObjects[i]->setJ(ji); |
406 | > | } |
407 | } | |
408 | + | |
409 | + | consFramework->doConstrainA(); |
410 | } | |
411 | ||
412 | ||
413 | < | void Integrator::moveB( void ){ |
413 | > | template<typename T> void Integrator<T>::moveB(void){ |
414 | int i, j; | |
376 | – | DirectionalAtom* dAtom; |
415 | double Tb[3], ji[3]; | |
416 | double vel[3], frc[3]; | |
417 | double mass; | |
418 | ||
419 | < | for( i=0; i<nAtoms; i++ ){ |
420 | < | |
421 | < | atoms[i]->getVel( vel ); |
384 | < | atoms[i]->getFrc( frc ); |
419 | > | for (i = 0; i < integrableObjects.size(); i++){ |
420 | > | integrableObjects[i]->getVel(vel); |
421 | > | integrableObjects[i]->getFrc(frc); |
422 | ||
423 | < | mass = atoms[i]->getMass(); |
423 | > | mass = integrableObjects[i]->getMass(); |
424 | ||
425 | // velocity half step | |
426 | < | for (j=0; j < 3; j++) |
427 | < | vel[j] += ( dt2 * frc[j] / mass ) * eConvert; |
391 | < | |
392 | < | atoms[i]->setVel( vel ); |
393 | < | |
394 | < | if( atoms[i]->isDirectional() ){ |
426 | > | for (j = 0; j < 3; j++) |
427 | > | vel[j] += (dt2 * frc[j] / mass) * eConvert; |
428 | ||
429 | < | dAtom = (DirectionalAtom *)atoms[i]; |
429 | > | integrableObjects[i]->setVel(vel); |
430 | ||
431 | < | // get and convert the torque to body frame |
431 | > | if (integrableObjects[i]->isDirectional()){ |
432 | ||
433 | < | dAtom->getTrq( Tb ); |
401 | < | dAtom->lab2Body( Tb ); |
433 | > | // get and convert the torque to body frame |
434 | ||
435 | + | integrableObjects[i]->getTrq(Tb); |
436 | + | integrableObjects[i]->lab2Body(Tb); |
437 | + | |
438 | // get the angular momentum, and propagate a half step | |
439 | ||
440 | < | dAtom->getJ( ji ); |
440 | > | integrableObjects[i]->getJ(ji); |
441 | ||
442 | < | for (j=0; j < 3; j++) |
442 | > | for (j = 0; j < 3; j++) |
443 | ji[j] += (dt2 * Tb[j]) * eConvert; | |
409 | – | |
444 | ||
445 | < | dAtom->setJ( ji ); |
445 | > | |
446 | > | integrableObjects[i]->setJ(ji); |
447 | } | |
448 | } | |
449 | + | |
450 | + | consFramework->doConstrainB(); |
451 | } | |
452 | ||
453 | < | void Integrator::preMove( void ){ |
453 | > | /* |
454 | > | template<typename T> void Integrator<T>::preMove(void){ |
455 | int i, j; | |
456 | double pos[3]; | |
457 | ||
458 | < | if( nConstrained ){ |
459 | < | |
460 | < | for(i=0; i < nAtoms; i++) { |
423 | < | |
424 | < | atoms[i]->getPos( pos ); |
425 | < | |
426 | < | for (j = 0; j < 3; j++) { |
427 | < | oldPos[3*i + j] = pos[j]; |
428 | < | } |
458 | > | if (nConstrained){ |
459 | > | for (i = 0; i < nAtoms; i++){ |
460 | > | atoms[i]->getPos(pos); |
461 | ||
462 | + | for (j = 0; j < 3; j++){ |
463 | + | oldPos[3 * i + j] = pos[j]; |
464 | + | } |
465 | } | |
466 | < | } |
466 | > | } |
467 | } | |
