ViewVC Help

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

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

Comparing:
branches/development/src/integrators/RNEMD.cpp (file contents), Revision 1465 by chuckv, Fri Jul 9 23:08:25 2010 UTC vs.
branches/development/src/rnemd/RNEMD.cpp (file contents), Revision 1866 by gezelter, Thu Apr 25 14:32:56 2013 UTC

#	Line 35 | Line 35
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).
38	>	* [3]  Sun, Lin & Gezelter, J. Chem. Phys. 128, 234107 (2008).
39		* [4]  Vardeman & Gezelter, in progress (2009).
40		*/
41
42		#include <cmath>
43	<	#include "integrators/RNEMD.hpp"
43	>	#include <sstream>
44	>	#include <string>
45	>
46	>	#include "rnemd/RNEMD.hpp"
47		#include "math/Vector3.hpp"
48	+	#include "math/Vector.hpp"
49		#include "math/SquareMatrix3.hpp"
50		#include "math/Polynomial.hpp"
51		#include "primitives/Molecule.hpp"
52		#include "primitives/StuntDouble.hpp"
53		#include "utils/PhysicalConstants.hpp"
54		#include "utils/Tuple.hpp"
55	+	#include "brains/Thermo.hpp"
56	+	#include "math/ConvexHull.hpp"
57	+	#ifdef IS_MPI
58	+	#include <mpi.h>
59	+	#endif
60
61	<	#ifndef IS_MPI
62	<	#include "math/SeqRandNumGen.hpp"
63	<	#else
55	<	#include "math/ParallelRandNumGen.hpp"
61	>	#ifdef _MSC_VER
62	>	#define isnan(x) _isnan((x))
63	>	#define isinf(x) (!_finite(x) && !_isnan(x))
64		#endif
65
66		#define HONKING_LARGE_VALUE 1.0e10
67
68	+	using namespace std;
69		namespace OpenMD {
70
71	<	RNEMD::RNEMD(SimInfo* info) : info_(info), evaluator_(info), seleMan_(info), usePeriodicBoundaryConditions_(info->getSimParams()->getUsePeriodicBoundaryConditions()) {
71	>	RNEMD::RNEMD(SimInfo* info) : info_(info), evaluator_(info), seleMan_(info),
72	>	evaluatorA_(info), seleManA_(info),
73	>	commonA_(info), evaluatorB_(info),
74	>	seleManB_(info), commonB_(info),
75	>	usePeriodicBoundaryConditions_(info->getSimParams()->getUsePeriodicBoundaryConditions()) {
76
77	+	trialCount_ = 0;
78		failTrialCount_ = 0;
79		failRootCount_ = 0;
80
81	<	int seedValue;
82	<	Globals * simParams = info->getSimParams();
81	>	Globals* simParams = info->getSimParams();
82	>	RNEMDParameters* rnemdParams = simParams->getRNEMDParameters();
83
84	<	stringToEnumMap_["KineticSwap"] = rnemdKineticSwap;
85	<	stringToEnumMap_["KineticScale"] = rnemdKineticScale;
72	<	stringToEnumMap_["PxScale"] = rnemdPxScale;
73	<	stringToEnumMap_["PyScale"] = rnemdPyScale;
74	<	stringToEnumMap_["PzScale"] = rnemdPzScale;
75	<	stringToEnumMap_["Px"] = rnemdPx;
76	<	stringToEnumMap_["Py"] = rnemdPy;
77	<	stringToEnumMap_["Pz"] = rnemdPz;
78	<	stringToEnumMap_["Unknown"] = rnemdUnknown;
84	>	doRNEMD_ = rnemdParams->getUseRNEMD();
85	>	if (!doRNEMD_) return;
86
87	<	rnemdObjectSelection_ = simParams->getRNEMD_objectSelection();
88	<	evaluator_.loadScriptString(rnemdObjectSelection_);
89	<	seleMan_.setSelectionSet(evaluator_.evaluate());
87	>	stringToMethod_["Swap"]  = rnemdSwap;
88	>	stringToMethod_["NIVS"]  = rnemdNIVS;
