ViewVC Help

View File | Revision Log | Show Annotations | View Changeset | Root Listing

root/OpenMD/branches/development/src/rnemd/RNEMD.cpp

Comparing:
trunk/src/integrators/RNEMD.cpp (file contents), Revision 1330 by skuang, Thu Mar 19 21:03:36 2009 UTC vs.
branches/development/src/rnemd/RNEMD.cpp (file contents), Revision 1731 by gezelter, Thu May 31 12:25:30 2012 UTC

#	Line 6 | Line 6
6		* redistribute this software in source and binary code form, provided
7		* that the following conditions are met:
8		*
9	<	* 1. Acknowledgement of the program authors must be made in any
10	<	*    publication of scientific results based in part on use of the
11	<	*    program.  An acceptable form of acknowledgement is citation of
12	<	*    the article in which the program was described (Matthew
13	<	*    A. Meineke, Charles F. Vardeman II, Teng Lin, Christopher
14	<	*    J. Fennell and J. Daniel Gezelter, "OOPSE: An Object-Oriented
15	<	*    Parallel Simulation Engine for Molecular Dynamics,"
16	<	*    J. Comput. Chem. 26, pp. 252-271 (2005))
17	<	*
18	<	* 2. Redistributions of source code must retain the above copyright
9	>	* 1. Redistributions of source code must retain the above copyright
10		*    notice, this list of conditions and the following disclaimer.
11		*
12	<	* 3. Redistributions in binary form must reproduce the above copyright
12	>	* 2. Redistributions in binary form must reproduce the above copyright
13		*    notice, this list of conditions and the following disclaimer in the
14		*    documentation and/or other materials provided with the
15		*    distribution.
#	Line 37 | Line 28
28		* arising out of the use of or inability to use software, even if the
29		* University of Notre Dame has been advised of the possibility of
30		* such damages.
31	+	*
32	+	* SUPPORT OPEN SCIENCE!  If you use OpenMD or its source code in your
33	+	* research, please cite the appropriate papers when you publish your
34	+	* work.  Good starting points are:
35	+	*
36	+	* [1]  Meineke, et al., J. Comp. Chem. 26, 252-271 (2005).
37	+	* [2]  Fennell & Gezelter, J. Chem. Phys. 124, 234104 (2006).
38	+	* [3]  Sun, Lin & Gezelter, J. Chem. Phys. 128, 24107 (2008).
39	+	* [4]  Vardeman & Gezelter, in progress (2009).
40		*/
41
42	<	#include "integrators/RNEMD.hpp"
42	>	#include <cmath>
43	>	#include "rnemd/RNEMD.hpp"
44	>	#include "math/Vector3.hpp"
45	>	#include "math/Vector.hpp"
46		#include "math/SquareMatrix3.hpp"
47	+	#include "math/Polynomial.hpp"
48		#include "primitives/Molecule.hpp"
49		#include "primitives/StuntDouble.hpp"
50	+	#include "utils/PhysicalConstants.hpp"
51	+	#include "utils/Tuple.hpp"
52
53		#ifndef IS_MPI
54		#include "math/SeqRandNumGen.hpp"
55		#else
56		#include "math/ParallelRandNumGen.hpp"
57	+	#include <mpi.h>
58		#endif
59
60	<	/* Remove me after testing*/
54	<	/*
55	<	#include <cstdio>
56	<	#include <iostream>
57	<	*/
58	<	/*End remove me*/
60	>	#define HONKING_LARGE_VALUE 1.0e10
61
62	<	namespace oopse {
62	>	using namespace std;
63	>	namespace OpenMD {
64
65	<	RNEMD::RNEMD(SimInfo* info) : info_(info) {
66	<
65	>	RNEMD::RNEMD(SimInfo* info) : info_(info), evaluator_(info), seleMan_(info),
66	>	usePeriodicBoundaryConditions_(info->getSimParams()->getUsePeriodicBoundaryConditions()) {
67	>
68	>	failTrialCount_ = 0;
69	>	failRootCount_ = 0;
70	>
71		int seedValue;
72		Globals * simParams = info->getSimParams();
73	+	RNEMDParameters* rnemdParams = simParams->getRNEMDParameters();
74
75	<	stringToEnumMap_["Kinetic"] = rnemdKinetic;
75	>	stringToEnumMap_["KineticSwap"] = rnemdKineticSwap;
76	>	stringToEnumMap_["KineticScale"] = rnemdKineticScale;
77	>	stringToEnumMap_["KineticScaleVAM"] = rnemdKineticScaleVAM;
78	>	stringToEnumMap_["KineticScaleAM"] = rnemdKineticScaleAM;
79	>	stringToEnumMap_["PxScale"] = rnemdPxScale;
80	>	stringToEnumMap_["PyScale"] = rnemdPyScale;
81	>	stringToEnumMap_["PzScale"] = rnemdPzScale;
82		stringToEnumMap_["Px"] = rnemdPx;
83		stringToEnumMap_["Py"] = rnemdPy;
84		stringToEnumMap_["Pz"] = rnemdPz;
85	+	stringToEnumMap_["ShiftScaleV"] = rnemdShiftScaleV;
86	+	stringToEnumMap_["ShiftScaleVAM"] = rnemdShiftScaleVAM;
87		stringToEnumMap_["Unknown"] = rnemdUnknown;
88
89	<	const std::string st = simParams->getRNEMD_swapType();
89	>	runTime_ = simParams->getRunTime();
90	>	statusTime_ = simParams->getStatusTime();
91
92	<	std::map<std::string, RNEMDTypeEnum>::iterator i;
93	<	i = stringToEnumMap_.find(st);
94	<	rnemdType_  = (i == stringToEnumMap_.end()) ? RNEMD::rnemdUnknown : i->second;
92	>	rnemdObjectSelection_ = rnemdParams->getObjectSelection();
93	>	evaluator_.loadScriptString(rnemdObjectSelection_);
94	>	seleMan_.setSelectionSet(evaluator_.evaluate());
95
96	+	// do some sanity checking
97
98	<	set_RNEMD_swapTime(simParams->getRNEMD_swapTime());
99	<	set_RNEMD_nBins(simParams->getRNEMD_nBins());
100	<	exchangeSum_ = 0.0;
98	>	int selectionCount = seleMan_.getSelectionCount();
99	>	int nIntegrable = info->getNGlobalIntegrableObjects();
100	>
101	>	if (selectionCount > nIntegrable) {
102	>	sprintf(painCave.errMsg,
103	>	"RNEMD: The current RNEMD_objectSelection,\n"
104	>	"\t\t%s\n"
105	>	"\thas resulted in %d selected objects.  However,\n"
106	>	"\tthe total number of integrable objects in the system\n"
107	>	"\tis only %d.  This is almost certainly not what you want\n"
108	>	"\tto do.  A likely cause of this is forgetting the _RB_0\n"
109	>	"\tselector in the selection script!\n",
110	>	rnemdObjectSelection_.c_str(),
111	>	selectionCount, nIntegrable);
112	>	painCave.isFatal = 0;
113	>	painCave.severity = OPENMD_WARNING;
114	>	simError();
115	>	}
116
117	+	const string st = rnemdParams->getExchangeType();
118	+
119	+	map<string, RNEMDTypeEnum>::iterator i;
120	+	i = stringToEnumMap_.find(st);
121	+	rnemdType_ = (i == stringToEnumMap_.end()) ? RNEMD::rnemdUnknown : i->second;
122	+	if (rnemdType_ == rnemdUnknown) {
123	+	sprintf(painCave.errMsg,
124	+	"RNEMD: The current RNEMD_exchangeType,\n"
125	+	"\t\t%s\n"
126	+	"\tis not one of the recognized exchange types.\n",
127	+	st.c_str());
128	+	painCave.isFatal = 1;
129	+	painCave.severity = OPENMD_ERROR;
130	+	simError();
131	+	}
132	+
133	+	outputTemp_ = false;
134	+	if (rnemdParams->haveOutputTemperature()) {
135	+	outputTemp_ = rnemdParams->getOutputTemperature();
136	+	} else if ((rnemdType_ == rnemdKineticSwap) ||
137	+	(rnemdType_ == rnemdKineticScale) ||
138	+	(rnemdType_ == rnemdKineticScaleVAM) ||
139	+	(rnemdType_ == rnemdKineticScaleAM)) {
140	+	outputTemp_ = true;
141	+	}
142	+	outputVx_ = false;
143	+	if (rnemdParams->haveOutputVx()) {
144	+	outputVx_ = rnemdParams->getOutputVx();
145	+	} else if ((rnemdType_ == rnemdPx) || (rnemdType_ == rnemdPxScale)) {
146	+	outputVx_ = true;
147	+	}
148	+	outputVy_ = false;
149	+	if (rnemdParams->haveOutputVy()) {
150	+	outputVy_ = rnemdParams->getOutputVy();
151	+	} else if ((rnemdType_ == rnemdPy) || (rnemdType_ == rnemdPyScale)) {
152	+	outputVy_ = true;
153	+	}
154	+	output3DTemp_ = false;
155	+	if (rnemdParams->haveOutputXyzTemperature()) {
156	+	output3DTemp_ = rnemdParams->getOutputXyzTemperature();
157	+	}
158	+	outputRotTemp_ = false;
159	+	if (rnemdParams->haveOutputRotTemperature()) {
160	+	outputRotTemp_ = rnemdParams->getOutputRotTemperature();
161	+	}
162	+	// James put this in.