468 | ||
469 | < | void Integrator::constrainA(){ |
470 | < | |
436 | < | int i,j,k; |
469 | > | template<typename T> void Integrator<T>::constrainA(){ |
470 | > | int i, j; |
471 | int done; | |
472 | double posA[3], posB[3]; | |
473 | double velA[3], velB[3]; | |
# | Line 448 | Line 482 | void Integrator::constrainA(){ | |
482 | double gab; | |
483 | int iteration; | |
484 | ||
485 | < | for( i=0; i<nAtoms; i++){ |
485 | > | for (i = 0; i < nAtoms; i++){ |
486 | moving[i] = 0; | |
487 | < | moved[i] = 1; |
487 | > | moved[i] = 1; |
488 | } | |
489 | ||
490 | iteration = 0; | |
491 | done = 0; | |
492 | < | while( !done && (iteration < maxIteration )){ |
459 | < | |
492 | > | while (!done && (iteration < maxIteration)){ |
493 | done = 1; | |
494 | < | for(i=0; i<nConstrained; i++){ |
462 | < | |
494 | > | for (i = 0; i < nConstrained; i++){ |
495 | a = constrainedA[i]; | |
496 | b = constrainedB[i]; | |
465 | – | |
466 | – | ax = (a*3) + 0; |
467 | – | ay = (a*3) + 1; |
468 | – | az = (a*3) + 2; |
497 | ||
498 | < | bx = (b*3) + 0; |
499 | < | by = (b*3) + 1; |
500 | < | bz = (b*3) + 2; |
498 | > | ax = (a * 3) + 0; |
499 | > | ay = (a * 3) + 1; |
500 | > | az = (a * 3) + 2; |
501 | ||
502 | < | if( moved[a] || moved[b] ){ |
503 | < | |
504 | < | atoms[a]->getPos( posA ); |
505 | < | atoms[b]->getPos( posB ); |
506 | < | |
507 | < | for (j = 0; j < 3; j++ ) |
502 | > | bx = (b * 3) + 0; |
503 | > | by = (b * 3) + 1; |
504 | > | bz = (b * 3) + 2; |
505 | > | |
506 | > | if (moved[a] || moved[b]){ |
507 | > | atoms[a]->getPos(posA); |
508 | > | atoms[b]->getPos(posB); |
509 | > | |
510 | > | for (j = 0; j < 3; j++) |
511 | pab[j] = posA[j] - posB[j]; | |
481 | – | |
482 | – | //periodic boundary condition |
512 | ||
513 | < | info->wrapVector( pab ); |
513 | > | //periodic boundary condition |
514 | ||
515 | < | pabsq = pab[0] * pab[0] + pab[1] * pab[1] + pab[2] * pab[2]; |
515 | > | info->wrapVector(pab); |
516 | ||
517 | < | rabsq = constrainedDsqr[i]; |
489 | < | diffsq = rabsq - pabsq; |
517 | > | pabsq = pab[0] * pab[0] + pab[1] * pab[1] + pab[2] * pab[2]; |
518 | ||
519 | < | // the original rattle code from alan tidesley |
520 | < | if (fabs(diffsq) > (tol*rabsq*2)) { |
493 | < | rab[0] = oldPos[ax] - oldPos[bx]; |
494 | < | rab[1] = oldPos[ay] - oldPos[by]; |
495 | < | rab[2] = oldPos[az] - oldPos[bz]; |
519 | > | rabsq = constrainedDsqr[i]; |
520 | > | diffsq = rabsq - pabsq; |
521 | ||
522 | < | info->wrapVector( rab ); |
522 | > | // the original rattle code from alan tidesley |
523 | > | if (fabs(diffsq) > (tol * rabsq * 2)){ |
524 | > | rab[0] = oldPos[ax] - oldPos[bx]; |
525 | > | rab[1] = oldPos[ay] - oldPos[by]; |
526 | > | rab[2] = oldPos[az] - oldPos[bz]; |
527 | ||