89	>	stringToMethod_["VSS"]   = rnemdVSS;
90
91	<	// do some sanity checking
91	>	stringToFluxType_["KE"]  = rnemdKE;
92	>	stringToFluxType_["Px"]  = rnemdPx;
93	>	stringToFluxType_["Py"]  = rnemdPy;
94	>	stringToFluxType_["Pz"]  = rnemdPz;
95	>	stringToFluxType_["Pvector"]  = rnemdPvector;
96	>	stringToFluxType_["Lx"]  = rnemdLx;
97	>	stringToFluxType_["Ly"]  = rnemdLy;
98	>	stringToFluxType_["Lz"]  = rnemdLz;
99	>	stringToFluxType_["Lvector"]  = rnemdLvector;
100	>	stringToFluxType_["KE+Px"]  = rnemdKePx;
101	>	stringToFluxType_["KE+Py"]  = rnemdKePy;
102	>	stringToFluxType_["KE+Pvector"]  = rnemdKePvector;
103	>	stringToFluxType_["KE+Lx"]  = rnemdKeLx;
104	>	stringToFluxType_["KE+Ly"]  = rnemdKeLy;
105	>	stringToFluxType_["KE+Lz"]  = rnemdKeLz;
106	>	stringToFluxType_["KE+Lvector"]  = rnemdKeLvector;
107
108	<	int selectionCount = seleMan_.getSelectionCount();
109	<	int nIntegrable = info->getNGlobalIntegrableObjects();
108	>	runTime_ = simParams->getRunTime();
109	>	statusTime_ = simParams->getStatusTime();
110
111	<	if (selectionCount > nIntegrable) {
111	>	const string methStr = rnemdParams->getMethod();
112	>	bool hasFluxType = rnemdParams->haveFluxType();
113	>
114	>	rnemdObjectSelection_ = rnemdParams->getObjectSelection();
115	>
116	>	string fluxStr;
117	>	if (hasFluxType) {
118	>	fluxStr = rnemdParams->getFluxType();
119	>	} else {
120		sprintf(painCave.errMsg,
121	<	"RNEMD warning: The current RNEMD_objectSelection,\n"
122	<	"\t\t%s\n"
123	<	"\thas resulted in %d selected objects.  However,\n"
124	<	"\tthe total number of integrable objects in the system\n"
125	<	"\tis only %d.  This is almost certainly not what you want\n"
126	<	"\tto do.  A likely cause of this is forgetting the _RB_0\n"
127	<	"\tselector in the selection script!\n",
98	<	rnemdObjectSelection_.c_str(),
99	<	selectionCount, nIntegrable);
100	<	painCave.isFatal = 0;
121	>	"RNEMD: No fluxType was set in the md file.  This parameter,\n"
122	>	"\twhich must be one of the following values:\n"
123	>	"\tKE, Px, Py, Pz, Pvector, Lx, Ly, Lz, Lvector,\n"
124	>	"\tKE+Px, KE+Py, KE+Pvector, KE+Lx, KE+Ly, KE+Lz, KE+Lvector\n"
125	>	"\tmust be set to use RNEMD\n");
126	>	painCave.isFatal = 1;
127	>	painCave.severity = OPENMD_ERROR;
128		simError();
102	–
129		}
130	+
131	+	bool hasKineticFlux = rnemdParams->haveKineticFlux();
132	+	bool hasMomentumFlux = rnemdParams->haveMomentumFlux();
133	+	bool hasMomentumFluxVector = rnemdParams->haveMomentumFluxVector();
134	+	bool hasAngularMomentumFlux = rnemdParams->haveAngularMomentumFlux();
135	+	bool hasAngularMomentumFluxVector = rnemdParams->haveAngularMomentumFluxVector();
136	+	hasSelectionA_ = rnemdParams->haveSelectionA();
137	+	hasSelectionB_ = rnemdParams->haveSelectionB();
138	+	bool hasSlabWidth = rnemdParams->haveSlabWidth();
139	+	bool hasSlabACenter = rnemdParams->haveSlabACenter();
140	+	bool hasSlabBCenter = rnemdParams->haveSlabBCenter();