163	+	outputDen_ = false;
164	+	if (rnemdParams->haveOutputDen()) {
165	+	outputDen_ = rnemdParams->getOutputDen();
166	+	}
167	+	outputAh_ = false;
168	+	if (rnemdParams->haveOutputAh()) {
169	+	outputAh_ = rnemdParams->getOutputAh();
170	+	}
171	+	outputVz_ = false;
172	+	if (rnemdParams->haveOutputVz()) {
173	+	outputVz_ = rnemdParams->getOutputVz();
174	+	} else if ((rnemdType_ == rnemdPz) || (rnemdType_ == rnemdPzScale)) {
175	+	outputVz_ = true;
176	+	}
177	+
178	+
179	+	#ifdef IS_MPI
180	+	if (worldRank == 0) {
181	+	#endif
182	+
183	+	//may have rnemdWriter separately
184	+	string rnemdFileName;
185	+
186	+	if (outputTemp_) {
187	+	rnemdFileName = "temperature.log";
188	+	tempLog_.open(rnemdFileName.c_str());
189	+	}
190	+	if (outputVx_) {
191	+	rnemdFileName = "velocityX.log";
192	+	vxzLog_.open(rnemdFileName.c_str());
193	+	}
194	+	if (outputVy_) {
195	+	rnemdFileName = "velocityY.log";
196	+	vyzLog_.open(rnemdFileName.c_str());
197	+	}
198	+
199	+	if (output3DTemp_) {
200	+	rnemdFileName = "temperatureX.log";
201	+	xTempLog_.open(rnemdFileName.c_str());
202	+	rnemdFileName = "temperatureY.log";
203	+	yTempLog_.open(rnemdFileName.c_str());
204	+	rnemdFileName = "temperatureZ.log";
205	+	zTempLog_.open(rnemdFileName.c_str());
206	+	}
207	+	if (outputRotTemp_) {
208	+	rnemdFileName = "temperatureR.log";
209	+	rotTempLog_.open(rnemdFileName.c_str());
210	+	}
211	+
212	+	//James put this in
213	+	if (outputDen_) {
214	+	rnemdFileName = "Density.log";
215	+	denLog_.open(rnemdFileName.c_str());
216	+	}
217	+	if (outputAh_) {
218	+	rnemdFileName = "Ah.log";
219	+	AhLog_.open(rnemdFileName.c_str());
220	+	}
221	+	if (outputVz_) {
222	+	rnemdFileName = "velocityZ.log";
223	+	vzzLog_.open(rnemdFileName.c_str());
224	+	}
225	+	logFrameCount_ = 0;
226	+	#ifdef IS_MPI
227	+	}
228	+	#endif
229	+
230	+	set_RNEMD_exchange_time(rnemdParams->getExchangeTime());
231	+	set_RNEMD_nBins(rnemdParams->getNbins());
232	+	midBin_ = nBins_ / 2;
233	+	if (rnemdParams->haveBinShift()) {
234	+	if (rnemdParams->getBinShift()) {
235	+	zShift_ = 0.5 / (RealType)(nBins_);
236	+	} else {
237	+	zShift_ = 0.0;
238	+	}
239	+	} else {
240	+	zShift_ = 0.0;
241	+	}
242	+	//cerr << "I shift slabs by " << zShift_ << " Lz\n";
243	+	//shift slabs by half slab width, maybe useful in heterogeneous systems
244	+	//set to 0.0 if not using it; N/A in status output yet
245	+	if (rnemdParams->haveLogWidth()) {
246	+	set_RNEMD_logWidth(rnemdParams->getLogWidth());
247	+	/*arbitary rnemdLogWidth_, no checking;
248	+	if (rnemdLogWidth_ != nBins_ && rnemdLogWidth_ != midBin_ + 1) {
249	+	cerr << "WARNING! RNEMD_logWidth has abnormal value!\n";
250	+	cerr << "Automaically set back to default.\n";
251	+	rnemdLogWidth_ = nBins_;
252	+	}*/
253	+	} else {
254	+	set_RNEMD_logWidth(nBins_);
255	+	}
256	+	tempHist_.resize(rnemdLogWidth_, 0.0);
257	+	tempCount_.resize(rnemdLogWidth_, 0);
258	+	pxzHist_.resize(rnemdLogWidth_, 0.0);
259	+	//vxzCount_.resize(rnemdLogWidth_, 0);
260	+	pyzHist_.resize(rnemdLogWidth_, 0.0);
261	+	//vyzCount_.resize(rnemdLogWidth_, 0);
262	+
263	+	mHist_.resize(rnemdLogWidth_, 0.0);
264	+	xTempHist_.resize(rnemdLogWidth_, 0.0);
265	+	yTempHist_.resize(rnemdLogWidth_, 0.0);
266	+	zTempHist_.resize(rnemdLogWidth_, 0.0);
267	+	xyzTempCount_.resize(rnemdLogWidth_, 0);
268	+	rotTempHist_.resize(rnemdLogWidth_, 0.0);
269	+	rotTempCount_.resize(rnemdLogWidth_, 0);
270	+	// James put this in
271	+	DenHist_.resize(rnemdLogWidth_, 0.0);
272	+	pzzHist_.resize(rnemdLogWidth_, 0.0);
273	+
274	+	set_RNEMD_exchange_total(0.0);
275	+	if (rnemdParams->haveTargetFlux()) {
276	+	set_RNEMD_target_flux(rnemdParams->getTargetFlux());
277	+	} else {
278	+	set_RNEMD_target_flux(0.0);
279	+	}
280	+	if (rnemdParams->haveTargetJzKE()) {
281	+	set_RNEMD_target_JzKE(rnemdParams->getTargetJzKE());
282	+	} else {
283	+	set_RNEMD_target_JzKE(0.0);
284	+	}
285	+	if (rnemdParams->haveTargetJzpx()) {
286	+	set_RNEMD_target_jzpx(rnemdParams->getTargetJzpx());
287	+	} else {
288	+	set_RNEMD_target_jzpx(0.0);
289	+	}
290	+	jzp_.x() = targetJzpx_;
291	+	njzp_.x() = -targetJzpx_;
292	+	if (rnemdParams->haveTargetJzpy()) {
293	+	set_RNEMD_target_jzpy(rnemdParams->getTargetJzpy());
294	+	} else {
295	+	set_RNEMD_target_jzpy(0.0);
296	+	}
297	+	jzp_.y() = targetJzpy_;
298	+	njzp_.y() = -targetJzpy_;
299	+	if (rnemdParams->haveTargetJzpz()) {
300	+	set_RNEMD_target_jzpz(rnemdParams->getTargetJzpz());
301	+	} else {
302	+	set_RNEMD_target_jzpz(0.0);
303	+	}
304	+	jzp_.z() = targetJzpz_;
305	+	njzp_.z() = -targetJzpz_;
306	+
307		#ifndef IS_MPI
308		if (simParams->haveSeed()) {
309		seedValue = simParams->getSeed();
#	Line 100 | Line 323 | namespace oopse {
323
324		RNEMD::~RNEMD() {
325		delete randNumGen_;
326	+
327	+	#ifdef IS_MPI
328	+	if (worldRank == 0) {
329	+	#endif
330	+
331	+	sprintf(painCave.errMsg,
332	+	"RNEMD: total failed trials: %d\n",
333	+	failTrialCount_);
334	+	painCave.isFatal = 0;
335	+	painCave.severity = OPENMD_INFO;
336	+	simError();
337	+
338	+	if (outputTemp_) tempLog_.close();
339	+	if (outputVx_)   vxzLog_.close();
340	+	if (outputVy_)   vyzLog_.close();
341	+
342	+	if (rnemdType_ == rnemdKineticScale || rnemdType_ == rnemdPxScale ||
343	+	rnemdType_ == rnemdPyScale) {
344	+	sprintf(painCave.errMsg,
345	+	"RNEMD: total root-checking warnings: %d\n",
346	+	failRootCount_);
347	+	painCave.isFatal = 0;
348	+	painCave.severity = OPENMD_INFO;
349	+	simError();
350	+	}
351	+	if (output3DTemp_) {
352	+	xTempLog_.close();
353	+	yTempLog_.close();
354	+	zTempLog_.close();
355	+	}
356	+	if (outputRotTemp_) rotTempLog_.close();
357	+	// James put this in
358	+	if (outputDen_) denLog_.close();
359	+	if (outputAh_)  AhLog_.close();
360	+	if (outputVz_)  vzzLog_.close();
361	+
362	+	#ifdef IS_MPI
363	+	}
364	+	#endif
365		}
366
367		void RNEMD::doSwap() {
368	<	std::cerr << "in RNEMD!\n";
369	<	std::cerr << "nBins = " << nBins_ << "\n";
370	<	std::cerr << "swapTime = " << swapTime_ << "\n";
371	<	std::cerr << "exchangeSum = " << exchangeSum_ << "\n";
372	<	std::cerr << "swapType = " << rnemdType_ << "\n";
373	<	}
368	>
369	>	Snapshot* currentSnap_ = info_->getSnapshotManager()->getCurrentSnapshot();
370	>	Mat3x3d hmat = currentSnap_->getHmat();
371	>
372	>	seleMan_.setSelectionSet(evaluator_.evaluate());
373	>
374	>	int selei;
375	>	StuntDouble* sd;
376	>	int idx;
377	>
378	>	RealType min_val;
379	>	bool min_found = false;
380	>	StuntDouble* min_sd;
381	>
382	>	RealType max_val;
383	>	bool max_found = false;
384	>	StuntDouble* max_sd;
385	>
386	>	for (sd = seleMan_.beginSelected(selei); sd != NULL;
387	>	sd = seleMan_.nextSelected(selei)) {
388	>
389	>	idx = sd->getLocalIndex();
390	>
391	>	Vector3d pos = sd->getPos();
392	>
393	>	// wrap the stuntdouble's position back into the box:
394	>
395	>	if (usePeriodicBoundaryConditions_)
396	>	currentSnap_->wrapVector(pos);
397	>
398	>	// which bin is this stuntdouble in?
