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 1350 by gezelter, Thu May 21 18:56:45 2009 UTC vs.
branches/development/src/rnemd/RNEMD.cpp (file contents), Revision 1776 by gezelter, Thu Aug 9 15:52:59 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/OOPSEConstant.hpp"
50	>	#include "utils/PhysicalConstants.hpp"
51		#include "utils/Tuple.hpp"
52	<
53	<	#ifndef IS_MPI
51	<	#include "math/SeqRandNumGen.hpp"
52	<	#else
53	<	#include "math/ParallelRandNumGen.hpp"
52	>	#ifdef IS_MPI
53	>	#include <mpi.h>
54		#endif
55
56		#define HONKING_LARGE_VALUE 1.0e10
57
58	<	namespace oopse {
58	>	using namespace std;
59	>	namespace OpenMD {
60
61	<	RNEMD::RNEMD(SimInfo* info) : info_(info), evaluator_(info), seleMan_(info), usePeriodicBoundaryConditions_(info->getSimParams()->getUsePeriodicBoundaryConditions()) {
62	<
61	>	RNEMD::RNEMD(SimInfo* info) : info_(info), evaluator_(info), seleMan_(info),
62	>	usePeriodicBoundaryConditions_(info->getSimParams()->getUsePeriodicBoundaryConditions()) {
63	>
64	>	trialCount_ = 0;
65	>	failTrialCount_ = 0;
66	>	failRootCount_ = 0;
67	>
68		int seedValue;
69		Globals * simParams = info->getSimParams();
70	+	RNEMDParameters* rnemdParams = simParams->getRNEMDParameters();
71
72	<	stringToEnumMap_["Kinetic"] = rnemdKinetic;
73	<	stringToEnumMap_["Px"] = rnemdPx;
67	<	stringToEnumMap_["Py"] = rnemdPy;
68	<	stringToEnumMap_["Pz"] = rnemdPz;
69	<	stringToEnumMap_["Unknown"] = rnemdUnknown;
72	>	doRNEMD_ = rnemdParams->getUseRNEMD();
73	>	if (!doRNEMD_) return;
74
75	<	rnemdObjectSelection_ = simParams->getRNEMD_objectSelection();
75	>	stringToMethod_["Swap"]  = rnemdSwap;
76	>	stringToMethod_["NIVS"]  = rnemdNIVS;
77	>	stringToMethod_["VSS"]   = rnemdVSS;
78	>
79	>	stringToFluxType_["KE"]  = rnemdKE;
80	>	stringToFluxType_["Px"]  = rnemdPx;
81	>	stringToFluxType_["Py"]  = rnemdPy;
82	>	stringToFluxType_["Pz"]  = rnemdPz;
83	>	stringToFluxType_["KE+Px"]  = rnemdKePx;
84	>	stringToFluxType_["KE+Py"]  = rnemdKePy;
85	>	stringToFluxType_["KE+Pvector"]  = rnemdKePvector;
86	>
87	>	runTime_ = simParams->getRunTime();
88	>	statusTime_ = simParams->getStatusTime();
89	>
90	>	rnemdObjectSelection_ = rnemdParams->getObjectSelection();
91		evaluator_.loadScriptString(rnemdObjectSelection_);
92		seleMan_.setSelectionSet(evaluator_.evaluate());
93
94	+	const string methStr = rnemdParams->getMethod();
95	+	bool hasFluxType = rnemdParams->haveFluxType();
96
97	+	string fluxStr;
98	+	if (hasFluxType) {
99	+	fluxStr = rnemdParams->getFluxType();
100	+	} else {
101	+	sprintf(painCave.errMsg,
102	+	"RNEMD: No fluxType was set in the md file.  This parameter,\n"
103	+	"\twhich must be one of the following values:\n"
104	+	"\tKE, Px, Py, Pz, KE+Px, KE+Py, KE+Pvector, must be set to\n"
105	+	"\tuse RNEMD\n");
106	+	painCave.isFatal = 1;
107	+	painCave.severity = OPENMD_ERROR;
108	+	simError();
109	+	}
110	+
111	+	bool hasKineticFlux = rnemdParams->haveKineticFlux();
112	+	bool hasMomentumFlux = rnemdParams->haveMomentumFlux();
113	+	bool hasMomentumFluxVector = rnemdParams->haveMomentumFluxVector();
114	+	bool hasSlabWidth = rnemdParams->haveSlabWidth();
115	+	bool hasSlabACenter = rnemdParams->haveSlabACenter();
116	+	bool hasSlabBCenter = rnemdParams->haveSlabBCenter();
117	+	bool hasOutputFileName = rnemdParams->haveOutputFileName();
118	+	bool hasOutputFields = rnemdParams->haveOutputFields();
119	+
120	+	map<string, RNEMDMethod>::iterator i;
121	+	i = stringToMethod_.find(methStr);
122	+	if (i != stringToMethod_.end())
123	+	rnemdMethod_ = i->second;
124	+	else {
125	+	sprintf(painCave.errMsg,
126	+	"RNEMD: The current method,\n"
127	+	"\t\t%s is not one of the recognized\n"
128	+	"\texchange methods: Swap, NIVS, or VSS\n",
129	+	methStr.c_str());
130	+	painCave.isFatal = 1;
131	+	painCave.severity = OPENMD_ERROR;
132	+	simError();
133	+	}
134	+
135	+	map<string, RNEMDFluxType>::iterator j;
136	+	j = stringToFluxType_.find(fluxStr);
137	+	if (j != stringToFluxType_.end())
138	+	rnemdFluxType_ = j->second;
139	+	else {
140	+	sprintf(painCave.errMsg,
141	+	"RNEMD: The current fluxType,\n"
142	+	"\t\t%s\n"
143	+	"\tis not one of the recognized flux types.\n",
144	+	fluxStr.c_str());
145	+	painCave.isFatal = 1;
146	+	painCave.severity = OPENMD_ERROR;
147	+	simError();
148	+	}
149	+
150	+	bool methodFluxMismatch = false;
151	+	bool hasCorrectFlux = false;
152	+	switch(rnemdMethod_) {
153	+	case rnemdSwap:
154	+	switch (rnemdFluxType_) {
155	+	case rnemdKE:
156	+	hasCorrectFlux = hasKineticFlux;
157	+	break;
158	+	case rnemdPx:
159	+	case rnemdPy:
160	+	case rnemdPz:
161	+	hasCorrectFlux = hasMomentumFlux;
162	+	break;
163	+	default :
164	+	methodFluxMismatch = true;
165	+	break;
166	+	}
167	+	break;
168	+	case rnemdNIVS:
169	+	switch (rnemdFluxType_) {
170	+	case rnemdKE:
171	+	case rnemdRotKE:
172	+	case rnemdFullKE:
173	+	hasCorrectFlux = hasKineticFlux;
174	+	break;
175	+	case rnemdPx:
176	+	case rnemdPy:
177	+	case rnemdPz:
178	+	hasCorrectFlux = hasMomentumFlux;
179	+	break;
180	+	case rnemdKePx:
181	+	case rnemdKePy:
182	+	hasCorrectFlux = hasMomentumFlux && hasKineticFlux;
183	+	break;
184	+	default:
185	+	methodFluxMismatch = true;
186	+	break;
187	+	}
188	+	break;
189	+	case rnemdVSS:
190	+	switch (rnemdFluxType_) {
191	+	case rnemdKE:
192	+	case rnemdRotKE:
193	+	case rnemdFullKE:
194	+	hasCorrectFlux = hasKineticFlux;
195	+	break;
196	+	case rnemdPx:
197	+	case rnemdPy:
198	+	case rnemdPz:
199	+	hasCorrectFlux = hasMomentumFlux;
200	+	break;
201	+	case rnemdPvector:
202	+	hasCorrectFlux = hasMomentumFluxVector;
203	+	case rnemdKePx:
204	+	case rnemdKePy:
205	+	hasCorrectFlux = hasMomentumFlux && hasKineticFlux;
206	+	break;
207	+	case rnemdKePvector:
208	+	hasCorrectFlux = hasMomentumFluxVector && hasKineticFlux;
209	+	break;
210	+	default:
211	+	methodFluxMismatch = true;
212	+	break;
213	+	}
214	+	default:
215	+	break;
216	+	}
217	+
218	+	if (methodFluxMismatch) {
219	+	sprintf(painCave.errMsg,
220	+	"RNEMD: The current method,\n"
221	+	"\t\t%s\n"
222	+	"\tcannot be used with the current flux type, %s\n",
223	+	methStr.c_str(), fluxStr.c_str());
224	+	painCave.isFatal = 1;
225	+	painCave.severity = OPENMD_ERROR;
226	+	simError();
227	+	}
228	+	if (!hasCorrectFlux) {
229	+	sprintf(painCave.errMsg,
230	+	"RNEMD: The current method,\n"
231	+	"\t%s, and flux type %s\n"
232	+	"\tdid not have the correct flux value specified. Options\n"
233	+	"\tinclude: kineticFlux, momentumFlux, and momentumFluxVector\n",
234	+	methStr.c_str(), fluxStr.c_str());
235	+	painCave.isFatal = 1;
236	+	painCave.severity = OPENMD_ERROR;
237	+	simError();
238	+	}
239	+
240	+	if (hasKineticFlux) {
241	+	kineticFlux_ = rnemdParams->getKineticFlux();
242	+	} else {
243	+	kineticFlux_ = 0.0;
244	+	}
245	+	if (hasMomentumFluxVector) {
246	+	momentumFluxVector_ = rnemdParams->getMomentumFluxVector();
247	+	} else {
248	+	momentumFluxVector_ = V3Zero;
249	+	if (hasMomentumFlux) {
250	+	RealType momentumFlux = rnemdParams->getMomentumFlux();
251	+	switch (rnemdFluxType_) {
252	+	case rnemdPx:
253	+	momentumFluxVector_.x() = momentumFlux;
254	+	break;
255	+	case rnemdPy:
256	+	momentumFluxVector_.y() = momentumFlux;
257	+	break;
258	+	case rnemdPz:
259	+	momentumFluxVector_.z() = momentumFlux;
260	+	break;
261	+	case rnemdKePx:
262	+	momentumFluxVector_.x() = momentumFlux;
263	+	break;
264	+	case rnemdKePy:
265	+	momentumFluxVector_.y() = momentumFlux;
266	+	break;
267	+	default:
268	+	break;
269	+	}
270	+	}
271	+	}
272	+
273		// do some sanity checking
274
275		int selectionCount = seleMan_.getSelectionCount();
#	Line 80 | Line 277 | namespace oopse {
277
278		if (selectionCount > nIntegrable) {
279		sprintf(painCave.errMsg,
280	<	"RNEMD warning: The current RNEMD_objectSelection,\n"
280	>	"RNEMD: The current objectSelection,\n"
281		"\t\t%s\n"
282		"\thas resulted in %d selected objects.  However,\n"
283		"\tthe total number of integrable objects in the system\n"
#	Line 90 | Line 287 | namespace oopse {
287		rnemdObjectSelection_.c_str(),
288		selectionCount, nIntegrable);
289		painCave.isFatal = 0;