528 | < | rpab = rab[0] * pab[0] + rab[1] * pab[1] + rab[2] * pab[2]; |
528 | > | info->wrapVector(rab); |
529 | ||
530 | < | rpabsq = rpab * rpab; |
530 | > | rpab = rab[0] * pab[0] + rab[1] * pab[1] + rab[2] * pab[2]; |
531 | ||
532 | + | rpabsq = rpab * rpab; |
533 | ||
504 | – | if (rpabsq < (rabsq * -diffsq)){ |
534 | ||
535 | + | if (rpabsq < (rabsq * -diffsq)){ |
536 | #ifdef IS_MPI | |
537 | < | a = atoms[a]->getGlobalIndex(); |
538 | < | b = atoms[b]->getGlobalIndex(); |
537 | > | a = atoms[a]->getGlobalIndex(); |
538 | > | b = atoms[b]->getGlobalIndex(); |
539 | #endif //is_mpi | |
540 | < | sprintf( painCave.errMsg, |
541 | < | "Constraint failure in constrainA at atom %d and %d.\n", |
542 | < | a, b ); |
543 | < | painCave.isFatal = 1; |
544 | < | simError(); |
545 | < | } |
540 | > | sprintf(painCave.errMsg, |
541 | > | "Constraint failure in constrainA at atom %d and %d.\n", a, |
542 | > | b); |
543 | > | painCave.isFatal = 1; |
544 | > | simError(); |
545 | > | } |
546 | ||
547 | < | rma = 1.0 / atoms[a]->getMass(); |
548 | < | rmb = 1.0 / atoms[b]->getMass(); |
547 | > | rma = 1.0 / atoms[a]->getMass(); |
548 | > | rmb = 1.0 / atoms[b]->getMass(); |
549 | ||
550 | < | gab = diffsq / ( 2.0 * ( rma + rmb ) * rpab ); |
550 | > | gab = diffsq / (2.0 * (rma + rmb) * rpab); |
551 | ||
552 | dx = rab[0] * gab; | |
553 | dy = rab[1] * gab; | |
554 | dz = rab[2] * gab; | |
555 | ||
556 | < | posA[0] += rma * dx; |
557 | < | posA[1] += rma * dy; |
558 | < | posA[2] += rma * dz; |
556 | > | posA[0] += rma * dx; |
557 | > | posA[1] += rma * dy; |
558 | > | posA[2] += rma * dz; |
559 | ||
560 | < | atoms[a]->setPos( posA ); |
560 | > | atoms[a]->setPos(posA); |
561 | ||
562 | < | posB[0] -= rmb * dx; |
563 | < | posB[1] -= rmb * dy; |
564 | < | posB[2] -= rmb * dz; |
562 | > | posB[0] -= rmb * dx; |
563 | > | posB[1] -= rmb * dy; |
564 | > | posB[2] -= rmb * dz; |
565 | ||
566 | < | atoms[b]->setPos( posB ); |
566 | > | atoms[b]->setPos(posB); |
567 | ||
568 | dx = dx / dt; | |
569 | dy = dy / dt; | |
570 | dz = dz / dt; | |
571 | ||
572 | < | atoms[a]->getVel( velA ); |
572 | > | atoms[a]->getVel(velA); |
573 | ||
574 | < | velA[0] += rma * dx; |
575 | < | velA[1] += rma * dy; |
576 | < | velA[2] += rma * dz; |
574 | > | velA[0] += rma * dx; |
575 | > | velA[1] += rma * dy; |
576 | > | velA[2] += rma * dz; |
577 | ||
578 | < | atoms[a]->setVel( velA ); |
578 | > | atoms[a]->setVel(velA); |
579 | ||
580 | < | atoms[b]->getVel( velB ); |
580 | > | atoms[b]->getVel(velB); |
581 | ||
582 | < | velB[0] -= rmb * dx; |
583 | < | velB[1] -= rmb * dy; |
584 | < | velB[2] -= rmb * dz; |
582 | > | velB[0] -= rmb * dx; |
583 | > | velB[1] -= rmb * dy; |
584 | > | velB[2] -= rmb * dz; |
585 | ||