399	>	// wrapped positions are in the range [-0.5*hmat(2,2), +0.5*hmat(2,2)]
400	>
401	>	int binNo = int(nBins_ * (pos.z() / hmat(2,2) + zShift_ + 0.5)) % nBins_;
402	>
403	>
404	>	// if we're in bin 0 or the middleBin
405	>	if (binNo == 0 || binNo == midBin_) {
406	>
407	>	RealType mass = sd->getMass();
408	>	Vector3d vel = sd->getVel();
409	>	RealType value;
410	>
411	>	switch(rnemdType_) {
412	>	case rnemdKineticSwap :
413	>
414	>	value = mass * vel.lengthSquare();
415	>
416	>	if (sd->isDirectional()) {
417	>	Vector3d angMom = sd->getJ();
418	>	Mat3x3d I = sd->getI();
419	>
420	>	if (sd->isLinear()) {
421	>	int i = sd->linearAxis();
422	>	int j = (i + 1) % 3;
423	>	int k = (i + 2) % 3;
424	>	value += angMom[j] * angMom[j] / I(j, j) +
425	>	angMom[k] * angMom[k] / I(k, k);
426	>	} else {
427	>	value += angMom[0]*angMom[0]/I(0, 0)
428	>	+ angMom[1]*angMom[1]/I(1, 1)
429	>	+ angMom[2]*angMom[2]/I(2, 2);
430	>	}
431	>	} //angular momenta exchange enabled
432	>	//energyConvert temporarily disabled
433	>	//make exchangeSum_ comparable between swap & scale
434	>	//value = value * 0.5 / PhysicalConstants::energyConvert;
435	>	value *= 0.5;
436	>	break;
437	>	case rnemdPx :
438	>	value = mass * vel[0];
439	>	break;
440	>	case rnemdPy :
441	>	value = mass * vel[1];
442	>	break;
443	>	case rnemdPz :
444	>	value = mass * vel[2];
445	>	break;
446	>	default :
447	>	break;
448	>	}
449	>
450	>	if (binNo == 0) {
451	>	if (!min_found) {
452	>	min_val = value;
453	>	min_sd = sd;
454	>	min_found = true;
455	>	} else {
456	>	if (min_val > value) {
457	>	min_val = value;
458	>	min_sd = sd;
459	>	}
460	>	}
461	>	} else { //midBin_
462	>	if (!max_found) {
463	>	max_val = value;
464	>	max_sd = sd;
465	>	max_found = true;
466	>	} else {
467	>	if (max_val < value) {
468	>	max_val = value;
469	>	max_sd = sd;
470	>	}
471	>	}
472	>	}
473	>	}
474	>	}
475	>
476	>	#ifdef IS_MPI
477	>	int nProc, worldRank;
478	>
479	>	nProc = MPI::COMM_WORLD.Get_size();
480	>	worldRank = MPI::COMM_WORLD.Get_rank();
481	>
482	>	bool my_min_found = min_found;
483	>	bool my_max_found = max_found;
484	>
485	>	// Even if we didn't find a minimum, did someone else?
486	>	MPI::COMM_WORLD.Allreduce(&my_min_found, &min_found, 1, MPI::BOOL, MPI::LOR);
487	>	// Even if we didn't find a maximum, did someone else?
488	>	MPI::COMM_WORLD.Allreduce(&my_max_found, &max_found, 1, MPI::BOOL, MPI::LOR);
489	>	#endif
490	>
491	>	if (max_found && min_found) {
492	>
493	>	#ifdef IS_MPI
494	>	struct {
495	>	RealType val;
496	>	int rank;
497	>	} max_vals, min_vals;
498	>
499	>	if (my_min_found) {
500	>	min_vals.val = min_val;
501	>	} else {
502	>	min_vals.val = HONKING_LARGE_VALUE;
503	>	}
504	>	min_vals.rank = worldRank;
505	>
506	>	// Who had the minimum?
507	>	MPI::COMM_WORLD.Allreduce(&min_vals, &min_vals,
508	>	1, MPI::REALTYPE_INT, MPI::MINLOC);
509	>	min_val = min_vals.val;
510	>
511	>	if (my_max_found) {
512	>	max_vals.val = max_val;
513	>	} else {
514	>	max_vals.val = -HONKING_LARGE_VALUE;
515	>	}
516	>	max_vals.rank = worldRank;
517	>
518	>	// Who had the maximum?
519	>	MPI::COMM_WORLD.Allreduce(&max_vals, &max_vals,
520	>	1, MPI::REALTYPE_INT, MPI::MAXLOC);
521	>	max_val = max_vals.val;
522	>	#endif
523	>
524	>	if (min_val < max_val) {
525	>
526	>	#ifdef IS_MPI
527	>	if (max_vals.rank == worldRank && min_vals.rank == worldRank) {
528	>	// I have both maximum and minimum, so proceed like a single
529	>	// processor version:
530	>	#endif
531	>
532	>	Vector3d min_vel = min_sd->getVel();
533	>	Vector3d max_vel = max_sd->getVel();
534	>	RealType temp_vel;
535	>
536	>	switch(rnemdType_) {
537	>	case rnemdKineticSwap :
538	>	min_sd->setVel(max_vel);
539	>	max_sd->setVel(min_vel);
540	>	if (min_sd->isDirectional() && max_sd->isDirectional()) {
541	>	Vector3d min_angMom = min_sd->getJ();
542	>	Vector3d max_angMom = max_sd->getJ();
543	>	min_sd->setJ(max_angMom);
544	>	max_sd->setJ(min_angMom);
545	>	}//angular momenta exchange enabled
546	>	//assumes same rigid body identity
547	>	break;
548	>	case rnemdPx :
549	>	temp_vel = min_vel.x();
550	>	min_vel.x() = max_vel.x();
551	>	max_vel.x() = temp_vel;
552	>	min_sd->setVel(min_vel);
553	>	max_sd->setVel(max_vel);
554	>	break;
555	>	case rnemdPy :
556	>	temp_vel = min_vel.y();
557	>	min_vel.y() = max_vel.y();
558	>	max_vel.y() = temp_vel;
559	>	min_sd->setVel(min_vel);
560	>	max_sd->setVel(max_vel);
561	>	break;
562	>	case rnemdPz :
563	>	temp_vel = min_vel.z();
564	>	min_vel.z() = max_vel.z();
565	>	max_vel.z() = temp_vel;
566	>	min_sd->setVel(min_vel);
567	>	max_sd->setVel(max_vel);
568	>	break;
569	>	default :
570	>	break;
571	>	}
572	>
573	>	#ifdef IS_MPI
574	>	// the rest of the cases only apply in parallel simulations:
575	>	} else if (max_vals.rank == worldRank) {
576	>	// I had the max, but not the minimum
577	>
578	>	Vector3d min_vel;
579	>	Vector3d max_vel = max_sd->getVel();
580	>	MPI::Status status;
581	>
582	>	// point-to-point swap of the velocity vector
583	>	MPI::COMM_WORLD.Sendrecv(max_vel.getArrayPointer(), 3, MPI::REALTYPE,
584	>	min_vals.rank, 0,
585	>	min_vel.getArrayPointer(), 3, MPI::REALTYPE,
586	>	min_vals.rank, 0, status);
587	>
588	>	switch(rnemdType_) {
589	>	case rnemdKineticSwap :
590	>	max_sd->setVel(min_vel);
591	>	//angular momenta exchange enabled
592	>	if (max_sd->isDirectional()) {
593	>	Vector3d min_angMom;
594	>	Vector3d max_angMom = max_sd->getJ();
595	>
596	>	// point-to-point swap of the angular momentum vector
597	>	MPI::COMM_WORLD.Sendrecv(max_angMom.getArrayPointer(), 3,
598	>	MPI::REALTYPE, min_vals.rank, 1,
599	>	min_angMom.getArrayPointer(), 3,
600	>	MPI::REALTYPE, min_vals.rank, 1,
601	>	status);
602	>
603	>	max_sd->setJ(min_angMom);
604	>	}
605	>	break;
606	>	case rnemdPx :
607	>	max_vel.x() = min_vel.x();
608	>	max_sd->setVel(max_vel);
609	>	break;
610	>	case rnemdPy :
611	>	max_vel.y() = min_vel.y();
612	>	max_sd->setVel(max_vel);
613	>	break;
614	>	case rnemdPz :
615	>	max_vel.z() = min_vel.z();
616	>	max_sd->setVel(max_vel);
617	>	break;
618	>	default :
619	>	break;
620	>	}
621	>	} else if (min_vals.rank == worldRank) {
622	>	// I had the minimum but not the maximum:
623	>
624	>	Vector3d max_vel;
625	>	Vector3d min_vel = min_sd->getVel();
626	>	MPI::Status status;
627	>
628	>	// point-to-point swap of the velocity vector
629	>	MPI::COMM_WORLD.Sendrecv(min_vel.getArrayPointer(), 3, MPI::REALTYPE,
630	>	max_vals.rank, 0,
631	>	max_vel.getArrayPointer(), 3, MPI::REALTYPE,
632	>	max_vals.rank, 0, status);
633	>
634	>	switch(rnemdType_) {
635	>	case rnemdKineticSwap :
636	>	min_sd->setVel(max_vel);
637	>	//angular momenta exchange enabled
638	>	if (min_sd->isDirectional()) {
639	>	Vector3d min_angMom = min_sd->getJ();
640	>	Vector3d max_angMom;
641	>
642	>	// point-to-point swap of the angular momentum vector
643	>	MPI::COMM_WORLD.Sendrecv(min_angMom.getArrayPointer(), 3,
644	>	MPI::REALTYPE, max_vals.rank, 1,
645	>	max_angMom.getArrayPointer(), 3,
646	>	MPI::REALTYPE, max_vals.rank, 1,
647	>	status);
648	>
649	>	min_sd->setJ(max_angMom);
650	>	}
651	>	break;
652	>	case rnemdPx :
653	>	min_vel.x() = max_vel.x();
654	>	min_sd->setVel(min_vel);
655	>	break;
656	>	case rnemdPy :
657	>	min_vel.y() = max_vel.y();
658	>	min_sd->setVel(min_vel);
659	>	break;
660	>	case rnemdPz :
661	>	min_vel.z() = max_vel.z();
662	>	min_sd->setVel(min_vel);
663	>	break;
664	>	default :
665	>	break;
666	>	}
667	>	}
668	>	#endif
669	>	exchangeSum_ += max_val - min_val;
670	>	} else {
671	>	sprintf(painCave.errMsg,
672	>	"RNEMD: exchange NOT performed because min_val > max_val\n");
673	>	painCave.isFatal = 0;
674	>	painCave.severity = OPENMD_INFO;
675	>	simError();
676	>	failTrialCount_++;
677	>	}
678	>	} else {
679	>	sprintf(painCave.errMsg,
680	>	"RNEMD: exchange NOT performed because selected object\n"
681	>	"\tnot present in at least one of the two slabs.\n");
682	>	painCave.isFatal = 0;
683	>	painCave.severity = OPENMD_INFO;
684	>	simError();
685	>	failTrialCount_++;
686	>	}
687	>
688	>	}
689	>
690	>	void RNEMD::doScale() {
691	>
692	>	Snapshot* currentSnap_ = info_->getSnapshotManager()->getCurrentSnapshot();
693	>	Mat3x3d hmat = currentSnap_->getHmat();
694	>
695	>	seleMan_.setSelectionSet(evaluator_.evaluate());
696	>
697	>	int selei;
698	>	StuntDouble* sd;
699	>	int idx;
700	>
701	>	vector<StuntDouble*> hotBin, coldBin;
702	>
703	>	RealType Phx = 0.0;
704	>	RealType Phy = 0.0;
705	>	RealType Phz = 0.0;
706	>	RealType Khx = 0.0;
707	>	RealType Khy = 0.0;
708	>	RealType Khz = 0.0;
709	>	RealType Khw = 0.0;
710	>	RealType Pcx = 0.0;
711	>	RealType Pcy = 0.0;
712	>	RealType Pcz = 0.0;
713	>	RealType Kcx = 0.0;
714	>	RealType Kcy = 0.0;
715	>	RealType Kcz = 0.0;
716	>	RealType Kcw = 0.0;
717	>
718	>	for (sd = seleMan_.beginSelected(selei); sd != NULL;
719	>	sd = seleMan_.nextSelected(selei)) {
720	>
721	>	idx = sd->getLocalIndex();
722	>
723	>	Vector3d pos = sd->getPos();
724	>
725	>	// wrap the stuntdouble's position back into the box:
726	>
727	>	if (usePeriodicBoundaryConditions_)
728	>	currentSnap_->wrapVector(pos);
729	>
730	>	// which bin is this stuntdouble in?