290	+	painCave.severity = OPENMD_WARNING;
291		simError();
292	+	}
293
294	+	areaAccumulator_ = new Accumulator();
295	+
296	+	nBins_ = rnemdParams->getOutputBins();
297	+
298	+	data_.resize(RNEMD::ENDINDEX);
299	+	OutputData z;
300	+	z.units =  "Angstroms";
301	+	z.title =  "Z";
302	+	z.dataType = "RealType";
303	+	z.accumulator.reserve(nBins_);
304	+	for (unsigned int i = 0; i < nBins_; i++)
305	+	z.accumulator.push_back( new Accumulator() );
306	+	data_[Z] = z;
307	+	outputMap_["Z"] =  Z;
308	+
309	+	OutputData temperature;
310	+	temperature.units =  "K";
311	+	temperature.title =  "Temperature";
312	+	temperature.dataType = "RealType";
313	+	temperature.accumulator.reserve(nBins_);
314	+	for (unsigned int i = 0; i < nBins_; i++)
315	+	temperature.accumulator.push_back( new Accumulator() );
316	+	data_[TEMPERATURE] = temperature;
317	+	outputMap_["TEMPERATURE"] =  TEMPERATURE;
318	+
319	+	OutputData velocity;
320	+	velocity.units = "amu/fs";
321	+	velocity.title =  "Velocity";
322	+	velocity.dataType = "Vector3d";
323	+	velocity.accumulator.reserve(nBins_);
324	+	for (unsigned int i = 0; i < nBins_; i++)
325	+	velocity.accumulator.push_back( new VectorAccumulator() );
326	+	data_[VELOCITY] = velocity;
327	+	outputMap_["VELOCITY"] = VELOCITY;
328	+
329	+	OutputData density;
330	+	density.units =  "g cm^-3";
331	+	density.title =  "Density";
332	+	density.dataType = "RealType";
333	+	density.accumulator.reserve(nBins_);
334	+	for (unsigned int i = 0; i < nBins_; i++)
335	+	density.accumulator.push_back( new Accumulator() );
336	+	data_[DENSITY] = density;
337	+	outputMap_["DENSITY"] =  DENSITY;
338	+
339	+	if (hasOutputFields) {
340	+	parseOutputFileFormat(rnemdParams->getOutputFields());
341	+	} else {
342	+	outputMask_.set(Z);
343	+	switch (rnemdFluxType_) {
344	+	case rnemdKE:
345	+	case rnemdRotKE:
346	+	case rnemdFullKE:
347	+	outputMask_.set(TEMPERATURE);
348	+	break;
349	+	case rnemdPx:
350	+	case rnemdPy:
351	+	outputMask_.set(VELOCITY);
352	+	break;
353	+	case rnemdPz:
354	+	case rnemdPvector:
355	+	outputMask_.set(VELOCITY);
356	+	outputMask_.set(DENSITY);
357	+	break;
358	+	case rnemdKePx:
359	+	case rnemdKePy:
360	+	outputMask_.set(TEMPERATURE);
361	+	outputMask_.set(VELOCITY);
362	+	break;
363	+	case rnemdKePvector:
364	+	outputMask_.set(TEMPERATURE);
365	+	outputMask_.set(VELOCITY);
366	+	outputMask_.set(DENSITY);
367	+	break;
368	+	default:
369	+	break;
370	+	}
371		}
372	<
373	<	const std::string st = simParams->getRNEMD_swapType();
372	>
373	>	if (hasOutputFileName) {
374	>	rnemdFileName_ = rnemdParams->getOutputFileName();
375	>	} else {
376	>	rnemdFileName_ = getPrefix(info->getFinalConfigFileName()) + ".rnemd";
377	>	}
378
379	<	std::map<std::string, RNEMDTypeEnum>::iterator i;
100	<	i = stringToEnumMap_.find(st);
101	<	rnemdType_  = (i == stringToEnumMap_.end()) ? RNEMD::rnemdUnknown : i->second;
379	>	exchangeTime_ = rnemdParams->getExchangeTime();
380
381	<	set_RNEMD_swapTime(simParams->getRNEMD_swapTime());
382	<	set_RNEMD_nBins(simParams->getRNEMD_nBins());
383	<	exchangeSum_ = 0.0;
381	>	Snapshot* currentSnap_ = info->getSnapshotManager()->getCurrentSnapshot();
382	>	Mat3x3d hmat = currentSnap_->getHmat();
383	>
384	>	// Target exchange quantities (in each exchange) =  2 Lx Ly dt flux
385	>	// Lx, Ly = box dimensions in x & y
386	>	// dt = exchange time interval
387	>	// flux = target flux
388
389	<	#ifndef IS_MPI
390	<	if (simParams->haveSeed()) {
391	<	seedValue = simParams->getSeed();
392	<	randNumGen_ = new SeqRandNumGen(seedValue);
393	<	}else {
394	<	randNumGen_ = new SeqRandNumGen();
395	<	}
396	<	#else
397	<	if (simParams->haveSeed()) {
398	<	seedValue = simParams->getSeed();
399	<	randNumGen_ = new ParallelRandNumGen(seedValue);
400	<	}else {
401	<	randNumGen_ = new ParallelRandNumGen();
120	<	}
121	<	#endif
122	<	}
389	>	RealType area = currentSnap_->getXYarea();
390	>	kineticTarget_ = 2.0 * kineticFlux_ * exchangeTime_ * area;
391	>	momentumTarget_ = 2.0 * momentumFluxVector_ * exchangeTime_ * area;
392	>
393	>	// total exchange sums are zeroed out at the beginning:
394	>
395	>	kineticExchange_ = 0.0;
396	>	momentumExchange_ = V3Zero;
397	>
398	>	if (hasSlabWidth)
399	>	slabWidth_ = rnemdParams->getSlabWidth();
400	>	else
401	>	slabWidth_ = hmat(2,2) / 10.0;
402
403	+	if (hasSlabACenter)
404	+	slabACenter_ = rnemdParams->getSlabACenter();
405	+	else
406	+	slabACenter_ = 0.0;
407	+
408	+	if (hasSlabBCenter)
409	+	slabBCenter_ = rnemdParams->getSlabBCenter();
410	+	else
411	+	slabBCenter_ = hmat(2,2) / 2.0;
412	+
413	+	}
414	+
415		RNEMD::~RNEMD() {
416	<	delete randNumGen_;
416	>	if (!doRNEMD_) return;
417	>	#ifdef IS_MPI
418	>	if (worldRank == 0) {
419	>	#endif
420	>
421	>	writeOutputFile();
422	>
423	>	rnemdFile_.close();
424	>
425	>	#ifdef IS_MPI
426	>	}
427	>	#endif
428		}
429	+
430	+	bool RNEMD::inSlabA(Vector3d pos) {
431	+	return (abs(pos.z() - slabACenter_) < 0.5*slabWidth_);
432	+	}
433	+	bool RNEMD::inSlabB(Vector3d pos) {
434	+	return (abs(pos.z() - slabBCenter_) < 0.5*slabWidth_);
435	+	}
436
437		void RNEMD::doSwap() {
438	<	int midBin = nBins_ / 2;
130	<
438	>	if (!doRNEMD_) return;
439		Snapshot* currentSnap_ = info_->getSnapshotManager()->getCurrentSnapshot();
440		Mat3x3d hmat = currentSnap_->getHmat();
441
#	Line 156 | Line 464 | namespace oopse {
464
465		if (usePeriodicBoundaryConditions_)
466		currentSnap_->wrapVector(pos);
467	+	bool inA = inSlabA(pos);
468	+	bool inB = inSlabB(pos);
469
470	<	// which bin is this stuntdouble in?
161	<	// wrapped positions are in the range [-0.5*hmat(2,2), +0.5*hmat(2,2)]
162	<
163	<	int binNo = int(nBins_ * (pos.z() / hmat(2,2) + 0.5)) % nBins_;
164	<
165	<
166	<	// if we're in bin 0 or the middleBin
167	<	if (binNo == 0 || binNo == midBin) {
470	>	if (inA || inB) {
471
472		RealType mass = sd->getMass();
473		Vector3d vel = sd->getVel();
474		RealType value;
475	<
476	<	switch(rnemdType_) {
477	<	case rnemdKinetic :
475	>
476	>	switch(rnemdFluxType_) {
477	>	case rnemdKE :
478
479	<	value = mass * (vel[0]*vel[0] + vel[1]*vel[1] +
480	<	vel[2]*vel[2]);
481	<	if (sd->isDirectional()) {
479	>	value = mass * vel.lengthSquare();
480	>
481	>	if (sd->isDirectional()) {
482		Vector3d angMom = sd->getJ();
483		Mat3x3d I = sd->getI();
484
485		if (sd->isLinear()) {
486	<	int i = sd->linearAxis();
487	<	int j = (i + 1) % 3;
488	<	int k = (i + 2) % 3;
489	<	value += angMom[j] * angMom[j] / I(j, j) +
490	<	angMom[k] * angMom[k] / I(k, k);
486	>	int i = sd->linearAxis();
487	>	int j = (i + 1) % 3;
488	>	int k = (i + 2) % 3;
489	>	value += angMom[j] * angMom[j] / I(j, j) +
490	>	angMom[k] * angMom[k] / I(k, k);
491		} else {
492	<	value += angMom[0]*angMom[0]/I(0, 0)
493	<	+ angMom[1]*angMom[1]/I(1, 1)
494	<	+ angMom[2]*angMom[2]/I(2, 2);
492	>	value += angMom[0]*angMom[0]/I(0, 0)
493	>	+ angMom[1]*angMom[1]/I(1, 1)
494	>	+ angMom[2]*angMom[2]/I(2, 2);
495		}
496	<	}
497	<	value = value * 0.5 / OOPSEConstant::energyConvert;
496	>	} //angular momenta exchange enabled
497	>	//energyConvert temporarily disabled
498	>	//make kineticExchange_ comparable between swap & scale
499	>	//value = value * 0.5 / PhysicalConstants::energyConvert;
500	>	value *= 0.5;
501		break;
502		case rnemdPx :
503		value = mass * vel[0];
#	Line 202 | Line 508 | namespace oopse {
508		case rnemdPz :
509		value = mass * vel[2];
510		break;
205	–	case rnemdUnknown :
511		default :
512		break;
513		}
514
515	<	if (binNo == 0) {
515	>	if (inA == 0) {
516		if (!min_found) {
517		min_val = value;
518		min_sd = sd;
#	Line 218 | Line 523 | namespace oopse {
523		min_sd = sd;
524		}
525		}
526	<	} else {
526	>	} else {
527		if (!max_found) {
528		max_val = value;
529		max_sd = sd;
#	Line 232 | Line 537 | namespace oopse {
537		}
538		}
539		}
540	<
540	>
541		#ifdef IS_MPI
542		int nProc, worldRank;
543	<
543	>
544		nProc = MPI::COMM_WORLD.Get_size();
545		worldRank = MPI::COMM_WORLD.Get_rank();
546
#	Line 243 | Line 548 | namespace oopse {
548		bool my_max_found = max_found;
549
550		// Even if we didn't find a minimum, did someone else?