586 | < | atoms[b]->setVel( velB ); |
586 | > | atoms[b]->setVel(velB); |
587 | ||
588 | < | moving[a] = 1; |
589 | < | moving[b] = 1; |
590 | < | done = 0; |
591 | < | } |
588 | > | moving[a] = 1; |
589 | > | moving[b] = 1; |
590 | > | done = 0; |
591 | > | } |
592 | } | |
593 | } | |
594 | < | |
595 | < | for(i=0; i<nAtoms; i++){ |
566 | < | |
594 | > | |
595 | > | for (i = 0; i < nAtoms; i++){ |
596 | moved[i] = moving[i]; | |
597 | moving[i] = 0; | |
598 | } | |
# | Line 571 | Line 600 | void Integrator::constrainA(){ | |
600 | iteration++; | |
601 | } | |
602 | ||
603 | < | if( !done ){ |
604 | < | |
605 | < | sprintf( painCave.errMsg, |
606 | < | "Constraint failure in constrainA, too many iterations: %d\n", |
578 | < | iteration ); |
603 | > | if (!done){ |
604 | > | sprintf(painCave.errMsg, |
605 | > | "Constraint failure in constrainA, too many iterations: %d\n", |
606 | > | iteration); |
607 | painCave.isFatal = 1; | |
608 | simError(); | |
609 | } | |
610 | ||
611 | } | |
612 | ||
613 | < | void Integrator::constrainB( void ){ |
614 | < | |
587 | < | int i,j,k; |
613 | > | template<typename T> void Integrator<T>::constrainB(void){ |
614 | > | int i, j; |
615 | int done; | |
616 | double posA[3], posB[3]; | |
617 | double velA[3], velB[3]; | |
# | Line 593 | Line 620 | void Integrator::constrainB( void ){ | |
620 | int a, b, ax, ay, az, bx, by, bz; | |
621 | double rma, rmb; | |
622 | double dx, dy, dz; | |
623 | < | double rabsq, pabsq, rvab; |
597 | < | double diffsq; |
623 | > | double rvab; |
624 | double gab; | |
625 | int iteration; | |
626 | ||
627 | < | for(i=0; i<nAtoms; i++){ |
627 | > | for (i = 0; i < nAtoms; i++){ |
628 | moving[i] = 0; | |
629 | moved[i] = 1; | |
630 | } | |
631 | ||
632 | done = 0; | |
633 | iteration = 0; | |
634 | < | while( !done && (iteration < maxIteration ) ){ |
609 | < | |
634 | > | while (!done && (iteration < maxIteration)){ |
635 | done = 1; | |
636 | ||
637 | < | for(i=0; i<nConstrained; i++){ |
613 | < | |
637 | > | for (i = 0; i < nConstrained; i++){ |
638 | a = constrainedA[i]; | |
639 | b = constrainedB[i]; | |
640 | ||
641 | < | ax = (a*3) + 0; |
642 | < | ay = (a*3) + 1; |
643 | < | az = (a*3) + 2; |
641 | > | ax = (a * 3) + 0; |
642 | > | ay = (a * 3) + 1; |
643 | > | az = (a * 3) + 2; |
644 | ||
645 | < | bx = (b*3) + 0; |
646 | < | by = (b*3) + 1; |
647 | < | bz = (b*3) + 2; |
645 | > | bx = (b * 3) + 0; |
646 | > | by = (b * 3) + 1; |
647 | > | bz = (b * 3) + 2; |
648 | ||
649 | < | if( moved[a] || moved[b] ){ |
649 | > | if (moved[a] || moved[b]){ |
650 | > | atoms[a]->getVel(velA); |
651 | > | atoms[b]->getVel(velB); |
652 | ||
653 | < | atoms[a]->getVel( velA ); |
654 | < | atoms[b]->getVel( velB ); |
655 | < | |
630 | < | vxab = velA[0] - velB[0]; |