731	>	// wrapped positions are in the range [-0.5*hmat(2,2), +0.5*hmat(2,2)]
732	>
733	>	int binNo = int(nBins_ * (pos.z() / hmat(2,2) + zShift_ + 0.5)) % nBins_;
734	>
735	>	// if we're in bin 0 or the middleBin
736	>	if (binNo == 0 || binNo == midBin_) {
737	>
738	>	RealType mass = sd->getMass();
739	>	Vector3d vel = sd->getVel();
740	>
741	>	if (binNo == 0) {
742	>	hotBin.push_back(sd);
743	>	Phx += mass * vel.x();
744	>	Phy += mass * vel.y();
745	>	Phz += mass * vel.z();
746	>	Khx += mass * vel.x() * vel.x();
747	>	Khy += mass * vel.y() * vel.y();
748	>	Khz += mass * vel.z() * vel.z();
749	>	//if (rnemdType_ == rnemdKineticScaleVAM) {
750	>	if (sd->isDirectional()) {
751	>	Vector3d angMom = sd->getJ();
752	>	Mat3x3d I = sd->getI();
753	>	if (sd->isLinear()) {
754	>	int i = sd->linearAxis();
755	>	int j = (i + 1) % 3;
756	>	int k = (i + 2) % 3;
757	>	Khw += angMom[j] * angMom[j] / I(j, j) +
758	>	angMom[k] * angMom[k] / I(k, k);
759	>	} else {
760	>	Khw += angMom[0]*angMom[0]/I(0, 0)
761	>	+ angMom[1]*angMom[1]/I(1, 1)
762	>	+ angMom[2]*angMom[2]/I(2, 2);
763	>	}
764	>	}
765	>	//}
766	>	} else { //midBin_
767	>	coldBin.push_back(sd);
768	>	Pcx += mass * vel.x();
769	>	Pcy += mass * vel.y();
770	>	Pcz += mass * vel.z();
771	>	Kcx += mass * vel.x() * vel.x();
772	>	Kcy += mass * vel.y() * vel.y();
773	>	Kcz += mass * vel.z() * vel.z();
774	>	//if (rnemdType_ == rnemdKineticScaleVAM) {
775	>	if (sd->isDirectional()) {
776	>	Vector3d angMom = sd->getJ();
777	>	Mat3x3d I = sd->getI();
778	>	if (sd->isLinear()) {
779	>	int i = sd->linearAxis();
780	>	int j = (i + 1) % 3;
781	>	int k = (i + 2) % 3;
782	>	Kcw += angMom[j] * angMom[j] / I(j, j) +
783	>	angMom[k] * angMom[k] / I(k, k);
784	>	} else {
785	>	Kcw += angMom[0]*angMom[0]/I(0, 0)
786	>	+ angMom[1]*angMom[1]/I(1, 1)
787	>	+ angMom[2]*angMom[2]/I(2, 2);
788	>	}
789	>	}
790	>	//}
791	>	}
792	>	}
793	>	}
794	>
795	>	Khx *= 0.5;
796	>	Khy *= 0.5;
797	>	Khz *= 0.5;
798	>	Khw *= 0.5;
799	>	Kcx *= 0.5;
800	>	Kcy *= 0.5;
801	>	Kcz *= 0.5;
802	>	Kcw *= 0.5;
803	>
804	>	// std::cerr << "Khx= " << Khx << "\tKhy= " << Khy << "\tKhz= " << Khz
805	>	//        << "\tKhw= " << Khw << "\tKcx= " << Kcx << "\tKcy= " << Kcy
806	>	//        << "\tKcz= " << Kcz << "\tKcw= " << Kcw << "\n";
807	>	// std::cerr << "Phx= " << Phx << "\tPhy= " << Phy << "\tPhz= " << Phz
808	>	//        << "\tPcx= " << Pcx << "\tPcy= " << Pcy << "\tPcz= " <<Pcz<<"\n";
809	>
810	>	#ifdef IS_MPI
811	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Phx, 1, MPI::REALTYPE, MPI::SUM);
812	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Phy, 1, MPI::REALTYPE, MPI::SUM);
813	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Phz, 1, MPI::REALTYPE, MPI::SUM);
814	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Pcx, 1, MPI::REALTYPE, MPI::SUM);
815	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Pcy, 1, MPI::REALTYPE, MPI::SUM);
816	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Pcz, 1, MPI::REALTYPE, MPI::SUM);
817	>
818	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Khx, 1, MPI::REALTYPE, MPI::SUM);
819	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Khy, 1, MPI::REALTYPE, MPI::SUM);
820	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Khz, 1, MPI::REALTYPE, MPI::SUM);
821	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Khw, 1, MPI::REALTYPE, MPI::SUM);
822	>
823	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Kcx, 1, MPI::REALTYPE, MPI::SUM);
824	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Kcy, 1, MPI::REALTYPE, MPI::SUM);
825	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Kcz, 1, MPI::REALTYPE, MPI::SUM);
826	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Kcw, 1, MPI::REALTYPE, MPI::SUM);
827	>	#endif
828	>
829	>	//solve coldBin coeff's first
830	>	RealType px = Pcx / Phx;
831	>	RealType py = Pcy / Phy;
832	>	RealType pz = Pcz / Phz;
833	>	RealType c, x, y, z;
834	>	bool successfulScale = false;
835	>	if ((rnemdType_ == rnemdKineticScaleVAM) ||
836	>	(rnemdType_ == rnemdKineticScaleAM)) {
837	>	//may need sanity check Khw & Kcw > 0
838	>
839	>	if (rnemdType_ == rnemdKineticScaleVAM) {
840	>	c = 1.0 - targetFlux_ / (Kcx + Kcy + Kcz + Kcw);
841	>	} else {
842	>	c = 1.0 - targetFlux_ / Kcw;
843	>	}
844	>
845	>	if ((c > 0.81) && (c < 1.21)) {//restrict scaling coefficients
846	>	c = sqrt(c);
847	>	std::cerr << "cold slab scaling coefficient: " << c << endl;
848	>	//now convert to hotBin coefficient
849	>	RealType w = 0.0;
850	>	if (rnemdType_ ==  rnemdKineticScaleVAM) {
851	>	x = 1.0 + px * (1.0 - c);
852	>	y = 1.0 + py * (1.0 - c);
853	>	z = 1.0 + pz * (1.0 - c);
854	>	/* more complicated way
855	>	w = 1.0 + (Kcw - Kcw * c * c - (c * c * (Kcx + Kcy + Kcz
856	>	+ Khx * px * px + Khy * py * py + Khz * pz * pz)
857	>	- 2.0 * c * (Khx * px * (1.0 + px) + Khy * py * (1.0 + py)
858	>	+ Khz * pz * (1.0 + pz)) + Khx * px * (2.0 + px)
859	>	+ Khy * py * (2.0 + py) + Khz * pz * (2.0 + pz)
860	>	- Kcx - Kcy - Kcz)) / Khw; the following is simpler
861	>	*/
862	>	if ((fabs(x - 1.0) < 0.1) && (fabs(y - 1.0) < 0.1) &&
863	>	(fabs(z - 1.0) < 0.1)) {
864	>	w = 1.0 + (targetFlux_ + Khx * (1.0 - x * x) + Khy * (1.0 - y * y)
865	>	+ Khz * (1.0 - z * z)) / Khw;
866	>	}//no need to calculate w if x, y or z is out of range
867	>	} else {
868	>	w = 1.0 + targetFlux_ / Khw;
869	>	}
870	>	if ((w > 0.81) && (w < 1.21)) {//restrict scaling coefficients
871	>	//if w is in the right range, so should be x, y, z.