551	<	MPI::COMM_WORLD.Allreduce(&my_min_found, &min_found,
247	<	1, MPI::BOOL, MPI::LAND);
248	<
551	>	MPI::COMM_WORLD.Allreduce(&my_min_found, &min_found, 1, MPI::BOOL, MPI::LOR);
552		// Even if we didn't find a maximum, did someone else?
553	<	MPI::COMM_WORLD.Allreduce(&my_max_found, &max_found,
554	<	1, MPI::BOOL, MPI::LAND);
555	<
556	<	struct {
557	<	RealType val;
558	<	int rank;
559	<	} max_vals, min_vals;
560	<
561	<	if (min_found) {
562	<	if (my_min_found)
553	>	MPI::COMM_WORLD.Allreduce(&my_max_found, &max_found, 1, MPI::BOOL, MPI::LOR);
554	>	#endif
555	>
556	>	if (max_found && min_found) {
557	>
558	>	#ifdef IS_MPI
559	>	struct {
560	>	RealType val;
561	>	int rank;
562	>	} max_vals, min_vals;
563	>
564	>	if (my_min_found) {
565		min_vals.val = min_val;
566	<	else
566	>	} else {
567		min_vals.val = HONKING_LARGE_VALUE;
568	<
568	>	}
569		min_vals.rank = worldRank;
570
571		// Who had the minimum?
572		MPI::COMM_WORLD.Allreduce(&min_vals, &min_vals,
573		1, MPI::REALTYPE_INT, MPI::MINLOC);
574		min_val = min_vals.val;
270	–	}
575
576	<	if (max_found) {
273	<	if (my_max_found)
576	>	if (my_max_found) {
577		max_vals.val = max_val;
578	<	else
578	>	} else {
579		max_vals.val = -HONKING_LARGE_VALUE;
580	<
580	>	}
581		max_vals.rank = worldRank;
582
583		// Who had the maximum?
584		MPI::COMM_WORLD.Allreduce(&max_vals, &max_vals,
585		1, MPI::REALTYPE_INT, MPI::MAXLOC);
586		max_val = max_vals.val;
284	–	}
587		#endif
588	<
589	<	if (max_found && min_found) {
590	<	if (min_val< max_val) {
289	<
588	>
589	>	if (min_val < max_val) {
590	>
591		#ifdef IS_MPI
592		if (max_vals.rank == worldRank && min_vals.rank == worldRank) {
593		// I have both maximum and minimum, so proceed like a single
594		// processor version:
595		#endif
596	<	// objects to be swapped: velocity & angular velocity
596	>
597		Vector3d min_vel = min_sd->getVel();
598		Vector3d max_vel = max_sd->getVel();
599		RealType temp_vel;
600
601	<	switch(rnemdType_) {
602	<	case rnemdKinetic :
601	>	switch(rnemdFluxType_) {
602	>	case rnemdKE :
603		min_sd->setVel(max_vel);
604		max_sd->setVel(min_vel);
605	<	if (min_sd->isDirectional() && max_sd->isDirectional()) {
605	>	if (min_sd->isDirectional() && max_sd->isDirectional()) {
606		Vector3d min_angMom = min_sd->getJ();
607		Vector3d max_angMom = max_sd->getJ();
608		min_sd->setJ(max_angMom);
609		max_sd->setJ(min_angMom);
610	<	}
610	>	}//angular momenta exchange enabled
611	>	//assumes same rigid body identity
612		break;
613		case rnemdPx :
614		temp_vel = min_vel.x();
#	Line 329 | Line 631 | namespace oopse {
631		min_sd->setVel(min_vel);
632		max_sd->setVel(max_vel);
633		break;
332	–	case rnemdUnknown :
634		default :
635		break;
636		}
637	+
638		#ifdef IS_MPI
639		// the rest of the cases only apply in parallel simulations:
640		} else if (max_vals.rank == worldRank) {
#	Line 348 | Line 650 | namespace oopse {
650		min_vel.getArrayPointer(), 3, MPI::REALTYPE,
651		min_vals.rank, 0, status);
652
653	<	switch(rnemdType_) {
654	<	case rnemdKinetic :
653	>	switch(rnemdFluxType_) {
654	>	case rnemdKE :
655		max_sd->setVel(min_vel);
656	<
656	>	//angular momenta exchange enabled
657		if (max_sd->isDirectional()) {
658		Vector3d min_angMom;
659		Vector3d max_angMom = max_sd->getJ();
660	<
660	>
661		// point-to-point swap of the angular momentum vector
662		MPI::COMM_WORLD.Sendrecv(max_angMom.getArrayPointer(), 3,
663		MPI::REALTYPE, min_vals.rank, 1,
664		min_angMom.getArrayPointer(), 3,
665		MPI::REALTYPE, min_vals.rank, 1,
666		status);
667	<
667	>
668		max_sd->setJ(min_angMom);
669	<	}
669	>	}
670		break;
671		case rnemdPx :
672		max_vel.x() = min_vel.x();
#	Line 378 | Line 680 | namespace oopse {
680		max_vel.z() = min_vel.z();
681		max_sd->setVel(max_vel);
682		break;
381	–	case rnemdUnknown :
683		default :
684		break;
685		}
#	Line 395 | Line 696 | namespace oopse {
696		max_vel.getArrayPointer(), 3, MPI::REALTYPE,
697		max_vals.rank, 0, status);
698
699	<	switch(rnemdType_) {
700	<	case rnemdKinetic :
699	>	switch(rnemdFluxType_) {
700	>	case rnemdKE :
701		min_sd->setVel(max_vel);
702	<
702	>	//angular momenta exchange enabled
703		if (min_sd->isDirectional()) {
704		Vector3d min_angMom = min_sd->getJ();
705		Vector3d max_angMom;
706	<
706	>
707		// point-to-point swap of the angular momentum vector
708		MPI::COMM_WORLD.Sendrecv(min_angMom.getArrayPointer(), 3,
709		MPI::REALTYPE, max_vals.rank, 1,
710		max_angMom.getArrayPointer(), 3,
711		MPI::REALTYPE, max_vals.rank, 1,
712		status);
713	<
713	>
714		min_sd->setJ(max_angMom);
715		}
716		break;
#	Line 425 | Line 726 | namespace oopse {
726		min_vel.z() = max_vel.z();
727		min_sd->setVel(min_vel);
728		break;
428	–	case rnemdUnknown :
729		default :
730		break;
731		}
732		}
733		#endif
734	<	exchangeSum_ += max_val - min_val;
735	<	} else {
736	<	std::cerr << "exchange NOT performed.\nmin_val > max_val.\n";
734	>
735	>	switch(rnemdFluxType_) {
736	>	case rnemdKE:
737	>	cerr << "KE\n";
738	>	kineticExchange_ += max_val - min_val;
739	>	break;
740	>	case rnemdPx:
741	>	momentumExchange_.x() += max_val - min_val;
742	>	break;
743	>	case rnemdPy:
744	>	momentumExchange_.y() += max_val - min_val;
745	>	break;
746	>	case rnemdPz:
747	>	momentumExchange_.z() += max_val - min_val;
748	>	break;
749	>	default:
750	>	cerr << "default\n";
751	>	break;
752	>	}
753	>	} else {
754	>	sprintf(painCave.errMsg,
755	>	"RNEMD::doSwap exchange NOT performed because min_val > max_val\n");
756	>	painCave.isFatal = 0;
757	>	painCave.severity = OPENMD_INFO;
758	>	simError();
759	>	failTrialCount_++;
760		}
761		} else {
762	<	std::cerr << "exchange NOT performed.\none of the two slabs empty.\n";
763	<	}
764	<
762	>	sprintf(painCave.errMsg,
763	>	"RNEMD::doSwap exchange NOT performed because selected object\n"
764	>	"\twas not present in at least one of the two slabs.\n");
765	>	painCave.isFatal = 0;
766	>	painCave.severity = OPENMD_INFO;
767	>	simError();
768	>	failTrialCount_++;
769	>	}
770		}
771
772	<	void RNEMD::getStatus() {
773	<
772	>	void RNEMD::doNIVS() {
773	>	if (!doRNEMD_) return;
774		Snapshot* currentSnap_ = info_->getSnapshotManager()->getCurrentSnapshot();
775		Mat3x3d hmat = currentSnap_->getHmat();
448	–	Stats& stat = currentSnap_->statData;
449	–	RealType time = currentSnap_->getTime();
776
451	–	stat[Stats::RNEMD_SWAP_TOTAL] = exchangeSum_;
452	–
777		seleMan_.setSelectionSet(evaluator_.evaluate());
778
779		int selei;
780		StuntDouble* sd;
781		int idx;
782
783	<	std::vector<RealType> valueHist(nBins_, 0.0); // keeps track of what's
460	<	// being averaged
461	<	std::vector<int> valueCount(nBins_, 0);       // keeps track of the
462	<	// number of degrees of
463	<	// freedom being averaged
783	>	vector<StuntDouble*> hotBin, coldBin;
784
785	+	RealType Phx = 0.0;
786	+	RealType Phy = 0.0;
787	+	RealType Phz = 0.0;
788	+	RealType Khx = 0.0;
789	+	RealType Khy = 0.0;
790	+	RealType Khz = 0.0;
791	+	RealType Khw = 0.0;
792	+	RealType Pcx = 0.0;
793	+	RealType Pcy = 0.0;
794	+	RealType Pcz = 0.0;
795	+	RealType Kcx = 0.0;
796	+	RealType Kcy = 0.0;
797	+	RealType Kcz = 0.0;
798	+	RealType Kcw = 0.0;
799	+
800		for (sd = seleMan_.beginSelected(selei); sd != NULL;
801		sd = seleMan_.nextSelected(selei)) {
802	<
802	>
803		idx = sd->getLocalIndex();
804	<
804	>
805		Vector3d pos = sd->getPos();
806
807		// wrap the stuntdouble's position back into the box:
808	<
808	>
809		if (usePeriodicBoundaryConditions_)
810		currentSnap_->wrapVector(pos);
811	<
811	>
812		// which bin is this stuntdouble in?