631 | < | vyab = velA[1] - velB[1]; |
632 | < | vzab = velA[2] - velB[2]; |
653 | > | vxab = velA[0] - velB[0]; |
654 | > | vyab = velA[1] - velB[1]; |
655 | > | vzab = velA[2] - velB[2]; |
656 | ||
657 | < | atoms[a]->getPos( posA ); |
658 | < | atoms[b]->getPos( posB ); |
657 | > | atoms[a]->getPos(posA); |
658 | > | atoms[b]->getPos(posB); |
659 | ||
660 | < | for (j = 0; j < 3; j++) |
660 | > | for (j = 0; j < 3; j++) |
661 | rab[j] = posA[j] - posB[j]; | |
639 | – | |
640 | – | info->wrapVector( rab ); |
641 | – | |
642 | – | rma = 1.0 / atoms[a]->getMass(); |
643 | – | rmb = 1.0 / atoms[b]->getMass(); |
662 | ||
663 | < | rvab = rab[0] * vxab + rab[1] * vyab + rab[2] * vzab; |
646 | < | |
647 | < | gab = -rvab / ( ( rma + rmb ) * constrainedDsqr[i] ); |
663 | > | info->wrapVector(rab); |
664 | ||
665 | < | if (fabs(gab) > tol) { |
666 | < | |
651 | < | dx = rab[0] * gab; |
652 | < | dy = rab[1] * gab; |
653 | < | dz = rab[2] * gab; |
654 | < | |
655 | < | velA[0] += rma * dx; |
656 | < | velA[1] += rma * dy; |
657 | < | velA[2] += rma * dz; |
665 | > | rma = 1.0 / atoms[a]->getMass(); |
666 | > | rmb = 1.0 / atoms[b]->getMass(); |
667 | ||
668 | < | atoms[a]->setVel( velA ); |
668 | > | rvab = rab[0] * vxab + rab[1] * vyab + rab[2] * vzab; |
669 | ||
670 | < | velB[0] -= rmb * dx; |
662 | < | velB[1] -= rmb * dy; |
663 | < | velB[2] -= rmb * dz; |
670 | > | gab = -rvab / ((rma + rmb) * constrainedDsqr[i]); |
671 | ||
672 | < | atoms[b]->setVel( velB ); |
673 | < | |
674 | < | moving[a] = 1; |
675 | < | moving[b] = 1; |
676 | < | done = 0; |
677 | < | } |
672 | > | if (fabs(gab) > tol){ |
673 | > | dx = rab[0] * gab; |
674 | > | dy = rab[1] * gab; |
675 | > | dz = rab[2] * gab; |
676 | > | |
677 | > | velA[0] += rma * dx; |
678 | > | velA[1] += rma * dy; |
679 | > | velA[2] += rma * dz; |
680 | > | |
681 | > | atoms[a]->setVel(velA); |
682 | > | |
683 | > | velB[0] -= rmb * dx; |
684 | > | velB[1] -= rmb * dy; |
685 | > | velB[2] -= rmb * dz; |
686 | > | |
687 | > | atoms[b]->setVel(velB); |
688 | > | |
689 | > | moving[a] = 1; |
690 | > | moving[b] = 1; |
691 | > | done = 0; |
692 | > | } |
693 | } | |
694 | } | |
695 | ||
696 | < | for(i=0; i<nAtoms; i++){ |
696 | > | for (i = 0; i < nAtoms; i++){ |
697 | moved[i] = moving[i]; | |
698 | moving[i] = 0; | |
699 | } | |
700 | < | |
700 | > | |
701 | iteration++; | |
702 | } | |
681 | – | |
682 | – | if( !done ){ |
703 | ||
704 | < | |
705 | < | sprintf( painCave.errMsg, |
706 | < | "Constraint failure in constrainB, too many iterations: %d\n", |
707 | < | iteration ); |
704 | > | if (!done){ |
705 | > | sprintf(painCave.errMsg, |
706 | > | "Constraint failure in constrainB, too many iterations: %d\n", |
707 | > | iteration); |