141	+	bool hasSphereARadius = rnemdParams->haveSphereARadius();
142	+	hasSphereBRadius_ = rnemdParams->haveSphereBRadius();
143	+	bool hasCoordinateOrigin = rnemdParams->haveCoordinateOrigin();
144	+	bool hasOutputFileName = rnemdParams->haveOutputFileName();
145	+	bool hasOutputFields = rnemdParams->haveOutputFields();
146
147	<	const std::string st = simParams->getRNEMD_exchangeType();
147	>	map<string, RNEMDMethod>::iterator i;
148	>	i = stringToMethod_.find(methStr);
149	>	if (i != stringToMethod_.end())
150	>	rnemdMethod_ = i->second;
151	>	else {
152	>	sprintf(painCave.errMsg,
153	>	"RNEMD: The current method,\n"
154	>	"\t\t%s is not one of the recognized\n"
155	>	"\texchange methods: Swap, NIVS, or VSS\n",
156	>	methStr.c_str());
157	>	painCave.isFatal = 1;
158	>	painCave.severity = OPENMD_ERROR;
159	>	simError();
160	>	}
161
162	<	std::map<std::string, RNEMDTypeEnum>::iterator i;
163	<	i = stringToEnumMap_.find(st);
164	<	rnemdType_ = (i == stringToEnumMap_.end()) ? RNEMD::rnemdUnknown : i->second;
165	<	if (rnemdType_ == rnemdUnknown) {
166	<	std::cerr << "WARNING! RNEMD Type Unknown!\n";
162	>	map<string, RNEMDFluxType>::iterator j;
163	>	j = stringToFluxType_.find(fluxStr);
164	>	if (j != stringToFluxType_.end())
165	>	rnemdFluxType_ = j->second;
166	>	else {
167	>	sprintf(painCave.errMsg,
168	>	"RNEMD: The current fluxType,\n"
169	>	"\t\t%s\n"
170	>	"\tis not one of the recognized flux types.\n",
171	>	fluxStr.c_str());
172	>	painCave.isFatal = 1;
173	>	painCave.severity = OPENMD_ERROR;
174	>	simError();
175		}
176
177	<	#ifdef IS_MPI
178	<	if (worldRank == 0) {
179	<	#endif
180	<
181	<	std::string rnemdFileName;
182	<	std::string xTempFileName;
183	<	std::string yTempFileName;
121	<	std::string zTempFileName;
122	<	switch(rnemdType_) {
123	<	case rnemdKineticSwap :
124	<	case rnemdKineticScale :
125	<	rnemdFileName = "temperature.log";
177	>	bool methodFluxMismatch = false;
178	>	bool hasCorrectFlux = false;
179	>	switch(rnemdMethod_) {
180	>	case rnemdSwap:
181	>	switch (rnemdFluxType_) {
182	>	case rnemdKE:
183	>	hasCorrectFlux = hasKineticFlux;
184		break;
185	<	case rnemdPx :
186	<	case rnemdPxScale :
187	<	case rnemdPy :
188	<	case rnemdPyScale :
131	<	rnemdFileName = "momemtum.log";
132	<	xTempFileName = "temperatureX.log";
133	<	yTempFileName = "temperatureY.log";
134	<	zTempFileName = "temperatureZ.log";
135	<	xTempLog_.open(xTempFileName.c_str());
136	<	yTempLog_.open(yTempFileName.c_str());
137	<	zTempLog_.open(zTempFileName.c_str());
185	>	case rnemdPx:
186	>	case rnemdPy:
187	>	case rnemdPz:
188	>	hasCorrectFlux = hasMomentumFlux;
189		break;
139	–	case rnemdPz :
140	–	case rnemdPzScale :
141	–	case rnemdUnknown :
190		default :
191	<	rnemdFileName = "rnemd.log";
191	>	methodFluxMismatch = true;
192		break;
193		}
194	<	rnemdLog_.open(rnemdFileName.c_str());
195	<
196	<	#ifdef IS_MPI
194	>	break;
195	>	case rnemdNIVS:
196	>	switch (rnemdFluxType_) {
197	>	case rnemdKE:
198	>	case rnemdRotKE:
199	>	case rnemdFullKE:
200	>	hasCorrectFlux = hasKineticFlux;
201	>	break;
202	>	case rnemdPx:
203	>	case rnemdPy:
204	>	case rnemdPz:
205	>	hasCorrectFlux = hasMomentumFlux;
206	>	break;
207	>	case rnemdKePx:
208	>	case rnemdKePy:
209	>	hasCorrectFlux = hasMomentumFlux && hasKineticFlux;
210	>	break;
211	>	default:
212	>	methodFluxMismatch = true;
213	>	break;
214	>	}
215	>	break;
216	>	case rnemdVSS:
217	>	switch (rnemdFluxType_) {
218	>	case rnemdKE:
219	>	case rnemdRotKE:
220	>	case rnemdFullKE:
221	>	hasCorrectFlux = hasKineticFlux;
222	>	break;
223	>	case rnemdPx:
224	>	case rnemdPy:
225	>	case rnemdPz:
226	>	hasCorrectFlux = hasMomentumFlux;
227	>	break;
228	>	case rnemdLx:
229	>	case rnemdLy:
230	>	case rnemdLz:
231	>	hasCorrectFlux = hasAngularMomentumFlux;
232	>	break;
233	>	case rnemdPvector:
234	>	hasCorrectFlux = hasMomentumFluxVector;
235	>	break;
236	>	case rnemdLvector:
237	>	hasCorrectFlux = hasAngularMomentumFluxVector;
238	>	break;
239	>	case rnemdKePx:
240	>	case rnemdKePy:
241	>	hasCorrectFlux = hasMomentumFlux && hasKineticFlux;
242	>	break;
243	>	case rnemdKeLx:
244	>	case rnemdKeLy:
245	>	case rnemdKeLz:
246	>	hasCorrectFlux = hasAngularMomentumFlux && hasKineticFlux;
247	>	break;
248	>	case rnemdKePvector:
249	>	hasCorrectFlux = hasMomentumFluxVector && hasKineticFlux;
250	>	break;
251	>	case rnemdKeLvector:
252	>	hasCorrectFlux = hasAngularMomentumFluxVector && hasKineticFlux;
253	>	break;
254	>	default:
255	>	methodFluxMismatch = true;
256	>	break;
257	>	}
258	>	default:
259	>	break;
260		}
150	–	#endif
261
262	<	set_RNEMD_exchange_time(simParams->getRNEMD_exchangeTime());
263	<	set_RNEMD_nBins(simParams->getRNEMD_nBins());
264	<	midBin_ = nBins_ / 2;
265	<	if (simParams->haveRNEMD_logWidth()) {
266	<	rnemdLogWidth_ = simParams->getRNEMD_logWidth();
267	<	if (rnemdLogWidth_ != nBins_ && rnemdLogWidth_ != midBin_ + 1) {
268	<	std::cerr << "WARNING! RNEMD_logWidth has abnormal value!\n";
269	<	std::cerr << "Automaically set back to default.\n";
270	<	rnemdLogWidth_ = nBins_;
271	<	}
262	>	if (methodFluxMismatch) {
263	>	sprintf(painCave.errMsg,
264	>	"RNEMD: The current method,\n"
265	>	"\t\t%s\n"
266	>	"\tcannot be used with the current flux type, %s\n",
267	>	methStr.c_str(), fluxStr.c_str());
268	>	painCave.isFatal = 1;
269	>	painCave.severity = OPENMD_ERROR;
270	>	simError();
271	>	}
272	>	if (!hasCorrectFlux) {
273	>	sprintf(painCave.errMsg,
274	>	"RNEMD: The current method, %s, and flux type, %s,\n"
275	>	"\tdid not have the correct flux value specified. Options\n"
276	>	"\tinclude: kineticFlux, momentumFlux, angularMomentumFlux,\n"
277	>	"\tmomentumFluxVector, and angularMomentumFluxVector.\n",
278	>	methStr.c_str(), fluxStr.c_str());
279	>	painCave.isFatal = 1;
280	>	painCave.severity = OPENMD_ERROR;
281	>	simError();
282	>	}
283	>
284	>	if (hasKineticFlux) {
285	>	// convert the kcal / mol / Angstroms^2 / fs values in the md file
286	>	// into  amu / fs^3:
287	>	kineticFlux_ = rnemdParams->getKineticFlux()
288	>	* PhysicalConstants::energyConvert;
289		} else {
290	<	rnemdLogWidth_ = nBins_;
290	>	kineticFlux_ = 0.0;
291		}
292	<	valueHist_.resize(rnemdLogWidth_, 0.0);
293	<	valueCount_.resize(rnemdLogWidth_, 0);
167	<	xTempHist_.resize(rnemdLogWidth_, 0.0);
168	<	yTempHist_.resize(rnemdLogWidth_, 0.0);
169	<	zTempHist_.resize(rnemdLogWidth_, 0.0);
170	<
171	<	set_RNEMD_exchange_total(0.0);
172	<	if (simParams->haveRNEMD_targetFlux()) {
173	<	set_RNEMD_target_flux(simParams->getRNEMD_targetFlux());
292	>	if (hasMomentumFluxVector) {
293	>	momentumFluxVector_ = rnemdParams->getMomentumFluxVector();
294		} else {
295	<	set_RNEMD_target_flux(0.0);
295	>	momentumFluxVector_ = V3Zero;
296	>	if (hasMomentumFlux) {
297	>	RealType momentumFlux = rnemdParams->getMomentumFlux();
298	>	switch (rnemdFluxType_) {
299	>	case rnemdPx:
300	>	momentumFluxVector_.x() = momentumFlux;
301	>	break;
302	>	case rnemdPy:
303	>	momentumFluxVector_.y() = momentumFlux;
304	>	break;
305	>	case rnemdPz:
306	>	momentumFluxVector_.z() = momentumFlux;
307	>	break;
308	>	case rnemdKePx:
309	>	momentumFluxVector_.x() = momentumFlux;
310	>	break;
311	>	case rnemdKePy:
312	>	momentumFluxVector_.y() = momentumFlux;
313	>	break;
314	>	default:
315	>	break;
316	>	}
317	>	}
318	>	if (hasAngularMomentumFluxVector) {
319	>	angularMomentumFluxVector_ = rnemdParams->getAngularMomentumFluxVector();
320	>	} else {
321	>	angularMomentumFluxVector_ = V3Zero;
322	>	if (hasAngularMomentumFlux) {
323	>	RealType angularMomentumFlux = rnemdParams->getAngularMomentumFlux();
324	>	switch (rnemdFluxType_) {
325	>	case rnemdLx:
326	>	angularMomentumFluxVector_.x() = angularMomentumFlux;
327	>	break;
328	>	case rnemdLy:
329	>	angularMomentumFluxVector_.y() = angularMomentumFlux;
330	>	break;
331	>	case rnemdLz:
332	>	angularMomentumFluxVector_.z() = angularMomentumFlux;
333	>	break;
334	>	case rnemdKeLx:
335	>	angularMomentumFluxVector_.x() = angularMomentumFlux;
336	>	break;
337	>	case rnemdKeLy:
338	>	angularMomentumFluxVector_.y() = angularMomentumFlux;
339	>	break;
340	>	case rnemdKeLz:
341	>	angularMomentumFluxVector_.z() = angularMomentumFlux;
342	>	break;
343	>	default:
344	>	break;
345	>	}
346	>	}
347	>	}
348	>
349	>	if (hasCoordinateOrigin) {
350	>	coordinateOrigin_ = rnemdParams->getCoordinateOrigin();
351	>	} else {
352	>	coordinateOrigin_ = V3Zero;
353	>	}
354	>
355	>	// do some sanity checking
356	>
357	>	int selectionCount = seleMan_.getSelectionCount();
358	>
359	>	int nIntegrable = info->getNGlobalIntegrableObjects();
360	>
361	>	if (selectionCount > nIntegrable) {
362	>	sprintf(painCave.errMsg,
363	>	"RNEMD: The current objectSelection,\n"
364	>	"\t\t%s\n"
365	>	"\thas resulted in %d selected objects.  However,\n"
366	>	"\tthe total number of integrable objects in the system\n"
367	>	"\tis only %d.  This is almost certainly not what you want\n"
368	>	"\tto do.  A likely cause of this is forgetting the _RB_0\n"