813	<	// wrapped positions are in the range [-0.5*hmat(2,2), +0.5*hmat(2,2)]
814	<
480	<	int binNo = int(nBins_ * (pos.z() / hmat(2,2) + 0.5)) % nBins_;
481	<
482	<	RealType mass = sd->getMass();
483	<	Vector3d vel = sd->getVel();
484	<	RealType value;
813	>	bool inA = inSlabA(pos);
814	>	bool inB = inSlabB(pos);
815
816	<	switch(rnemdType_) {
817	<	case rnemdKinetic :
818	<
819	<	value = mass * (vel[0]*vel[0] + vel[1]*vel[1] +
820	<	vel[2]*vel[2]);
821	<
822	<	valueCount[binNo] += 3;
823	<	if (sd->isDirectional()) {
824	<	Vector3d angMom = sd->getJ();
825	<	Mat3x3d I = sd->getI();
826	<
827	<	if (sd->isLinear()) {
828	<	int i = sd->linearAxis();
829	<	int j = (i + 1) % 3;
830	<	int k = (i + 2) % 3;
831	<	value += angMom[j] * angMom[j] / I(j, j) +
832	<	angMom[k] * angMom[k] / I(k, k);
816	>	if (inA || inB) {
817	>
818	>	RealType mass = sd->getMass();
819	>	Vector3d vel = sd->getVel();
820	>
821	>	if (inA) {
822	>	hotBin.push_back(sd);
823	>	Phx += mass * vel.x();
824	>	Phy += mass * vel.y();
825	>	Phz += mass * vel.z();
826	>	Khx += mass * vel.x() * vel.x();
827	>	Khy += mass * vel.y() * vel.y();
828	>	Khz += mass * vel.z() * vel.z();
829	>	if (sd->isDirectional()) {
830	>	Vector3d angMom = sd->getJ();
831	>	Mat3x3d I = sd->getI();
832	>	if (sd->isLinear()) {
833	>	int i = sd->linearAxis();
834	>	int j = (i + 1) % 3;
835	>	int k = (i + 2) % 3;
836	>	Khw += angMom[j] * angMom[j] / I(j, j) +
837	>	angMom[k] * angMom[k] / I(k, k);
838	>	} else {
839	>	Khw += angMom[0]*angMom[0]/I(0, 0)
840	>	+ angMom[1]*angMom[1]/I(1, 1)
841	>	+ angMom[2]*angMom[2]/I(2, 2);
842	>	}
843	>	}
844	>	} else {
845	>	coldBin.push_back(sd);
846	>	Pcx += mass * vel.x();
847	>	Pcy += mass * vel.y();
848	>	Pcz += mass * vel.z();
849	>	Kcx += mass * vel.x() * vel.x();
850	>	Kcy += mass * vel.y() * vel.y();
851	>	Kcz += mass * vel.z() * vel.z();
852	>	if (sd->isDirectional()) {
853	>	Vector3d angMom = sd->getJ();
854	>	Mat3x3d I = sd->getI();
855	>	if (sd->isLinear()) {
856	>	int i = sd->linearAxis();
857	>	int j = (i + 1) % 3;
858	>	int k = (i + 2) % 3;
859	>	Kcw += angMom[j] * angMom[j] / I(j, j) +
860	>	angMom[k] * angMom[k] / I(k, k);
861	>	} else {
862	>	Kcw += angMom[0]*angMom[0]/I(0, 0)
863	>	+ angMom[1]*angMom[1]/I(1, 1)
864	>	+ angMom[2]*angMom[2]/I(2, 2);
865	>	}
866	>	}
867	>	}
868	>	}
869	>	}
870	>
871	>	Khx *= 0.5;
872	>	Khy *= 0.5;
873	>	Khz *= 0.5;
874	>	Khw *= 0.5;
875	>	Kcx *= 0.5;
876	>	Kcy *= 0.5;
877	>	Kcz *= 0.5;
878	>	Kcw *= 0.5;
879
880	<	valueCount[binNo] +=2;
880	>	#ifdef IS_MPI
881	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Phx, 1, MPI::REALTYPE, MPI::SUM);
882	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Phy, 1, MPI::REALTYPE, MPI::SUM);
883	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Phz, 1, MPI::REALTYPE, MPI::SUM);
884	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Pcx, 1, MPI::REALTYPE, MPI::SUM);
885	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Pcy, 1, MPI::REALTYPE, MPI::SUM);
886	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Pcz, 1, MPI::REALTYPE, MPI::SUM);
887
888	<	} else {
889	<	value += angMom[0]*angMom[0]/I(0, 0)
890	<	+ angMom[1]*angMom[1]/I(1, 1)
891	<	+ angMom[2]*angMom[2]/I(2, 2);
892	<	valueCount[binNo] +=3;
888	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Khx, 1, MPI::REALTYPE, MPI::SUM);
889	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Khy, 1, MPI::REALTYPE, MPI::SUM);
890	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Khz, 1, MPI::REALTYPE, MPI::SUM);
891	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Khw, 1, MPI::REALTYPE, MPI::SUM);
892	>
893	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Kcx, 1, MPI::REALTYPE, MPI::SUM);
894	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Kcy, 1, MPI::REALTYPE, MPI::SUM);
895	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Kcz, 1, MPI::REALTYPE, MPI::SUM);
896	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Kcw, 1, MPI::REALTYPE, MPI::SUM);
897	>	#endif
898	>
899	>	//solve coldBin coeff's first
900	>	RealType px = Pcx / Phx;
901	>	RealType py = Pcy / Phy;
902	>	RealType pz = Pcz / Phz;
903	>	RealType c, x, y, z;
904	>	bool successfulScale = false;
905	>	if ((rnemdFluxType_ == rnemdFullKE) ||
906	>	(rnemdFluxType_ == rnemdRotKE)) {
907	>	//may need sanity check Khw & Kcw > 0
908	>
909	>	if (rnemdFluxType_ == rnemdFullKE) {
910	>	c = 1.0 - kineticTarget_ / (Kcx + Kcy + Kcz + Kcw);
911	>	} else {
912	>	c = 1.0 - kineticTarget_ / Kcw;
913	>	}
914	>
915	>	if ((c > 0.81) && (c < 1.21)) {//restrict scaling coefficients
916	>	c = sqrt(c);
917	>	//std::cerr << "cold slab scaling coefficient: " << c << endl;
918	>	//now convert to hotBin coefficient
919	>	RealType w = 0.0;
920	>	if (rnemdFluxType_ ==  rnemdFullKE) {
921	>	x = 1.0 + px * (1.0 - c);
922	>	y = 1.0 + py * (1.0 - c);
923	>	z = 1.0 + pz * (1.0 - c);
924	>	/* more complicated way
925	>	w = 1.0 + (Kcw - Kcw * c * c - (c * c * (Kcx + Kcy + Kcz
926	>	+ Khx * px * px + Khy * py * py + Khz * pz * pz)
927	>	- 2.0 * c * (Khx * px * (1.0 + px) + Khy * py * (1.0 + py)
928	>	+ Khz * pz * (1.0 + pz)) + Khx * px * (2.0 + px)
929	>	+ Khy * py * (2.0 + py) + Khz * pz * (2.0 + pz)
930	>	- Kcx - Kcy - Kcz)) / Khw; the following is simpler
931	>	*/
932	>	if ((fabs(x - 1.0) < 0.1) && (fabs(y - 1.0) < 0.1) &&
933	>	(fabs(z - 1.0) < 0.1)) {
934	>	w = 1.0 + (kineticTarget_
935	>	+ Khx * (1.0 - x * x) + Khy * (1.0 - y * y)
936	>	+ Khz * (1.0 - z * z)) / Khw;
937	>	}//no need to calculate w if x, y or z is out of range
938	>	} else {
939	>	w = 1.0 + kineticTarget_ / Khw;
940	>	}
941	>	if ((w > 0.81) && (w < 1.21)) {//restrict scaling coefficients
942	>	//if w is in the right range, so should be x, y, z.