872	>	vector<StuntDouble*>::iterator sdi;
873	>	Vector3d vel;
874	>	for (sdi = coldBin.begin(); sdi != coldBin.end(); sdi++) {
875	>	if (rnemdType_ == rnemdKineticScaleVAM) {
876	>	vel = (*sdi)->getVel() * c;
877	>	//vel.x() *= c;
878	>	//vel.y() *= c;
879	>	//vel.z() *= c;
880	>	(*sdi)->setVel(vel);
881	>	}
882	>	if ((*sdi)->isDirectional()) {
883	>	Vector3d angMom = (*sdi)->getJ() * c;
884	>	//angMom[0] *= c;
885	>	//angMom[1] *= c;
886	>	//angMom[2] *= c;
887	>	(*sdi)->setJ(angMom);
888	>	}
889	>	}
890	>	w = sqrt(w);
891	>	std::cerr << "xh= " << x << "\tyh= " << y << "\tzh= " << z
892	>	<< "\twh= " << w << endl;
893	>	for (sdi = hotBin.begin(); sdi != hotBin.end(); sdi++) {
894	>	if (rnemdType_ == rnemdKineticScaleVAM) {
895	>	vel = (*sdi)->getVel();
896	>	vel.x() *= x;
897	>	vel.y() *= y;
898	>	vel.z() *= z;
899	>	(*sdi)->setVel(vel);
900	>	}
901	>	if ((*sdi)->isDirectional()) {
902	>	Vector3d angMom = (*sdi)->getJ() * w;
903	>	//angMom[0] *= w;
904	>	//angMom[1] *= w;
905	>	//angMom[2] *= w;
906	>	(*sdi)->setJ(angMom);
907	>	}
908	>	}
909	>	successfulScale = true;
910	>	exchangeSum_ += targetFlux_;
911	>	}
912	>	}
913	>	} else {
914	>	RealType a000, a110, c0, a001, a111, b01, b11, c1;
915	>	switch(rnemdType_) {
916	>	case rnemdKineticScale :
917	>	/* used hotBin coeff's & only scale x & y dimensions
918	>	RealType px = Phx / Pcx;
919	>	RealType py = Phy / Pcy;
920	>	a110 = Khy;
921	>	c0 = - Khx - Khy - targetFlux_;
922	>	a000 = Khx;
923	>	a111 = Kcy * py * py;
924	>	b11 = -2.0 * Kcy * py * (1.0 + py);
925	>	c1 = Kcy * py * (2.0 + py) + Kcx * px * ( 2.0 + px) + targetFlux_;
926	>	b01 = -2.0 * Kcx * px * (1.0 + px);
927	>	a001 = Kcx * px * px;
928	>	*/
929	>	//scale all three dimensions, let c_x = c_y
930	>	a000 = Kcx + Kcy;
931	>	a110 = Kcz;
932	>	c0 = targetFlux_ - Kcx - Kcy - Kcz;
933	>	a001 = Khx * px * px + Khy * py * py;
934	>	a111 = Khz * pz * pz;
935	>	b01 = -2.0 * (Khx * px * (1.0 + px) + Khy * py * (1.0 + py));
936	>	b11 = -2.0 * Khz * pz * (1.0 + pz);
937	>	c1 = Khx * px * (2.0 + px) + Khy * py * (2.0 + py)
938	>	+ Khz * pz * (2.0 + pz) - targetFlux_;
939	>	break;
940	>	case rnemdPxScale :
941	>	c = 1 - targetFlux_ / Pcx;
942	>	a000 = Kcy;
943	>	a110 = Kcz;
944	>	c0 = Kcx * c * c - Kcx - Kcy - Kcz;
945	>	a001 = py * py * Khy;
946	>	a111 = pz * pz * Khz;
947	>	b01 = -2.0 * Khy * py * (1.0 + py);
948	>	b11 = -2.0 * Khz * pz * (1.0 + pz);
949	>	c1 = Khy * py * (2.0 + py) + Khz * pz * (2.0 + pz)
950	>	+ Khx * (fastpow(c * px - px - 1.0, 2) - 1.0);
951	>	break;
952	>	case rnemdPyScale :
953	>	c = 1 - targetFlux_ / Pcy;
954	>	a000 = Kcx;
955	>	a110 = Kcz;
956	>	c0 = Kcy * c * c - Kcx - Kcy - Kcz;
957	>	a001 = px * px * Khx;
958	>	a111 = pz * pz * Khz;
959	>	b01 = -2.0 * Khx * px * (1.0 + px);
960	>	b11 = -2.0 * Khz * pz * (1.0 + pz);
961	>	c1 = Khx * px * (2.0 + px) + Khz * pz * (2.0 + pz)
962	>	+ Khy * (fastpow(c * py - py - 1.0, 2) - 1.0);
963	>	break;
964	>	case rnemdPzScale ://we don't really do this, do we?
965	>	c = 1 - targetFlux_ / Pcz;
966	>	a000 = Kcx;
967	>	a110 = Kcy;
968	>	c0 = Kcz * c * c - Kcx - Kcy - Kcz;
969	>	a001 = px * px * Khx;
970	>	a111 = py * py * Khy;
971	>	b01 = -2.0 * Khx * px * (1.0 + px);
972	>	b11 = -2.0 * Khy * py * (1.0 + py);
973	>	c1 = Khx * px * (2.0 + px) + Khy * py * (2.0 + py)
974	>	+ Khz * (fastpow(c * pz - pz - 1.0, 2) - 1.0);
975	>	break;
976	>	default :
977	>	break;
978	>	}
979	>
980	>	RealType v1 = a000 * a111 - a001 * a110;
981	>	RealType v2 = a000 * b01;
982	>	RealType v3 = a000 * b11;
983	>	RealType v4 = a000 * c1 - a001 * c0;
984	>	RealType v8 = a110 * b01;
985	>	RealType v10 = - b01 * c0;
986	>
987	>	RealType u0 = v2 * v10 - v4 * v4;
988	>	RealType u1 = -2.0 * v3 * v4;
989	>	RealType u2 = -v2 * v8 - v3 * v3 - 2.0 * v1 * v4;
990	>	RealType u3 = -2.0 * v1 * v3;
991	>	RealType u4 = - v1 * v1;
992	>	//rescale coefficients
993	>	RealType maxAbs = fabs(u0);
994	>	if (maxAbs < fabs(u1)) maxAbs = fabs(u1);
995	>	if (maxAbs < fabs(u2)) maxAbs = fabs(u2);
996	>	if (maxAbs < fabs(u3)) maxAbs = fabs(u3);
997	>	if (maxAbs < fabs(u4)) maxAbs = fabs(u4);
998	>	u0 /= maxAbs;
999	>	u1 /= maxAbs;
1000	>	u2 /= maxAbs;
1001	>	u3 /= maxAbs;
1002	>	u4 /= maxAbs;
1003	>	//max_element(start, end) is also available.
1004	>	Polynomial<RealType> poly; //same as DoublePolynomial poly;
1005	>	poly.setCoefficient(4, u4);
1006	>	poly.setCoefficient(3, u3);
1007	>	poly.setCoefficient(2, u2);
1008	>	poly.setCoefficient(1, u1);
1009	>	poly.setCoefficient(0, u0);
1010	>	vector<RealType> realRoots = poly.FindRealRoots();
1011	>
1012	>	vector<RealType>::iterator ri;
1013	>	RealType r1, r2, alpha0;
1014	>	vector<pair<RealType,RealType> > rps;
1015	>	for (ri = realRoots.begin(); ri !=realRoots.end(); ri++) {
1016	>	r2 = *ri;
1017	>	//check if FindRealRoots() give the right answer
1018	>	if ( fabs(u0 + r2 * (u1 + r2 * (u2 + r2 * (u3 + r2 * u4)))) > 1e-6 ) {
1019	>	sprintf(painCave.errMsg,
1020	>	"RNEMD Warning: polynomial solve seems to have an error!");
1021	>	painCave.isFatal = 0;
1022	>	simError();
1023	>	failRootCount_++;
1024	>	}
1025	>	//might not be useful w/o rescaling coefficients
1026	>	alpha0 = -c0 - a110 * r2 * r2;
1027	>	if (alpha0 >= 0.0) {
1028	>	r1 = sqrt(alpha0 / a000);
1029	>	if (fabs(c1 + r1 * (b01 + r1 * a001) + r2 * (b11 + r2 * a111))
1030	>	< 1e-6)
1031	>	{ rps.push_back(make_pair(r1, r2)); }
1032	>	if (r1 > 1e-6) { //r1 non-negative
1033	>	r1 = -r1;
1034	>	if (fabs(c1 + r1 * (b01 + r1 * a001) + r2 * (b11 + r2 * a111))
1035	>	< 1e-6)
1036	>	{ rps.push_back(make_pair(r1, r2)); }
1037	>	}
1038	>	}
1039	>	}
1040	>	// Consider combining together the solving pair part w/ the searching
1041	>	// best solution part so that we don't need the pairs vector
1042	>	if (!rps.empty()) {
1043	>	RealType smallestDiff = HONKING_LARGE_VALUE;
1044	>	RealType diff;
1045	>	pair<RealType,RealType> bestPair = make_pair(1.0, 1.0);
1046	>	vector<pair<RealType,RealType> >::iterator rpi;
1047	>	for (rpi = rps.begin(); rpi != rps.end(); rpi++) {
1048	>	r1 = (*rpi).first;
1049	>	r2 = (*rpi).second;
1050	>	switch(rnemdType_) {
1051	>	case rnemdKineticScale :
1052	>	diff = fastpow(1.0 - r1, 2) + fastpow(1.0 - r2, 2)
1053	>	+ fastpow(r1 * r1 / r2 / r2 - Kcz/Kcx, 2)
1054	>	+ fastpow(r1 * r1 / r2 / r2 - Kcz/Kcy, 2);
1055	>	break;
1056	>	case rnemdPxScale :
1057	>	diff = fastpow(1.0 - r1, 2) + fastpow(1.0 - r2, 2)
1058	>	+ fastpow(r1 * r1 / r2 / r2 - Kcz/Kcy, 2);
1059	>	break;
1060	>	case rnemdPyScale :
1061	>	diff = fastpow(1.0 - r1, 2) + fastpow(1.0 - r2, 2)
1062	>	+ fastpow(r1 * r1 / r2 / r2 - Kcz/Kcx, 2);
1063	>	break;
1064	>	case rnemdPzScale :
1065	>	diff = fastpow(1.0 - r1, 2) + fastpow(1.0 - r2, 2)
1066	>	+ fastpow(r1 * r1 / r2 / r2 - Kcy/Kcx, 2);
1067	>	default :
1068	>	break;
1069	>	}
1070	>	if (diff < smallestDiff) {
1071	>	smallestDiff = diff;
1072	>	bestPair = *rpi;
1073	>	}
1074	>	}
1075	>	#ifdef IS_MPI
1076	>	if (worldRank == 0) {
1077	>	#endif
1078	>	sprintf(painCave.errMsg,
1079	>	"RNEMD: roots r1= %lf\tr2 = %lf\n",
1080	>	bestPair.first, bestPair.second);
1081	>	painCave.isFatal = 0;
1082	>	painCave.severity = OPENMD_INFO;
1083	>	simError();
1084	>	#ifdef IS_MPI
1085	>	}
1086	>	#endif
1087	>