369	>	"\tselector in the selection script!\n",
370	>	rnemdObjectSelection_.c_str(),
371	>	selectionCount, nIntegrable);
372	>	painCave.isFatal = 0;
373	>	painCave.severity = OPENMD_WARNING;
374	>	simError();
375	>	}
376	>
377	>	areaAccumulator_ = new Accumulator();
378	>
379	>	nBins_ = rnemdParams->getOutputBins();
380	>	binWidth_ = rnemdParams->getOutputBinWidth();
381	>
382	>	data_.resize(RNEMD::ENDINDEX);
383	>	OutputData z;
384	>	z.units =  "Angstroms";
385	>	z.title =  "Z";
386	>	z.dataType = "RealType";
387	>	z.accumulator.reserve(nBins_);
388	>	for (int i = 0; i < nBins_; i++)
389	>	z.accumulator.push_back( new Accumulator() );
390	>	data_[Z] = z;
391	>	outputMap_["Z"] =  Z;
392	>
393	>	OutputData r;
394	>	r.units =  "Angstroms";
395	>	r.title =  "R";
396	>	r.dataType = "RealType";
397	>	r.accumulator.reserve(nBins_);
398	>	for (int i = 0; i < nBins_; i++)
399	>	r.accumulator.push_back( new Accumulator() );
400	>	data_[R] = r;
401	>	outputMap_["R"] =  R;
402	>
403	>	OutputData temperature;
404	>	temperature.units =  "K";
405	>	temperature.title =  "Temperature";
406	>	temperature.dataType = "RealType";
407	>	temperature.accumulator.reserve(nBins_);
408	>	for (int i = 0; i < nBins_; i++)
409	>	temperature.accumulator.push_back( new Accumulator() );
410	>	data_[TEMPERATURE] = temperature;
411	>	outputMap_["TEMPERATURE"] =  TEMPERATURE;
412	>
413	>	OutputData velocity;
414	>	velocity.units = "angstroms/fs";
415	>	velocity.title =  "Velocity";
416	>	velocity.dataType = "Vector3d";
417	>	velocity.accumulator.reserve(nBins_);
418	>	for (int i = 0; i < nBins_; i++)
419	>	velocity.accumulator.push_back( new VectorAccumulator() );
420	>	data_[VELOCITY] = velocity;
421	>	outputMap_["VELOCITY"] = VELOCITY;
422	>
423	>	OutputData angularVelocity;
424	>	angularVelocity.units = "angstroms^2/fs";
425	>	angularVelocity.title =  "AngularVelocity";
426	>	angularVelocity.dataType = "Vector3d";
427	>	angularVelocity.accumulator.reserve(nBins_);
428	>	for (int i = 0; i < nBins_; i++)
429	>	angularVelocity.accumulator.push_back( new VectorAccumulator() );
430	>	data_[ANGULARVELOCITY] = angularVelocity;
431	>	outputMap_["ANGULARVELOCITY"] = ANGULARVELOCITY;
432	>
433	>	OutputData density;
434	>	density.units =  "g cm^-3";
435	>	density.title =  "Density";
436	>	density.dataType = "RealType";
437	>	density.accumulator.reserve(nBins_);
438	>	for (int i = 0; i < nBins_; i++)
439	>	density.accumulator.push_back( new Accumulator() );
440	>	data_[DENSITY] = density;
441	>	outputMap_["DENSITY"] =  DENSITY;
442	>
443	>	if (hasOutputFields) {
444	>	parseOutputFileFormat(rnemdParams->getOutputFields());
445	>	} else {
446	>	if (usePeriodicBoundaryConditions_)
447	>	outputMask_.set(Z);
448	>	else
449	>	outputMask_.set(R);
450	>	switch (rnemdFluxType_) {
451	>	case rnemdKE:
452	>	case rnemdRotKE:
453	>	case rnemdFullKE:
454	>	outputMask_.set(TEMPERATURE);
455	>	break;
456	>	case rnemdPx:
457	>	case rnemdPy:
458	>	outputMask_.set(VELOCITY);