943	>	vector<StuntDouble*>::iterator sdi;
944	>	Vector3d vel;
945	>	for (sdi = coldBin.begin(); sdi != coldBin.end(); sdi++) {
946	>	if (rnemdFluxType_ == rnemdFullKE) {
947	>	vel = (*sdi)->getVel() * c;
948	>	(*sdi)->setVel(vel);
949	>	}
950	>	if ((*sdi)->isDirectional()) {
951	>	Vector3d angMom = (*sdi)->getJ() * c;
952	>	(*sdi)->setJ(angMom);
953	>	}
954		}
955	+	w = sqrt(w);
956	+	// std::cerr << "xh= " << x << "\tyh= " << y << "\tzh= " << z
957	+	//           << "\twh= " << w << endl;
958	+	for (sdi = hotBin.begin(); sdi != hotBin.end(); sdi++) {
959	+	if (rnemdFluxType_ == rnemdFullKE) {
960	+	vel = (*sdi)->getVel();
961	+	vel.x() *= x;
962	+	vel.y() *= y;
963	+	vel.z() *= z;
964	+	(*sdi)->setVel(vel);
965	+	}
966	+	if ((*sdi)->isDirectional()) {
967	+	Vector3d angMom = (*sdi)->getJ() * w;
968	+	(*sdi)->setJ(angMom);
969	+	}
970	+	}
971	+	successfulScale = true;
972	+	kineticExchange_ += kineticTarget_;
973		}
974	<	value = value / OOPSEConstant::energyConvert / OOPSEConstant::kb;
975	<
976	<	break;
977	<	case rnemdPx :
978	<	value = mass * vel[0];
979	<	valueCount[binNo]++;
974	>	}
975	>	} else {
976	>	RealType a000, a110, c0, a001, a111, b01, b11, c1;
977	>	switch(rnemdFluxType_) {
978	>	case rnemdKE :
979	>	/* used hotBin coeff's & only scale x & y dimensions
980	>	RealType px = Phx / Pcx;
981	>	RealType py = Phy / Pcy;
982	>	a110 = Khy;
983	>	c0 = - Khx - Khy - kineticTarget_;
984	>	a000 = Khx;
985	>	a111 = Kcy * py * py;
986	>	b11 = -2.0 * Kcy * py * (1.0 + py);
987	>	c1 = Kcy * py * (2.0 + py) + Kcx * px * ( 2.0 + px) + kineticTarget_;
988	>	b01 = -2.0 * Kcx * px * (1.0 + px);
989	>	a001 = Kcx * px * px;
990	>	*/
991	>	//scale all three dimensions, let c_x = c_y
992	>	a000 = Kcx + Kcy;
993	>	a110 = Kcz;
994	>	c0 = kineticTarget_ - Kcx - Kcy - Kcz;
995	>	a001 = Khx * px * px + Khy * py * py;
996	>	a111 = Khz * pz * pz;
997	>	b01 = -2.0 * (Khx * px * (1.0 + px) + Khy * py * (1.0 + py));
998	>	b11 = -2.0 * Khz * pz * (1.0 + pz);
999	>	c1 = Khx * px * (2.0 + px) + Khy * py * (2.0 + py)
1000	>	+ Khz * pz * (2.0 + pz) - kineticTarget_;
1001		break;
1002	+	case rnemdPx :
1003	+	c = 1 - momentumTarget_.x() / Pcx;
1004	+	a000 = Kcy;
1005	+	a110 = Kcz;
1006	+	c0 = Kcx * c * c - Kcx - Kcy - Kcz;
1007	+	a001 = py * py * Khy;
1008	+	a111 = pz * pz * Khz;
1009	+	b01 = -2.0 * Khy * py * (1.0 + py);
1010	+	b11 = -2.0 * Khz * pz * (1.0 + pz);
1011	+	c1 = Khy * py * (2.0 + py) + Khz * pz * (2.0 + pz)
1012	+	+ Khx * (fastpow(c * px - px - 1.0, 2) - 1.0);
1013	+	break;
1014		case rnemdPy :
1015	<	value = mass * vel[1];
1016	<	valueCount[binNo]++;
1015	>	c = 1 - momentumTarget_.y() / Pcy;
1016	>	a000 = Kcx;
1017	>	a110 = Kcz;
1018	>	c0 = Kcy * c * c - Kcx - Kcy - Kcz;
1019	>	a001 = px * px * Khx;
1020	>	a111 = pz * pz * Khz;
1021	>	b01 = -2.0 * Khx * px * (1.0 + px);
1022	>	b11 = -2.0 * Khz * pz * (1.0 + pz);
1023	>	c1 = Khx * px * (2.0 + px) + Khz * pz * (2.0 + pz)
1024	>	+ Khy * (fastpow(c * py - py - 1.0, 2) - 1.0);
1025		break;
1026	<	case rnemdPz :
1027	<	value = mass * vel[2];
1028	<	valueCount[binNo]++;
1026	>	case rnemdPz ://we don't really do this, do we?
1027	>	c = 1 - momentumTarget_.z() / Pcz;
1028	>	a000 = Kcx;
1029	>	a110 = Kcy;
1030	>	c0 = Kcz * c * c - Kcx - Kcy - Kcz;
1031	>	a001 = px * px * Khx;
1032	>	a111 = py * py * Khy;
1033	>	b01 = -2.0 * Khx * px * (1.0 + px);
1034	>	b11 = -2.0 * Khy * py * (1.0 + py);
1035	>	c1 = Khx * px * (2.0 + px) + Khy * py * (2.0 + py)
1036	>	+ Khz * (fastpow(c * pz - pz - 1.0, 2) - 1.0);
1037		break;
528	–	case rnemdUnknown :
1038		default :
1039		break;
1040		}
1041	<	valueHist[binNo] += value;
1041	>
1042	>	RealType v1 = a000 * a111 - a001 * a110;
1043	>	RealType v2 = a000 * b01;
1044	>	RealType v3 = a000 * b11;
1045	>	RealType v4 = a000 * c1 - a001 * c0;
1046	>	RealType v8 = a110 * b01;
1047	>	RealType v10 = - b01 * c0;
1048	>
1049	>	RealType u0 = v2 * v10 - v4 * v4;
1050	>	RealType u1 = -2.0 * v3 * v4;
1051	>	RealType u2 = -v2 * v8 - v3 * v3 - 2.0 * v1 * v4;
1052	>	RealType u3 = -2.0 * v1 * v3;
1053	>	RealType u4 = - v1 * v1;
1054	>	//rescale coefficients
1055	>	RealType maxAbs = fabs(u0);
1056	>	if (maxAbs < fabs(u1)) maxAbs = fabs(u1);
1057	>	if (maxAbs < fabs(u2)) maxAbs = fabs(u2);
1058	>	if (maxAbs < fabs(u3)) maxAbs = fabs(u3);
1059	>	if (maxAbs < fabs(u4)) maxAbs = fabs(u4);
1060	>	u0 /= maxAbs;
1061	>	u1 /= maxAbs;
1062	>	u2 /= maxAbs;
1063	>	u3 /= maxAbs;
1064	>	u4 /= maxAbs;
1065	>	//max_element(start, end) is also available.
1066	>	Polynomial<RealType> poly; //same as DoublePolynomial poly;
1067	>	poly.setCoefficient(4, u4);
1068	>	poly.setCoefficient(3, u3);
1069	>	poly.setCoefficient(2, u2);
1070	>	poly.setCoefficient(1, u1);
1071	>	poly.setCoefficient(0, u0);
1072	>	vector<RealType> realRoots = poly.FindRealRoots();
1073	>
1074	>	vector<RealType>::iterator ri;
1075	>	RealType r1, r2, alpha0;
1076	>	vector<pair<RealType,RealType> > rps;
1077	>	for (ri = realRoots.begin(); ri !=realRoots.end(); ri++) {
1078	>	r2 = *ri;
1079	>	//check if FindRealRoots() give the right answer
1080	>	if ( fabs(u0 + r2 * (u1 + r2 * (u2 + r2 * (u3 + r2 * u4)))) > 1e-6 ) {
1081	>	sprintf(painCave.errMsg,
1082	>	"RNEMD Warning: polynomial solve seems to have an error!");
1083	>	painCave.isFatal = 0;
1084	>	simError();
1085	>	failRootCount_++;
1086	>	}
1087	>	//might not be useful w/o rescaling coefficients
1088	>	alpha0 = -c0 - a110 * r2 * r2;
1089	>	if (alpha0 >= 0.0) {
1090	>	r1 = sqrt(alpha0 / a000);
1091	>	if (fabs(c1 + r1 * (b01 + r1 * a001) + r2 * (b11 + r2 * a111))
1092	>	< 1e-6)
1093	>	{ rps.push_back(make_pair(r1, r2)); }
1094	>	if (r1 > 1e-6) { //r1 non-negative
1095	>	r1 = -r1;
1096	>	if (fabs(c1 + r1 * (b01 + r1 * a001) + r2 * (b11 + r2 * a111))
1097	>	< 1e-6)
1098	>	{ rps.push_back(make_pair(r1, r2)); }
1099	>	}
1100	>	}
1101	>	}
1102	>	// Consider combining together the solving pair part w/ the searching
1103	>	// best solution part so that we don't need the pairs vector
1104	>	if (!rps.empty()) {
1105	>	RealType smallestDiff = HONKING_LARGE_VALUE;
1106	>	RealType diff;
1107	>	pair<RealType,RealType> bestPair = make_pair(1.0, 1.0);
1108	>	vector<pair<RealType,RealType> >::iterator rpi;
1109	>	for (rpi = rps.begin(); rpi != rps.end(); rpi++) {
1110	>	r1 = (*rpi).first;
1111	>	r2 = (*rpi).second;
1112	>	switch(rnemdFluxType_) {
1113	>	case rnemdKE :
1114	>	diff = fastpow(1.0 - r1, 2) + fastpow(1.0 - r2, 2)
1115	>	+ fastpow(r1 * r1 / r2 / r2 - Kcz/Kcx, 2)
1116	>	+ fastpow(r1 * r1 / r2 / r2 - Kcz/Kcy, 2);
1117	>	break;
1118	>	case rnemdPx :
1119	>	diff = fastpow(1.0 - r1, 2) + fastpow(1.0 - r2, 2)
1120	>	+ fastpow(r1 * r1 / r2 / r2 - Kcz/Kcy, 2);
1121	>	break;
1122	>	case rnemdPy :
1123	>	diff = fastpow(1.0 - r1, 2) + fastpow(1.0 - r2, 2)
1124	>	+ fastpow(r1 * r1 / r2 / r2 - Kcz/Kcx, 2);
1125	>	break;
1126	>	case rnemdPz :
1127	>	diff = fastpow(1.0 - r1, 2) + fastpow(1.0 - r2, 2)
1128	>	+ fastpow(r1 * r1 / r2 / r2 - Kcy/Kcx, 2);
1129	>	default :
1130	>	break;
1131	>	}
1132	>	if (diff < smallestDiff) {
1133	>	smallestDiff = diff;
1134	>	bestPair = *rpi;
1135	>	}
1136	>	}
1137	>	#ifdef IS_MPI
1138	>	if (worldRank == 0) {
1139	>	#endif
1140	>	// sprintf(painCave.errMsg,
1141	>	//         "RNEMD: roots r1= %lf\tr2 = %lf\n",
1142	>	//         bestPair.first, bestPair.second);
1143	>	// painCave.isFatal = 0;
1144	>	// painCave.severity = OPENMD_INFO;
1145	>	// simError();
1146	>	#ifdef IS_MPI
1147	>	}
1148	>	#endif
1149	>
1150	>	switch(rnemdFluxType_) {