1088	>	switch(rnemdType_) {
1089	>	case rnemdKineticScale :
1090	>	x = bestPair.first;
1091	>	y = bestPair.first;
1092	>	z = bestPair.second;
1093	>	break;
1094	>	case rnemdPxScale :
1095	>	x = c;
1096	>	y = bestPair.first;
1097	>	z = bestPair.second;
1098	>	break;
1099	>	case rnemdPyScale :
1100	>	x = bestPair.first;
1101	>	y = c;
1102	>	z = bestPair.second;
1103	>	break;
1104	>	case rnemdPzScale :
1105	>	x = bestPair.first;
1106	>	y = bestPair.second;
1107	>	z = c;
1108	>	break;
1109	>	default :
1110	>	break;
1111	>	}
1112	>	vector<StuntDouble*>::iterator sdi;
1113	>	Vector3d vel;
1114	>	for (sdi = coldBin.begin(); sdi != coldBin.end(); sdi++) {
1115	>	vel = (*sdi)->getVel();
1116	>	vel.x() *= x;
1117	>	vel.y() *= y;
1118	>	vel.z() *= z;
1119	>	(*sdi)->setVel(vel);
1120	>	}
1121	>	//convert to hotBin coefficient
1122	>	x = 1.0 + px * (1.0 - x);
1123	>	y = 1.0 + py * (1.0 - y);
1124	>	z = 1.0 + pz * (1.0 - z);
1125	>	for (sdi = hotBin.begin(); sdi != hotBin.end(); sdi++) {
1126	>	vel = (*sdi)->getVel();
1127	>	vel.x() *= x;
1128	>	vel.y() *= y;
1129	>	vel.z() *= z;
1130	>	(*sdi)->setVel(vel);
1131	>	}
1132	>	successfulScale = true;
1133	>	exchangeSum_ += targetFlux_;
1134	>	}
1135	>	}
1136	>	if (successfulScale != true) {
1137	>	sprintf(painCave.errMsg,
1138	>	"RNEMD: exchange NOT performed!\n");
1139	>	painCave.isFatal = 0;
1140	>	painCave.severity = OPENMD_INFO;
1141	>	simError();
1142	>	failTrialCount_++;
1143	>	}
1144	>	}
1145	>
1146	>	void RNEMD::doShiftScale() {
1147	>
1148	>	Snapshot* currentSnap_ = info_->getSnapshotManager()->getCurrentSnapshot();
1149	>	RealType time = currentSnap_->getTime();
1150	>	Mat3x3d hmat = currentSnap_->getHmat();
1151	>
1152	>	seleMan_.setSelectionSet(evaluator_.evaluate());
1153	>
1154	>	int selei;
1155	>	StuntDouble* sd;
1156	>	int idx;
1157	>
1158	>	vector<StuntDouble*> hotBin, coldBin;
1159	>
1160	>	Vector3d Ph(V3Zero);
1161	>	RealType Mh = 0.0;
1162	>	RealType Kh = 0.0;
1163	>	Vector3d Pc(V3Zero);
1164	>	RealType Mc = 0.0;
1165	>	RealType Kc = 0.0;
1166	>
1167	>
1168	>	for (sd = seleMan_.beginSelected(selei); sd != NULL;
1169	>	sd = seleMan_.nextSelected(selei)) {
1170	>
1171	>	idx = sd->getLocalIndex();
1172	>
1173	>	Vector3d pos = sd->getPos();
1174	>
1175	>	// wrap the stuntdouble's position back into the box:
1176	>
1177	>	if (usePeriodicBoundaryConditions_)
1178	>	currentSnap_->wrapVector(pos);
1179	>
1180	>	// which bin is this stuntdouble in?
1181	>	// wrapped positions are in the range [-0.5*hmat(2,2), +0.5*hmat(2,2)]
1182	>
1183	>	int binNo = int(nBins_ * (pos.z() / hmat(2,2) + zShift_ + 0.5)) % nBins_;
1184	>
1185	>	// if we're in bin 0 or the middleBin
1186	>	if (binNo == 0 || binNo == midBin_) {
1187	>
1188	>	RealType mass = sd->getMass();
1189	>	Vector3d vel = sd->getVel();
1190	>
1191	>	if (binNo == 0) {
1192	>	hotBin.push_back(sd);
1193	>	//std::cerr << "before, velocity = " << vel << endl;
1194	>	Ph += mass * vel;
1195	>	//std::cerr << "after, velocity = " << vel << endl;
1196	>	Mh += mass;
1197	>	Kh += mass * vel.lengthSquare();
1198	>	if (rnemdType_ == rnemdShiftScaleVAM) {
1199	>	if (sd->isDirectional()) {
1200	>	Vector3d angMom = sd->getJ();
1201	>	Mat3x3d I = sd->getI();
1202	>	if (sd->isLinear()) {
1203	>	int i = sd->linearAxis();
1204	>	int j = (i + 1) % 3;
1205	>	int k = (i + 2) % 3;
1206	>	Kh += angMom[j] * angMom[j] / I(j, j) +
1207	>	angMom[k] * angMom[k] / I(k, k);
1208	>	} else {
1209	>	Kh += angMom[0] * angMom[0] / I(0, 0) +
1210	>	angMom[1] * angMom[1] / I(1, 1) +
1211	>	angMom[2] * angMom[2] / I(2, 2);
1212	>	}
1213	>	}
1214	>	}
1215	>	} else { //midBin_
1216	>	coldBin.push_back(sd);
1217	>	Pc += mass * vel;
1218	>	Mc += mass;
1219	>	Kc += mass * vel.lengthSquare();
1220	>	if (rnemdType_ == rnemdShiftScaleVAM) {
1221	>	if (sd->isDirectional()) {
1222	>	Vector3d angMom = sd->getJ();
1223	>	Mat3x3d I = sd->getI();
1224	>	if (sd->isLinear()) {
1225	>	int i = sd->linearAxis();
1226	>	int j = (i + 1) % 3;
1227	>	int k = (i + 2) % 3;
1228	>	Kc += angMom[j] * angMom[j] / I(j, j) +
1229	>	angMom[k] * angMom[k] / I(k, k);
1230	>	} else {
1231	>	Kc += angMom[0] * angMom[0] / I(0, 0) +
1232	>	angMom[1] * angMom[1] / I(1, 1) +
1233	>	angMom[2] * angMom[2] / I(2, 2);
1234	>	}
1235	>	}
1236	>	}
1237	>	}
1238	>	}
1239	>	}
1240	>
1241	>	Kh *= 0.5;
1242	>	Kc *= 0.5;
1243	>
1244	>	// std::cerr << "Mh= " << Mh << "\tKh= " << Kh << "\tMc= " << Mc
1245	>	//        << "\tKc= " << Kc << endl;
1246	>	// std::cerr << "Ph= " << Ph << "\tPc= " << Pc << endl;
1247	>
1248	>	#ifdef IS_MPI
1249	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Ph[0], 3, MPI::REALTYPE, MPI::SUM);
1250	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Pc[0], 3, MPI::REALTYPE, MPI::SUM);
1251	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Mh, 1, MPI::REALTYPE, MPI::SUM);
1252	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Kh, 1, MPI::REALTYPE, MPI::SUM);
1253	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Mc, 1, MPI::REALTYPE, MPI::SUM);
1254	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Kc, 1, MPI::REALTYPE, MPI::SUM);
1255	>	#endif
1256	>
1257	>	bool successfulExchange = false;
1258	>	if ((Mh > 0.0) && (Mc > 0.0)) {//both slabs are not empty
1259	>	Vector3d vc = Pc / Mc;
1260	>	Vector3d ac = njzp_ / Mc + vc;
1261	>	Vector3d acrec = njzp_ / Mc;
1262	>	RealType cNumerator = Kc - targetJzKE_ - 0.5 * Mc * ac.lengthSquare();
1263	>	if (cNumerator > 0.0) {
1264	>	RealType cDenominator = Kc - 0.5 * Mc * vc.lengthSquare();
1265	>	if (cDenominator > 0.0) {
1266	>	RealType c = sqrt(cNumerator / cDenominator);
1267	>	if ((c > 0.9) && (c < 1.1)) {//restrict scaling coefficients
1268	>	Vector3d vh = Ph / Mh;
1269	>	Vector3d ah = jzp_ / Mh + vh;
1270	>	Vector3d ahrec = jzp_ / Mh;
1271	>	RealType hNumerator = Kh + targetJzKE_
1272	>	- 0.5 * Mh * ah.lengthSquare();
1273	>	if (hNumerator > 0.0) {
1274	>	RealType hDenominator = Kh - 0.5 * Mh * vh.lengthSquare();
1275	>	if (hDenominator > 0.0) {
1276	>	RealType h = sqrt(hNumerator / hDenominator);
1277	>	if ((h > 0.9) && (h < 1.1)) {
1278	>	// std::cerr << "cold slab scaling coefficient: " << c << "\n";
1279	>	// std::cerr << "hot slab scaling coefficient: " << h <<  "\n";
1280	>	vector<StuntDouble*>::iterator sdi;
1281	>	Vector3d vel;
1282	>	for (sdi = coldBin.begin(); sdi != coldBin.end(); sdi++) {
1283	>	//vel = (*sdi)->getVel();
1284	>	vel = ((*sdi)->getVel() - vc) * c + ac;
1285	>	(*sdi)->setVel(vel);
1286	>	if (rnemdType_ == rnemdShiftScaleVAM) {
1287	>	if ((*sdi)->isDirectional()) {
1288	>	Vector3d angMom = (*sdi)->getJ() * c;
1289	>	(*sdi)->setJ(angMom);
1290	>	}
1291	>	}
1292	>	}
1293	>	for (sdi = hotBin.begin(); sdi != hotBin.end(); sdi++) {
1294	>	//vel = (*sdi)->getVel();
1295	>	vel = ((*sdi)->getVel() - vh) * h + ah;
1296	>	(*sdi)->setVel(vel);
1297	>	if (rnemdType_ == rnemdShiftScaleVAM) {
1298	>	if ((*sdi)->isDirectional()) {
1299	>	Vector3d angMom = (*sdi)->getJ() * h;
1300	>	(*sdi)->setJ(angMom);
1301	>	}
1302	>	}
1303	>	}
1304	>	successfulExchange = true;
1305	>	exchangeSum_ += targetFlux_;
1306	>	// this is a redundant variable for doShiftScale() so that
1307	>	// RNEMD can output one exchange quantity needed in a job.