708 | painCave.isFatal = 1; | |
709 | simError(); | |
710 | < | } |
691 | < | |
710 | > | } |
711 | } | |
712 | + | */ |
713 | + | template<typename T> void Integrator<T>::rotationPropagation |
714 | + | ( StuntDouble* sd, double ji[3] ){ |
715 | ||
716 | < | void Integrator::rotate( int axes1, int axes2, double angle, double ji[3], |
717 | < | double A[3][3] ){ |
716 | > | double angle; |
717 | > | double A[3][3], I[3][3]; |
718 | > | int i, j, k; |
719 | ||
720 | < | int i,j,k; |
720 | > | // use the angular velocities to propagate the rotation matrix a |
721 | > | // full time step |
722 | > | |
723 | > | sd->getA(A); |
724 | > | sd->getI(I); |
725 | > | |
726 | > | if (sd->isLinear()) { |
727 | > | i = sd->linearAxis(); |
728 | > | j = (i+1)%3; |
729 | > | k = (i+2)%3; |
730 | > | |
731 | > | angle = dt2 * ji[j] / I[j][j]; |
732 | > | this->rotate( k, i, angle, ji, A ); |
733 | > | |
734 | > | angle = dt * ji[k] / I[k][k]; |
735 | > | this->rotate( i, j, angle, ji, A); |
736 | > | |
737 | > | angle = dt2 * ji[j] / I[j][j]; |
738 | > | this->rotate( k, i, angle, ji, A ); |
739 | > | |
740 | > | } else { |
741 | > | // rotate about the x-axis |
742 | > | angle = dt2 * ji[0] / I[0][0]; |
743 | > | this->rotate( 1, 2, angle, ji, A ); |
744 | > | |
745 | > | // rotate about the y-axis |
746 | > | angle = dt2 * ji[1] / I[1][1]; |
747 | > | this->rotate( 2, 0, angle, ji, A ); |
748 | > | |
749 | > | // rotate about the z-axis |
750 | > | angle = dt * ji[2] / I[2][2]; |
751 | > | sd->addZangle(angle); |
752 | > | this->rotate( 0, 1, angle, ji, A); |
753 | > | |
754 | > | // rotate about the y-axis |
755 | > | angle = dt2 * ji[1] / I[1][1]; |
756 | > | this->rotate( 2, 0, angle, ji, A ); |
757 | > | |
758 | > | // rotate about the x-axis |
759 | > | angle = dt2 * ji[0] / I[0][0]; |
760 | > | this->rotate( 1, 2, angle, ji, A ); |
761 | > | |
762 | > | } |
763 | > | sd->setA( A ); |
764 | > | } |
765 | > | |
766 | > | template<typename T> void Integrator<T>::rotate(int axes1, int axes2, |
767 | > | double angle, double ji[3], |
768 | > | double A[3][3]){ |
769 | > | int i, j, k; |
770 | double sinAngle; | |
771 | double cosAngle; | |
772 | double angleSqr; | |
# | Line 706 | Line 778 | void Integrator::rotate( int axes1, int axes2, double | |
778 | ||
779 | // initialize the tempA | |
780 | ||
781 | < | for(i=0; i<3; i++){ |
782 | < | for(j=0; j<3; j++){ |
781 | > | for (i = 0; i < 3; i++){ |
782 | > | for (j = 0; j < 3; j++){ |
783 | tempA[j][i] = A[i][j]; | |
784 | } | |
785 | } | |
786 | ||
787 | // initialize the tempJ | |
788 | ||
789 | < | for( i=0; i<3; i++) tempJ[i] = ji[i]; |
790 | < | |
789 | > | for (i = 0; i < 3; i++) |
790 | > | tempJ[i] = ji[i]; |
791 | > | |
792 | // initalize rot as a unit matrix | |