1151	>	case rnemdKE :
1152	>	x = bestPair.first;
1153	>	y = bestPair.first;
1154	>	z = bestPair.second;
1155	>	break;
1156	>	case rnemdPx :
1157	>	x = c;
1158	>	y = bestPair.first;
1159	>	z = bestPair.second;
1160	>	break;
1161	>	case rnemdPy :
1162	>	x = bestPair.first;
1163	>	y = c;
1164	>	z = bestPair.second;
1165	>	break;
1166	>	case rnemdPz :
1167	>	x = bestPair.first;
1168	>	y = bestPair.second;
1169	>	z = c;
1170	>	break;
1171	>	default :
1172	>	break;
1173	>	}
1174	>	vector<StuntDouble*>::iterator sdi;
1175	>	Vector3d vel;
1176	>	for (sdi = coldBin.begin(); sdi != coldBin.end(); sdi++) {
1177	>	vel = (*sdi)->getVel();
1178	>	vel.x() *= x;
1179	>	vel.y() *= y;
1180	>	vel.z() *= z;
1181	>	(*sdi)->setVel(vel);
1182	>	}
1183	>	//convert to hotBin coefficient
1184	>	x = 1.0 + px * (1.0 - x);
1185	>	y = 1.0 + py * (1.0 - y);
1186	>	z = 1.0 + pz * (1.0 - z);
1187	>	for (sdi = hotBin.begin(); sdi != hotBin.end(); sdi++) {
1188	>	vel = (*sdi)->getVel();
1189	>	vel.x() *= x;
1190	>	vel.y() *= y;
1191	>	vel.z() *= z;
1192	>	(*sdi)->setVel(vel);
1193	>	}
1194	>	successfulScale = true;
1195	>	switch(rnemdFluxType_) {
1196	>	case rnemdKE :
1197	>	kineticExchange_ += kineticTarget_;
1198	>	break;
1199	>	case rnemdPx :
1200	>	case rnemdPy :
1201	>	case rnemdPz :
1202	>	momentumExchange_ += momentumTarget_;
1203	>	break;
1204	>	default :
1205	>	break;
1206	>	}
1207	>	}
1208		}
1209	+	if (successfulScale != true) {
1210	+	sprintf(painCave.errMsg,
1211	+	"RNEMD::doNIVS exchange NOT performed - roots that solve\n"
1212	+	"\tthe constraint equations may not exist or there may be\n"
1213	+	"\tno selected objects in one or both slabs.\n");
1214	+	painCave.isFatal = 0;
1215	+	painCave.severity = OPENMD_INFO;
1216	+	simError();
1217	+	failTrialCount_++;
1218	+	}
1219	+	}
1220
1221	<	#ifdef IS_MPI
1221	>	void RNEMD::doVSS() {
1222	>	if (!doRNEMD_) return;
1223	>	Snapshot* currentSnap_ = info_->getSnapshotManager()->getCurrentSnapshot();
1224	>	RealType time = currentSnap_->getTime();
1225	>	Mat3x3d hmat = currentSnap_->getHmat();
1226
1227	<	// all processors have the same number of bins, and STL vectors pack their
538	<	// arrays, so in theory, this should be safe:
1227	>	seleMan_.setSelectionSet(evaluator_.evaluate());
1228
1229	<	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &valueHist[0],
1230	<	nBins_, MPI::REALTYPE, MPI::SUM);
1231	<	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &valueCount[0],
543	<	nBins_, MPI::INT, MPI::SUM);
1229	>	int selei;
1230	>	StuntDouble* sd;
1231	>	int idx;
1232
1233	+	vector<StuntDouble*> hotBin, coldBin;
1234	+
1235	+	Vector3d Ph(V3Zero);
1236	+	RealType Mh = 0.0;
1237	+	RealType Kh = 0.0;
1238	+	Vector3d Pc(V3Zero);
1239	+	RealType Mc = 0.0;
1240	+	RealType Kc = 0.0;
1241	+
1242	+
1243	+	for (sd = seleMan_.beginSelected(selei); sd != NULL;
1244	+	sd = seleMan_.nextSelected(selei)) {
1245	+
1246	+	idx = sd->getLocalIndex();
1247	+
1248	+	Vector3d pos = sd->getPos();
1249	+
1250	+	// wrap the stuntdouble's position back into the box:
1251	+
1252	+	if (usePeriodicBoundaryConditions_)
1253	+	currentSnap_->wrapVector(pos);
1254	+
1255	+	// which bin is this stuntdouble in?
1256	+	bool inA = inSlabA(pos);
1257	+	bool inB = inSlabB(pos);
1258	+
1259	+	if (inA || inB) {
1260	+
1261	+	RealType mass = sd->getMass();
1262	+	Vector3d vel = sd->getVel();
1263	+
1264	+	if (inA) {
1265	+	hotBin.push_back(sd);
1266	+	//std::cerr << "before, velocity = " << vel << endl;
1267	+	Ph += mass * vel;
1268	+	//std::cerr << "after, velocity = " << vel << endl;
1269	+	Mh += mass;
1270	+	Kh += mass * vel.lengthSquare();
1271	+	if (rnemdFluxType_ == rnemdFullKE) {
1272	+	if (sd->isDirectional()) {
1273	+	Vector3d angMom = sd->getJ();
1274	+	Mat3x3d I = sd->getI();
1275	+	if (sd->isLinear()) {
1276	+	int i = sd->linearAxis();
1277	+	int j = (i + 1) % 3;
1278	+	int k = (i + 2) % 3;
1279	+	Kh += angMom[j] * angMom[j] / I(j, j) +
1280	+	angMom[k] * angMom[k] / I(k, k);
1281	+	} else {
1282	+	Kh += angMom[0] * angMom[0] / I(0, 0) +
1283	+	angMom[1] * angMom[1] / I(1, 1) +
1284	+	angMom[2] * angMom[2] / I(2, 2);
1285	+	}
1286	+	}
1287	+	}
1288	+	} else { //midBin_
1289	+	coldBin.push_back(sd);
1290	+	Pc += mass * vel;
1291	+	Mc += mass;
1292	+	Kc += mass * vel.lengthSquare();
1293	+	if (rnemdFluxType_ == rnemdFullKE) {
1294	+	if (sd->isDirectional()) {
1295	+	Vector3d angMom = sd->getJ();
1296	+	Mat3x3d I = sd->getI();
1297	+	if (sd->isLinear()) {
1298	+	int i = sd->linearAxis();
1299	+	int j = (i + 1) % 3;
1300	+	int k = (i + 2) % 3;
1301	+	Kc += angMom[j] * angMom[j] / I(j, j) +
1302	+	angMom[k] * angMom[k] / I(k, k);
1303	+	} else {
1304	+	Kc += angMom[0] * angMom[0] / I(0, 0) +
1305	+	angMom[1] * angMom[1] / I(1, 1) +
1306	+	angMom[2] * angMom[2] / I(2, 2);
1307	+	}
1308	+	}
1309	+	}
1310	+	}
1311	+	}
1312	+	}
1313	+
1314	+	Kh *= 0.5;
1315	+	Kc *= 0.5;
1316	+
1317	+	// std::cerr << "Mh= " << Mh << "\tKh= " << Kh << "\tMc= " << Mc
1318	+	//        << "\tKc= " << Kc << endl;
1319	+	// std::cerr << "Ph= " << Ph << "\tPc= " << Pc << endl;
1320	+
1321	+	#ifdef IS_MPI
1322	+	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Ph[0], 3, MPI::REALTYPE, MPI::SUM);
1323	+	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Pc[0], 3, MPI::REALTYPE, MPI::SUM);
1324	+	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Mh, 1, MPI::REALTYPE, MPI::SUM);
1325	+	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Kh, 1, MPI::REALTYPE, MPI::SUM);
1326	+	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Mc, 1, MPI::REALTYPE, MPI::SUM);
1327	+	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Kc, 1, MPI::REALTYPE, MPI::SUM);
1328	+	#endif
1329	+
1330	+	bool successfulExchange = false;
1331	+	if ((Mh > 0.0) && (Mc > 0.0)) {//both slabs are not empty
1332	+	Vector3d vc = Pc / Mc;
1333	+	Vector3d ac = -momentumTarget_ / Mc + vc;
1334	+	Vector3d acrec = -momentumTarget_ / Mc;
1335	+	RealType cNumerator = Kc - kineticTarget_ - 0.5 * Mc * ac.lengthSquare();
1336	+	if (cNumerator > 0.0) {
1337	+	RealType cDenominator = Kc - 0.5 * Mc * vc.lengthSquare();
1338	+	if (cDenominator > 0.0) {
1339	+	RealType c = sqrt(cNumerator / cDenominator);
1340	+	if ((c > 0.9) && (c < 1.1)) {//restrict scaling coefficients
1341	+	Vector3d vh = Ph / Mh;
1342	+	Vector3d ah = momentumTarget_ / Mh + vh;
1343	+	Vector3d ahrec = momentumTarget_ / Mh;
1344	+	RealType hNumerator = Kh + kineticTarget_
1345	+	- 0.5 * Mh * ah.lengthSquare();
1346	+	if (hNumerator > 0.0) {
1347	+	RealType hDenominator = Kh - 0.5 * Mh * vh.lengthSquare();
1348	+	if (hDenominator > 0.0) {
1349	+	RealType h = sqrt(hNumerator / hDenominator);
1350	+	if ((h > 0.9) && (h < 1.1)) {
1351	+	// std::cerr << "cold slab scaling coefficient: " << c << "\n";
1352	+	// std::cerr << "hot slab scaling coefficient: " << h <<  "\n";
1353	+	vector<StuntDouble*>::iterator sdi;
1354	+	Vector3d vel;
1355	+	for (sdi = coldBin.begin(); sdi != coldBin.end(); sdi++) {
1356	+	//vel = (*sdi)->getVel();
1357	+	vel = ((*sdi)->getVel() - vc) * c + ac;
1358	+	(*sdi)->setVel(vel);
1359	+	if (rnemdFluxType_ == rnemdFullKE) {
1360	+	if ((*sdi)->isDirectional()) {
1361	+	Vector3d angMom = (*sdi)->getJ() * c;
1362	+	(*sdi)->setJ(angMom);
1363	+	}
1364	+	}
1365	+	}
1366	+	for (sdi = hotBin.begin(); sdi != hotBin.end(); sdi++) {
1367	+	//vel = (*sdi)->getVel();
1368	+	vel = ((*sdi)->getVel() - vh) * h + ah;
1369	+	(*sdi)->setVel(vel);
1370	+	if (rnemdFluxType_ == rnemdFullKE) {