1308	>	// need a better way to do this.
1309	>	//cerr << "acx =" << ac.x() << "ahx =" << ah.x() << '\n';
1310	>	//cerr << "acy =" << ac.y() << "ahy =" << ah.y() << '\n';
1311	>	//cerr << "acz =" << ac.z() << "ahz =" << ah.z() << '\n';
1312	>	Asum_ += (ahrec.z() - acrec.z());
1313	>	Jsum_ += (jzp_.z()*((1/Mh)+(1/Mc)));
1314	>	AhCount_ = ahrec.z();
1315	>	if (outputAh_) {
1316	>	AhLog_ << time << "   ";
1317	>	AhLog_ << AhCount_;
1318	>	AhLog_ << endl;
1319	>	}
1320	>	}
1321	>	}
1322	>	}
1323	>	}
1324	>	}
1325	>	}
1326	>	}
1327	>	if (successfulExchange != true) {
1328	>	//   sprintf(painCave.errMsg,
1329	>	//              "RNEMD: exchange NOT performed!\n");
1330	>	//   painCave.isFatal = 0;
1331	>	//   painCave.severity = OPENMD_INFO;
1332	>	//   simError();
1333	>	failTrialCount_++;
1334	>	}
1335	>	}
1336	>
1337	>	void RNEMD::doRNEMD() {
1338	>
1339	>	switch(rnemdType_) {
1340	>	case rnemdKineticScale :
1341	>	case rnemdKineticScaleVAM :
1342	>	case rnemdKineticScaleAM :
1343	>	case rnemdPxScale :
1344	>	case rnemdPyScale :
1345	>	case rnemdPzScale :
1346	>	doScale();
1347	>	break;
1348	>	case rnemdKineticSwap :
1349	>	case rnemdPx :
1350	>	case rnemdPy :
1351	>	case rnemdPz :
1352	>	doSwap();
1353	>	break;
1354	>	case rnemdShiftScaleV :
1355	>	case rnemdShiftScaleVAM :
1356	>	doShiftScale();
1357	>	break;
1358	>	case rnemdUnknown :
1359	>	default :
1360	>	break;
1361	>	}
1362	>	}
1363	>
1364	>	void RNEMD::collectData() {
1365	>
1366	>	Snapshot* currentSnap_ = info_->getSnapshotManager()->getCurrentSnapshot();
1367	>	Mat3x3d hmat = currentSnap_->getHmat();
1368	>
1369	>	seleMan_.setSelectionSet(evaluator_.evaluate());
1370	>
1371	>	int selei;
1372	>	StuntDouble* sd;
1373	>	int idx;
1374	>
1375	>	logFrameCount_++;
1376	>
1377	>	// alternative approach, track all molecules instead of only those
1378	>	// selected for scaling/swapping:
1379	>	/*
1380	>	SimInfo::MoleculeIterator miter;
1381	>	vector<StuntDouble*>::iterator iiter;
1382	>	Molecule* mol;
1383	>	StuntDouble* integrableObject;
1384	>	for (mol = info_->beginMolecule(miter); mol != NULL;
1385	>	mol = info_->nextMolecule(miter))
1386	>	integrableObject is essentially sd
1387	>	for (integrableObject = mol->beginIntegrableObject(iiter);
1388	>	integrableObject != NULL;
1389	>	integrableObject = mol->nextIntegrableObject(iiter))
1390	>	*/
1391	>	for (sd = seleMan_.beginSelected(selei); sd != NULL;
1392	>	sd = seleMan_.nextSelected(selei)) {
1393	>
1394	>	idx = sd->getLocalIndex();
1395	>
1396	>	Vector3d pos = sd->getPos();
1397	>
1398	>	// wrap the stuntdouble's position back into the box:
1399	>
1400	>	if (usePeriodicBoundaryConditions_)
1401	>	currentSnap_->wrapVector(pos);
1402	>
1403	>	// which bin is this stuntdouble in?
1404	>	// wrapped positions are in the range [-0.5*hmat(2,2), +0.5*hmat(2,2)]
1405	>
1406	>	int binNo = int(rnemdLogWidth_ * (pos.z() / hmat(2,2) + 0.5)) %
1407	>	rnemdLogWidth_;
1408	>	// no symmetrization allowed due to arbitary rnemdLogWidth_
1409	>	/*
1410	>	if (rnemdLogWidth_ == midBin_ + 1)
1411	>	if (binNo > midBin_)
1412	>	binNo = nBins_ - binNo;
1413	>	*/
1414	>	RealType mass = sd->getMass();
1415	>	mHist_[binNo] += mass;
1416	>	Vector3d vel = sd->getVel();
1417	>	RealType value;
1418	>	//RealType xVal, yVal, zVal;
1419	>
1420	>	if (outputTemp_) {
1421	>	value = mass * vel.lengthSquare();
1422	>	tempCount_[binNo] += 3;
1423	>	if (sd->isDirectional()) {
1424	>	Vector3d angMom = sd->getJ();
1425	>	Mat3x3d I = sd->getI();
1426	>	if (sd->isLinear()) {
1427	>	int i = sd->linearAxis();
1428	>	int j = (i + 1) % 3;
1429	>	int k = (i + 2) % 3;
1430	>	value += angMom[j] * angMom[j] / I(j, j) +
1431	>	angMom[k] * angMom[k] / I(k, k);
1432	>	tempCount_[binNo] +=2;
1433	>	} else {
1434	>	value += angMom[0] * angMom[0] / I(0, 0) +
1435	>	angMom[1]*angMom[1]/I(1, 1) +
1436	>	angMom[2]*angMom[2]/I(2, 2);
1437	>	tempCount_[binNo] +=3;
1438	>	}
1439	>	}
1440	>	value = value / PhysicalConstants::energyConvert
1441	>	/ PhysicalConstants::kb;//may move to getStatus()
1442	>	tempHist_[binNo] += value;
1443	>	}
1444	>	if (outputVx_) {
1445	>	value = mass * vel[0];
1446	>	//vxzCount_[binNo]++;
1447	>	pxzHist_[binNo] += value;
1448	>	}
1449	>	if (outputVy_) {
1450	>	value = mass * vel[1];
1451	>	//vyzCount_[binNo]++;
1452	>	pyzHist_[binNo] += value;
1453	>	}
1454	>
1455	>	if (output3DTemp_) {
1456	>	value = mass * vel.x() * vel.x();
1457	>	xTempHist_[binNo] += value;
1458	>	value = mass * vel.y() * vel.y() / PhysicalConstants::energyConvert
1459	>	/ PhysicalConstants::kb;
1460	>	yTempHist_[binNo] += value;
1461	>	value = mass * vel.z() * vel.z() / PhysicalConstants::energyConvert
1462	>	/ PhysicalConstants::kb;
1463	>	zTempHist_[binNo] += value;
1464	>	xyzTempCount_[binNo]++;
1465	>	}
1466	>	if (outputRotTemp_) {
1467	>	if (sd->isDirectional()) {
1468	>	Vector3d angMom = sd->getJ();
1469	>	Mat3x3d I = sd->getI();
1470	>	if (sd->isLinear()) {
1471	>	int i = sd->linearAxis();
1472	>	int j = (i + 1) % 3;
1473	>	int k = (i + 2) % 3;
1474	>	value = angMom[j] * angMom[j] / I(j, j) +
1475	>	angMom[k] * angMom[k] / I(k, k);
1476	>	rotTempCount_[binNo] +=2;
1477	>	} else {
1478	>	value = angMom[0] * angMom[0] / I(0, 0) +
1479	>	angMom[1] * angMom[1] / I(1, 1) +
1480	>	angMom[2] * angMom[2] / I(2, 2);
1481	>	rotTempCount_[binNo] +=3;
1482	>	}
1483	>	}
1484	>	value = value / PhysicalConstants::energyConvert
1485	>	/ PhysicalConstants::kb;//may move to getStatus()
1486	>	rotTempHist_[binNo] += value;
1487	>	}
1488	>	// James put this in.