793 | ||
794 | rot[0][0] = 1.0; | |
# | Line 725 | Line 798 | void Integrator::rotate( int axes1, int axes2, double | |
798 | rot[1][0] = 0.0; | |
799 | rot[1][1] = 1.0; | |
800 | rot[1][2] = 0.0; | |
801 | < | |
801 | > | |
802 | rot[2][0] = 0.0; | |
803 | rot[2][1] = 0.0; | |
804 | rot[2][2] = 1.0; | |
805 | < | |
805 | > | |
806 | // use a small angle aproximation for sin and cosine | |
807 | ||
808 | < | angleSqr = angle * angle; |
808 | > | angleSqr = angle * angle; |
809 | angleSqrOver4 = angleSqr / 4.0; | |
810 | top = 1.0 - angleSqrOver4; | |
811 | bottom = 1.0 + angleSqrOver4; | |
# | Line 745 | Line 818 | void Integrator::rotate( int axes1, int axes2, double | |
818 | ||
819 | rot[axes1][axes2] = sinAngle; | |
820 | rot[axes2][axes1] = -sinAngle; | |
821 | < | |
821 | > | |
822 | // rotate the momentum acoording to: ji[] = rot[][] * ji[] | |
823 | < | |
824 | < | for(i=0; i<3; i++){ |
823 | > | |
824 | > | for (i = 0; i < 3; i++){ |
825 | ji[i] = 0.0; | |
826 | < | for(k=0; k<3; k++){ |
826 | > | for (k = 0; k < 3; k++){ |
827 | ji[i] += rot[i][k] * tempJ[k]; | |
828 | } | |
829 | } | |
830 | ||
831 | < | // rotate the Rotation matrix acording to: |
831 | > | // rotate the Rotation matrix acording to: |
832 | // A[][] = A[][] * transpose(rot[][]) | |
833 | ||
834 | ||
# | Line 763 | Line 836 | void Integrator::rotate( int axes1, int axes2, double | |
836 | // calculation as: | |
837 | // transpose(A[][]) = transpose(A[][]) * transpose(rot[][]) | |
838 | ||
839 | < | for(i=0; i<3; i++){ |
840 | < | for(j=0; j<3; j++){ |
839 | > | for (i = 0; i < 3; i++){ |
840 | > | for (j = 0; j < 3; j++){ |
841 | A[j][i] = 0.0; | |
842 | < | for(k=0; k<3; k++){ |
843 | < | A[j][i] += tempA[i][k] * rot[j][k]; |
842 | > | for (k = 0; k < 3; k++){ |
843 | > | A[j][i] += tempA[i][k] * rot[j][k]; |
844 | } | |
845 | } | |
846 | } | |
847 | } | |
848 | + | |
849 | + | template<typename T> void Integrator<T>::calcForce(int calcPot, int calcStress){ |
850 | + | myFF->doForces(calcPot, calcStress); |
851 | + | } |
852 | + | |
853 | + | template<typename T> void Integrator<T>::thermalize(){ |
854 | + | tStats->velocitize(); |
855 | + | } |
856 | + | |
857 | + | template<typename T> double Integrator<T>::getConservedQuantity(void){ |
858 | + | return tStats->getTotalE(); |
859 | + | } |
860 | + | template<typename T> string Integrator<T>::getAdditionalParameters(void){ |
861 | + | //By default, return a null string |
862 | + | //The reason we use string instead of char* is that if we use char*, we will |
863 | + | //return a pointer point to local variable which might cause problem |
864 | + | return string(); |
865 | + | } |
– | Removed lines |
+ | Added lines |
< | Changed lines |
> | Changed lines |