1371	+	if ((*sdi)->isDirectional()) {
1372	+	Vector3d angMom = (*sdi)->getJ() * h;
1373	+	(*sdi)->setJ(angMom);
1374	+	}
1375	+	}
1376	+	}
1377	+	successfulExchange = true;
1378	+	kineticExchange_ += kineticTarget_;
1379	+	momentumExchange_ += momentumTarget_;
1380	+	}
1381	+	}
1382	+	}
1383	+	}
1384	+	}
1385	+	}
1386	+	}
1387	+	if (successfulExchange != true) {
1388	+	sprintf(painCave.errMsg,
1389	+	"RNEMD::doVSS exchange NOT performed - roots that solve\n"
1390	+	"\tthe constraint equations may not exist or there may be\n"
1391	+	"\tno selected objects in one or both slabs.\n");
1392	+	painCave.isFatal = 0;
1393	+	painCave.severity = OPENMD_INFO;
1394	+	simError();
1395	+	failTrialCount_++;
1396	+	}
1397	+	}
1398	+
1399	+	void RNEMD::doRNEMD() {
1400	+	if (!doRNEMD_) return;
1401	+	trialCount_++;
1402	+	switch(rnemdMethod_) {
1403	+	case rnemdSwap:
1404	+	doSwap();
1405	+	break;
1406	+	case rnemdNIVS:
1407	+	doNIVS();
1408	+	break;
1409	+	case rnemdVSS:
1410	+	doVSS();
1411	+	break;
1412	+	case rnemdUnkownMethod:
1413	+	default :
1414	+	break;
1415	+	}
1416	+	}
1417	+
1418	+	void RNEMD::collectData() {
1419	+	if (!doRNEMD_) return;
1420	+	Snapshot* currentSnap_ = info_->getSnapshotManager()->getCurrentSnapshot();
1421	+	Mat3x3d hmat = currentSnap_->getHmat();
1422	+
1423	+	areaAccumulator_->add(currentSnap_->getXYarea());
1424	+
1425	+	seleMan_.setSelectionSet(evaluator_.evaluate());
1426	+
1427	+	int selei;
1428	+	StuntDouble* sd;
1429	+	int idx;
1430	+
1431	+	vector<RealType> binMass(nBins_, 0.0);
1432	+	vector<RealType> binPx(nBins_, 0.0);
1433	+	vector<RealType> binPy(nBins_, 0.0);
1434	+	vector<RealType> binPz(nBins_, 0.0);
1435	+	vector<RealType> binKE(nBins_, 0.0);
1436	+	vector<int> binDOF(nBins_, 0);
1437	+	vector<int> binCount(nBins_, 0);
1438	+
1439	+	// alternative approach, track all molecules instead of only those
1440	+	// selected for scaling/swapping:
1441	+	/*
1442	+	SimInfo::MoleculeIterator miter;
1443	+	vector<StuntDouble*>::iterator iiter;
1444	+	Molecule* mol;
1445	+	StuntDouble* sd;
1446	+	for (mol = info_->beginMolecule(miter); mol != NULL;
1447	+	mol = info_->nextMolecule(miter))
1448	+	sd is essentially sd
1449	+	for (sd = mol->beginIntegrableObject(iiter);
1450	+	sd != NULL;
1451	+	sd = mol->nextIntegrableObject(iiter))
1452	+	*/
1453	+	for (sd = seleMan_.beginSelected(selei); sd != NULL;
1454	+	sd = seleMan_.nextSelected(selei)) {
1455	+
1456	+	idx = sd->getLocalIndex();
1457	+
1458	+	Vector3d pos = sd->getPos();
1459	+
1460	+	// wrap the stuntdouble's position back into the box:
1461	+
1462	+	if (usePeriodicBoundaryConditions_)
1463	+	currentSnap_->wrapVector(pos);
1464	+
1465	+
1466	+	// which bin is this stuntdouble in?
1467	+	// wrapped positions are in the range [-0.5*hmat(2,2), +0.5*hmat(2,2)]
1468	+	// Shift molecules by half a box to have bins start at 0
1469	+	// The modulo operator is used to wrap the case when we are
1470	+	// beyond the end of the bins back to the beginning.
1471	+	int binNo = int(nBins_ * (pos.z() / hmat(2,2) + 0.5)) % nBins_;
1472	+
1473	+	RealType mass = sd->getMass();
1474	+	Vector3d vel = sd->getVel();
1475	+
1476	+	binCount[binNo]++;
1477	+	binMass[binNo] += mass;
1478	+	binPx[binNo] += mass*vel.x();
1479	+	binPy[binNo] += mass*vel.y();
1480	+	binPz[binNo] += mass*vel.z();
1481	+	binKE[binNo] += 0.5 * (mass * vel.lengthSquare());
1482	+	binDOF[binNo] += 3;
1483	+
1484	+	if (sd->isDirectional()) {
1485	+	Vector3d angMom = sd->getJ();
1486	+	Mat3x3d I = sd->getI();
1487	+	if (sd->isLinear()) {
1488	+	int i = sd->linearAxis();
1489	+	int j = (i + 1) % 3;
1490	+	int k = (i + 2) % 3;
1491	+	binKE[binNo] += 0.5 * (angMom[j] * angMom[j] / I(j, j) +
1492	+	angMom[k] * angMom[k] / I(k, k));
1493	+	binDOF[binNo] += 2;
1494	+	} else {
1495	+	binKE[binNo] += 0.5 * (angMom[0] * angMom[0] / I(0, 0) +
1496	+	angMom[1] * angMom[1] / I(1, 1) +
1497	+	angMom[2] * angMom[2] / I(2, 2));
1498	+	binDOF[binNo] += 3;
1499	+	}
1500	+	}
1501	+	}
1502	+
1503	+
1504	+	#ifdef IS_MPI
1505	+	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &binCount[0],
1506	+	nBins_, MPI::INT, MPI::SUM);
1507	+	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &binMass[0],
1508	+	nBins_, MPI::REALTYPE, MPI::SUM);
1509	+	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &binPx[0],
1510	+	nBins_, MPI::REALTYPE, MPI::SUM);
1511	+	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &binPy[0],
1512	+	nBins_, MPI::REALTYPE, MPI::SUM);
1513	+	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &binPz[0],
1514	+	nBins_, MPI::REALTYPE, MPI::SUM);
1515	+	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &binKE[0],
1516	+	nBins_, MPI::REALTYPE, MPI::SUM);
1517	+	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &binDOF[0],
1518	+	nBins_, MPI::INT, MPI::SUM);
1519	+	#endif
1520	+
1521	+	Vector3d vel;
1522	+	RealType den;
1523	+	RealType temp;
1524	+	RealType z;
1525	+	for (int i = 0; i < nBins_; i++) {
1526	+	z = (((RealType)i + 0.5) / (RealType)nBins_) * hmat(2,2);
1527	+	vel.x() = binPx[i] / binMass[i];
1528	+	vel.y() = binPy[i] / binMass[i];
1529	+	vel.z() = binPz[i] / binMass[i];
1530	+	den = binCount[i] * nBins_ / currentSnap_->getVolume();
1531	+	temp = 2.0 * binKE[i] / (binDOF[i] * PhysicalConstants::kb *
1532	+	PhysicalConstants::energyConvert);
1533	+
1534	+	for (unsigned int j = 0; j < outputMask_.size(); ++j) {
1535	+	if(outputMask_[j]) {
1536	+	switch(j) {
1537	+	case Z:
1538	+	(data_[j].accumulator[i])->add(z);
1539	+	break;
1540	+	case TEMPERATURE:
1541	+	data_[j].accumulator[i]->add(temp);
1542	+	break;
1543	+	case VELOCITY:
1544	+	dynamic_cast<VectorAccumulator *>(data_[j].accumulator[i])->add(vel);
1545	+	break;
1546	+	case DENSITY:
1547	+	data_[j].accumulator[i]->add(den);
1548	+	break;
1549	+	}
1550	+	}
1551	+	}
1552	+	}
1553	+	}
1554	+
1555	+	void RNEMD::getStarted() {
1556	+	if (!doRNEMD_) return;
1557	+	collectData();
1558	+	writeOutputFile();
1559	+	}
1560	+
1561	+	void RNEMD::parseOutputFileFormat(const std::string& format) {
1562	+	if (!doRNEMD_) return;
1563	+	StringTokenizer tokenizer(format, " ,;|\t\n\r");
1564	+
1565	+	while(tokenizer.hasMoreTokens()) {
1566	+	std::string token(tokenizer.nextToken());
1567	+	toUpper(token);
1568	+	OutputMapType::iterator i = outputMap_.find(token);
1569	+	if (i != outputMap_.end()) {
1570	+	outputMask_.set(i->second);
1571	+	} else {
1572	+	sprintf( painCave.errMsg,
1573	+	"RNEMD::parseOutputFileFormat: %s is not a recognized\n"
1574	+	"\toutputFileFormat keyword.\n", token.c_str() );
1575	+	painCave.isFatal = 0;
1576	+	painCave.severity = OPENMD_ERROR;
1577	+	simError();
1578	+	}
1579	+	}
1580	+	}
1581	+
1582	+	void RNEMD::writeOutputFile() {
1583	+	if (!doRNEMD_) return;
1584	+
1585	+	#ifdef IS_MPI
1586		// If we're the root node, should we print out the results
1587		int worldRank = MPI::COMM_WORLD.Get_rank();
1588		if (worldRank == 0) {
1589		#endif
1590	<
550	<	std::cout << time;
551	<	for (int j = 0; j < nBins_; j++)
552	<	std::cout << "\t" << valueHist[j] / (RealType)valueCount[j];
553	<	std::cout << "\n";
1590	>	rnemdFile_.open(rnemdFileName_.c_str(), std::ios::out | std::ios::trunc );
1591
1592	+	if( !rnemdFile_ ){
1593	+	sprintf( painCave.errMsg,
1594	+	"Could not open \"%s\" for RNEMD output.\n",
1595	+	rnemdFileName_.c_str());
1596	+	painCave.isFatal = 1;
1597	+	simError();
1598	+	}
1599	+
1600	+	Snapshot* currentSnap_ = info_->getSnapshotManager()->getCurrentSnapshot();