1489	>	if (outputDen_) {
1490	>	//value = 1.0;
1491	>	DenHist_[binNo] += 1;
1492	>	}
1493	>	if (outputVz_) {
1494	>	value = mass * vel[2];
1495	>	//vyzCount_[binNo]++;
1496	>	pzzHist_[binNo] += value;
1497	>	}
1498	>	}
1499	>	}
1500	>
1501	>	void RNEMD::getStarted() {
1502	>	collectData();
1503	>	/*now can output profile in step 0, but might not be useful;
1504	>	Snapshot* currentSnap_ = info_->getSnapshotManager()->getCurrentSnapshot();
1505	>	Stats& stat = currentSnap_->statData;
1506	>	stat[Stats::RNEMD_EXCHANGE_TOTAL] = exchangeSum_;
1507	>	*/
1508	>	//may output a header for the log file here
1509	>	getStatus();
1510	>	}
1511	>
1512	>	void RNEMD::getStatus() {
1513	>
1514	>	Snapshot* currentSnap_ = info_->getSnapshotManager()->getCurrentSnapshot();
1515	>	Stats& stat = currentSnap_->statData;
1516	>	RealType time = currentSnap_->getTime();
1517	>
1518	>	stat[Stats::RNEMD_EXCHANGE_TOTAL] = exchangeSum_;
1519	>	//or to be more meaningful, define another item as exchangeSum_ / time
1520	>	int j;
1521	>
1522	>	#ifdef IS_MPI
1523	>
1524	>	// all processors have the same number of bins, and STL vectors pack their
1525	>	// arrays, so in theory, this should be safe:
1526	>
1527	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &mHist_[0],
1528	>	rnemdLogWidth_, MPI::REALTYPE, MPI::SUM);
1529	>	if (outputTemp_) {
1530	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &tempHist_[0],
1531	>	rnemdLogWidth_, MPI::REALTYPE, MPI::SUM);
1532	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &tempCount_[0],
1533	>	rnemdLogWidth_, MPI::INT, MPI::SUM);
1534	>	}
1535	>	if (outputVx_) {
1536	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &pxzHist_[0],
1537	>	rnemdLogWidth_, MPI::REALTYPE, MPI::SUM);
1538	>	//MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &vxzCount_[0],
1539	>	//                        rnemdLogWidth_, MPI::INT, MPI::SUM);
1540	>	}
1541	>	if (outputVy_) {
1542	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &pyzHist_[0],
1543	>	rnemdLogWidth_, MPI::REALTYPE, MPI::SUM);
1544	>	//MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &vyzCount_[0],
1545	>	//                        rnemdLogWidth_, MPI::INT, MPI::SUM);
1546	>	}
1547	>	if (output3DTemp_) {
1548	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &xTempHist_[0],
1549	>	rnemdLogWidth_, MPI::REALTYPE, MPI::SUM);
1550	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &yTempHist_[0],
1551	>	rnemdLogWidth_, MPI::REALTYPE, MPI::SUM);
1552	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &zTempHist_[0],
1553	>	rnemdLogWidth_, MPI::REALTYPE, MPI::SUM);
1554	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &xyzTempCount_[0],
1555	>	rnemdLogWidth_, MPI::INT, MPI::SUM);
1556	>	}
1557	>	if (outputRotTemp_) {
1558	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &rotTempHist_[0],
1559	>	rnemdLogWidth_, MPI::REALTYPE, MPI::SUM);
1560	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &rotTempCount_[0],
1561	>	rnemdLogWidth_, MPI::INT, MPI::SUM);
1562	>	}
1563	>	// James put this in
1564	>	if (outputDen_) {
1565	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &DenHist_[0],
1566	>	rnemdLogWidth_, MPI::REALTYPE, MPI::SUM);
1567	>	}
1568	>	if (outputAh_) {
1569	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &AhCount_,
1570	>	1, MPI::REALTYPE, MPI::SUM);
1571	>	}
1572	>	if (outputVz_) {
1573	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &pzzHist_[0],
1574	>	rnemdLogWidth_, MPI::REALTYPE, MPI::SUM);
1575	>	}
1576	>
1577	>	// If we're the root node, should we print out the results
1578	>	int worldRank = MPI::COMM_WORLD.Get_rank();
1579	>	if (worldRank == 0) {
1580	>	#endif
1581	>
1582	>	if (outputTemp_) {
1583	>	tempLog_ << time;
1584	>	for (j = 0; j < rnemdLogWidth_; j++) {
1585	>	tempLog_ << "\t" << tempHist_[j] / (RealType)tempCount_[j];
1586	>	}
1587	>	tempLog_ << endl;
1588	>	}
1589	>	if (outputVx_) {
1590	>	vxzLog_ << time;
1591	>	for (j = 0; j < rnemdLogWidth_; j++) {
1592	>	vxzLog_ << "\t" << pxzHist_[j] / mHist_[j];
1593	>	}
1594	>	vxzLog_ << endl;
1595	>	}
1596	>	if (outputVy_) {
1597	>	vyzLog_ << time;
1598	>	for (j = 0; j < rnemdLogWidth_; j++) {
1599	>	vyzLog_ << "\t" << pyzHist_[j] / mHist_[j];
1600	>	}
1601	>	vyzLog_ << endl;
1602	>	}
1603	>
1604	>	if (output3DTemp_) {
1605	>	RealType temp;
1606	>	xTempLog_ << time;
1607	>	for (j = 0; j < rnemdLogWidth_; j++) {
1608	>	if (outputVx_)
1609	>	xTempHist_[j] -= pxzHist_[j] * pxzHist_[j] / mHist_[j];
1610	>	temp = xTempHist_[j] / (RealType)xyzTempCount_[j]
1611	>	/ PhysicalConstants::energyConvert / PhysicalConstants::kb;
1612	>	xTempLog_ << "\t" << temp;
1613	>	}
1614	>	xTempLog_ << endl;
1615	>	yTempLog_ << time;
1616	>	for (j = 0; j < rnemdLogWidth_; j++) {
1617	>	yTempLog_ << "\t" << yTempHist_[j] / (RealType)xyzTempCount_[j];
1618	>	}
1619	>	yTempLog_ << endl;
1620	>	zTempLog_ << time;
1621	>	for (j = 0; j < rnemdLogWidth_; j++) {
1622	>	zTempLog_ << "\t" << zTempHist_[j] / (RealType)xyzTempCount_[j];
1623	>	}
1624	>	zTempLog_ << endl;
1625	>	}
1626	>	if (outputRotTemp_) {
1627	>	rotTempLog_ << time;
1628	>	for (j = 0; j < rnemdLogWidth_; j++) {
1629	>	rotTempLog_ << "\t" << rotTempHist_[j] / (RealType)rotTempCount_[j];
1630	>	}
1631	>	rotTempLog_ << endl;
1632	>	}
1633	>	// James put this in.
1634	>	Mat3x3d hmat = currentSnap_->getHmat();
1635	>	if (outputDen_) {
1636	>	denLog_ << time;
1637	>	for (j = 0; j < rnemdLogWidth_; j++) {
1638	>
1639	>	RealType binVol = hmat(0,0) * hmat(1,1) * (hmat(2,2) / float(nBins_));
1640	>	denLog_ << "\t" << DenHist_[j] / (float(logFrameCount_) * binVol);
1641	>	}
1642	>	denLog_ << endl;
1643	>	}
1644	>	if (outputVz_) {
1645	>	vzzLog_ << time;
1646	>	for (j = 0; j < rnemdLogWidth_; j++) {
1647	>	vzzLog_ << "\t" << pzzHist_[j] / mHist_[j];
1648	>	}
1649	>	vzzLog_ << endl;
1650	>	}
1651	>	#ifdef IS_MPI
1652	>	}
1653	>	#endif
1654	>
1655	>	for (j = 0; j < rnemdLogWidth_; j++) {
1656	>	mHist_[j] = 0.0;
1657	>	}
1658	>	if (outputTemp_)
1659	>	for (j = 0; j < rnemdLogWidth_; j++) {
1660	>	tempCount_[j] = 0;
1661	>	tempHist_[j] = 0.0;
1662	>	}
1663	>	if (outputVx_)
1664	>	for (j = 0; j < rnemdLogWidth_; j++) {
1665	>	//pxzCount_[j] = 0;
1666	>	pxzHist_[j] = 0.0;
1667	>	}
1668	>	if (outputVy_)
1669	>	for (j = 0; j < rnemdLogWidth_; j++) {
1670	>	//pyzCount_[j] = 0;
1671	>	pyzHist_[j] = 0.0;
1672	>	}
1673	>
1674	>	if (output3DTemp_)
1675	>	for (j = 0; j < rnemdLogWidth_; j++) {
1676	>	xTempHist_[j] = 0.0;
1677	>	yTempHist_[j] = 0.0;
1678	>	zTempHist_[j] = 0.0;
1679	>	xyzTempCount_[j] = 0;
1680	>	}
1681	>	if (outputRotTemp_)
1682	>	for (j = 0; j < rnemdLogWidth_; j++) {
1683	>	rotTempCount_[j] = 0;
1684	>	rotTempHist_[j] = 0.0;
1685	>	}
1686	>	// James put this in
1687	>	if (outputDen_)
1688	>	for (j = 0; j < rnemdLogWidth_; j++) {
1689	>	//pyzCount_[j] = 0;
1690	>	DenHist_[j] = 0.0;
1691	>	}
1692	>	if (outputVz_)
1693	>	for (j = 0; j < rnemdLogWidth_; j++) {
1694	>	//pyzCount_[j] = 0;
1695	>	pzzHist_[j] = 0.0;
1696	>	}
1697	>	// reset the counter
1698	>
1699	>	Numcount_++;
1700	>	if (Numcount_ > int(runTime_/statusTime_))
1701	>	cerr << "time =" << time << "  Asum =" << Asum_ << '\n';
1702	>	if (Numcount_ > int(runTime_/statusTime_))
1703	>	cerr << "time =" << time << "  Jsum =" << Jsum_ << '\n';
1704	>
1705	>	logFrameCount_ = 0;
1706	>	}
1707		}
1708	+

Comparing:
trunk/src/integrators/RNEMD.cpp (property svn:keywords), Revision 1330 by skuang, Thu Mar 19 21:03:36 2009 UTC vs.
branches/development/src/rnemd/RNEMD.cpp (property svn:keywords), Revision 1731 by gezelter, Thu May 31 12:25:30 2012 UTC

#	Line 0 | Line 1
1	+	Author Id Revision Date

Diff Legend

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