1601	+
1602	+	RealType time = currentSnap_->getTime();
1603	+	RealType avgArea;
1604	+	areaAccumulator_->getAverage(avgArea);
1605	+	RealType Jz = kineticExchange_ / (2.0 * time * avgArea);
1606	+	Vector3d JzP = momentumExchange_ / (2.0 * time * avgArea);
1607	+
1608	+	rnemdFile_ << "#######################################################\n";
1609	+	rnemdFile_ << "# RNEMD {\n";
1610	+
1611	+	map<string, RNEMDMethod>::iterator mi;
1612	+	for(mi = stringToMethod_.begin(); mi != stringToMethod_.end(); ++mi) {
1613	+	if ( (*mi).second == rnemdMethod_)
1614	+	rnemdFile_ << "#    exchangeMethod  = \"" << (*mi).first << "\";\n";
1615	+	}
1616	+	map<string, RNEMDFluxType>::iterator fi;
1617	+	for(fi = stringToFluxType_.begin(); fi != stringToFluxType_.end(); ++fi) {
1618	+	if ( (*fi).second == rnemdFluxType_)
1619	+	rnemdFile_ << "#    fluxType  = \"" << (*fi).first << "\";\n";
1620	+	}
1621	+
1622	+	rnemdFile_ << "#    exchangeTime = " << exchangeTime_ << ";\n";
1623	+
1624	+	rnemdFile_ << "#    objectSelection = \""
1625	+	<< rnemdObjectSelection_ << "\";\n";
1626	+	rnemdFile_ << "#    slabWidth = " << slabWidth_ << ";\n";
1627	+	rnemdFile_ << "#    slabAcenter = " << slabACenter_ << ";\n";
1628	+	rnemdFile_ << "#    slabBcenter = " << slabBCenter_ << ";\n";
1629	+	rnemdFile_ << "# }\n";
1630	+	rnemdFile_ << "#######################################################\n";
1631	+	rnemdFile_ << "# RNEMD report:\n";
1632	+	rnemdFile_ << "#     running time = " << time << " fs\n";
1633	+	rnemdFile_ << "#     target flux:\n";
1634	+	rnemdFile_ << "#         kinetic = " << kineticFlux_ << "\n";
1635	+	rnemdFile_ << "#         momentum = " << momentumFluxVector_ << "\n";
1636	+	rnemdFile_ << "#     target one-time exchanges:\n";
1637	+	rnemdFile_ << "#         kinetic = " << kineticTarget_ << "\n";
1638	+	rnemdFile_ << "#         momentum = " << momentumTarget_ << "\n";
1639	+	rnemdFile_ << "#     actual exchange totals:\n";
1640	+	rnemdFile_ << "#         kinetic = " << kineticExchange_ << "\n";
1641	+	rnemdFile_ << "#         momentum = " << momentumExchange_  << "\n";
1642	+	rnemdFile_ << "#     actual flux:\n";
1643	+	rnemdFile_ << "#         kinetic = " << Jz << "\n";
1644	+	rnemdFile_ << "#         momentum = " << JzP  << "\n";
1645	+	rnemdFile_ << "#     exchange statistics:\n";
1646	+	rnemdFile_ << "#         attempted = " << trialCount_ << "\n";
1647	+	rnemdFile_ << "#         failed = " << failTrialCount_ << "\n";
1648	+	if (rnemdMethod_ == rnemdNIVS) {
1649	+	rnemdFile_ << "#         NIVS root-check errors = "
1650	+	<< failRootCount_ << "\n";
1651	+	}
1652	+	rnemdFile_ << "#######################################################\n";
1653	+
1654	+
1655	+
1656	+	//write title
1657	+	rnemdFile_ << "#";
1658	+	for (unsigned int i = 0; i < outputMask_.size(); ++i) {
1659	+	if (outputMask_[i]) {
1660	+	rnemdFile_ << "\t" << data_[i].title <<
1661	+	"(" << data_[i].units << ")";
1662	+	}
1663	+	}
1664	+	rnemdFile_ << std::endl;
1665	+
1666	+	rnemdFile_.precision(8);
1667	+
1668	+	for (unsigned int j = 0; j < nBins_; j++) {
1669	+
1670	+	for (unsigned int i = 0; i < outputMask_.size(); ++i) {
1671	+	if (outputMask_[i]) {
1672	+	if (data_[i].dataType == "RealType")
1673	+	writeReal(i,j);
1674	+	else if (data_[i].dataType == "Vector3d")
1675	+	writeVector(i,j);
1676	+	else {
1677	+	sprintf( painCave.errMsg,
1678	+	"RNEMD found an unknown data type for: %s ",
1679	+	data_[i].title.c_str());
1680	+	painCave.isFatal = 1;
1681	+	simError();
1682	+	}
1683	+	}
1684	+	}
1685	+	rnemdFile_ << std::endl;
1686	+
1687	+	}
1688	+
1689	+	rnemdFile_ << "#######################################################\n";
1690	+	rnemdFile_ << "# Standard Deviations in those quantities follow:\n";
1691	+	rnemdFile_ << "#######################################################\n";
1692	+
1693	+
1694	+	for (unsigned int j = 0; j < nBins_; j++) {
1695	+	rnemdFile_ << "#";
1696	+	for (unsigned int i = 0; i < outputMask_.size(); ++i) {
1697	+	if (outputMask_[i]) {
1698	+	if (data_[i].dataType == "RealType")
1699	+	writeRealStdDev(i,j);
1700	+	else if (data_[i].dataType == "Vector3d")
1701	+	writeVectorStdDev(i,j);
1702	+	else {
1703	+	sprintf( painCave.errMsg,
1704	+	"RNEMD found an unknown data type for: %s ",
1705	+	data_[i].title.c_str());
1706	+	painCave.isFatal = 1;
1707	+	simError();
1708	+	}
1709	+	}
1710	+	}
1711	+	rnemdFile_ << std::endl;
1712	+
1713	+	}
1714	+
1715	+	rnemdFile_.flush();
1716	+	rnemdFile_.close();
1717	+
1718		#ifdef IS_MPI
1719		}
1720		#endif
1721	+
1722		}
1723	+
1724	+	void RNEMD::writeReal(int index, unsigned int bin) {
1725	+	if (!doRNEMD_) return;
1726	+	assert(index >=0 && index < ENDINDEX);
1727	+	assert(bin < nBins_);
1728	+	RealType s;
1729	+
1730	+	data_[index].accumulator[bin]->getAverage(s);
1731	+
1732	+	if (! isinf(s) && ! isnan(s)) {
1733	+	rnemdFile_ << "\t" << s;
1734	+	} else{
1735	+	sprintf( painCave.errMsg,
1736	+	"RNEMD detected a numerical error writing: %s for bin %d",
1737	+	data_[index].title.c_str(), bin);
1738	+	painCave.isFatal = 1;
1739	+	simError();
1740	+	}
1741	+	}
1742	+
1743	+	void RNEMD::writeVector(int index, unsigned int bin) {
1744	+	if (!doRNEMD_) return;
1745	+	assert(index >=0 && index < ENDINDEX);
1746	+	assert(bin < nBins_);
1747	+	Vector3d s;
1748	+	dynamic_cast<VectorAccumulator*>(data_[index].accumulator[bin])->getAverage(s);
1749	+	if (isinf(s[0]) || isnan(s[0]) ||
1750	+	isinf(s[1]) || isnan(s[1]) ||
1751	+	isinf(s[2]) || isnan(s[2]) ) {
1752	+	sprintf( painCave.errMsg,
1753	+	"RNEMD detected a numerical error writing: %s for bin %d",
1754	+	data_[index].title.c_str(), bin);
1755	+	painCave.isFatal = 1;
1756	+	simError();
1757	+	} else {
1758	+	rnemdFile_ << "\t" << s[0] << "\t" << s[1] << "\t" << s[2];
1759	+	}
1760	+	}
1761	+
1762	+	void RNEMD::writeRealStdDev(int index, unsigned int bin) {
1763	+	if (!doRNEMD_) return;
1764	+	assert(index >=0 && index < ENDINDEX);
1765	+	assert(bin < nBins_);
1766	+	RealType s;
1767	+
1768	+	data_[index].accumulator[bin]->getStdDev(s);
1769	+
1770	+	if (! isinf(s) && ! isnan(s)) {
1771	+	rnemdFile_ << "\t" << s;
1772	+	} else{
1773	+	sprintf( painCave.errMsg,
1774	+	"RNEMD detected a numerical error writing: %s std. dev. for bin %d",
1775	+	data_[index].title.c_str(), bin);
1776	+	painCave.isFatal = 1;
1777	+	simError();
1778	+	}
1779	+	}
1780	+
1781	+	void RNEMD::writeVectorStdDev(int index, unsigned int bin) {
1782	+	if (!doRNEMD_) return;
1783	+	assert(index >=0 && index < ENDINDEX);
1784	+	assert(bin < nBins_);
1785	+	Vector3d s;
1786	+	dynamic_cast<VectorAccumulator*>(data_[index].accumulator[bin])->getStdDev(s);
1787	+	if (isinf(s[0]) || isnan(s[0]) ||
1788	+	isinf(s[1]) || isnan(s[1]) ||
1789	+	isinf(s[2]) || isnan(s[2]) ) {
1790	+	sprintf( painCave.errMsg,
1791	+	"RNEMD detected a numerical error writing: %s std. dev. for bin %d",
1792	+	data_[index].title.c_str(), bin);
1793	+	painCave.isFatal = 1;
1794	+	simError();
1795	+	} else {
1796	+	rnemdFile_ << "\t" << s[0] << "\t" << s[1] << "\t" << s[2];
1797	+	}
1798	+	}
1799		}
1800	+

Comparing:
trunk/src/integrators/RNEMD.cpp (property svn:keywords), Revision 1350 by gezelter, Thu May 21 18:56:45 2009 UTC vs.
branches/development/src/rnemd/RNEMD.cpp (property svn:keywords), Revision 1776 by gezelter, Thu Aug 9 15:52:59 2012 UTC

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

Diff Legend

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