459	>	break;
460	>	case rnemdPz:
461	>	case rnemdPvector:
462	>	outputMask_.set(VELOCITY);
463	>	outputMask_.set(DENSITY);
464	>	break;
465	>	case rnemdLx:
466	>	case rnemdLy:
467	>	case rnemdLz:
468	>	case rnemdLvector:
469	>	outputMask_.set(ANGULARVELOCITY);
470	>	break;
471	>	case rnemdKeLx:
472	>	case rnemdKeLy:
473	>	case rnemdKeLz:
474	>	case rnemdKeLvector:
475	>	outputMask_.set(TEMPERATURE);
476	>	outputMask_.set(ANGULARVELOCITY);
477	>	break;
478	>	case rnemdKePx:
479	>	case rnemdKePy:
480	>	outputMask_.set(TEMPERATURE);
481	>	outputMask_.set(VELOCITY);
482	>	break;
483	>	case rnemdKePvector:
484	>	outputMask_.set(TEMPERATURE);
485	>	outputMask_.set(VELOCITY);
486	>	outputMask_.set(DENSITY);
487	>	break;
488	>	default:
489	>	break;
490	>	}
491	>	}
492	>
493	>	if (hasOutputFileName) {
494	>	rnemdFileName_ = rnemdParams->getOutputFileName();
495	>	} else {
496	>	rnemdFileName_ = getPrefix(info->getFinalConfigFileName()) + ".rnemd";
497	>	}
498	>
499	>	exchangeTime_ = rnemdParams->getExchangeTime();
500	>
501	>	Snapshot* currentSnap_ = info->getSnapshotManager()->getCurrentSnapshot();
502	>	// total exchange sums are zeroed out at the beginning:
503	>
504	>	kineticExchange_ = 0.0;
505	>	momentumExchange_ = V3Zero;
506	>	angularMomentumExchange_ = V3Zero;
507	>
508	>	std::ostringstream selectionAstream;
509	>	std::ostringstream selectionBstream;
510	>
511	>	if (hasSelectionA_) {
512	>	selectionA_ = rnemdParams->getSelectionA();
513	>	} else {
514	>	if (usePeriodicBoundaryConditions_) {
515	>	Mat3x3d hmat = currentSnap_->getHmat();
516	>
517	>	if (hasSlabWidth)
518	>	slabWidth_ = rnemdParams->getSlabWidth();
519	>	else
520	>	slabWidth_ = hmat(2,2) / 10.0;
521	>
522	>	if (hasSlabACenter)
523	>	slabACenter_ = rnemdParams->getSlabACenter();
524	>	else
525	>	slabACenter_ = 0.0;
526	>
527	>	selectionAstream << "select wrappedz > "
528	>	<< slabACenter_ - 0.5*slabWidth_
529	>	<<  " && wrappedz < "
530	>	<< slabACenter_ + 0.5*slabWidth_;
531	>	selectionA_ = selectionAstream.str();
532	>	} else {
533	>	if (hasSphereARadius)
534	>	sphereARadius_ = rnemdParams->getSphereARadius();
535	>	else {
536	>	// use an initial guess to the size of the inner slab to be 1/10 the
537	>	// radius of an approximately spherical hull:
538	>	Thermo thermo(info);
539	>	RealType hVol = thermo.getHullVolume();
540	>	sphereARadius_ = 0.1 * pow((3.0 * hVol / (4.0 * M_PI)), 1.0/3.0);
541	>	}
542	>	selectionAstream << "select r < " << sphereARadius_;
543	>	selectionA_ = selectionAstream.str();
544	>	}
545	>	}
546	>
547	>	if (hasSelectionB_) {
548	>	selectionB_ = rnemdParams->getSelectionB();
549	>	} else {
550	>	if (usePeriodicBoundaryConditions_) {
551	>	Mat3x3d hmat = currentSnap_->getHmat();
552	>
553	>	if (hasSlabWidth)
554	>	slabWidth_ = rnemdParams->getSlabWidth();
555	>	else
556	>	slabWidth_ = hmat(2,2) / 10.0;
557	>
558	>	if (hasSlabBCenter)
559	>	slabBCenter_ = rnemdParams->getSlabBCenter();
560	>	else
561	>	slabBCenter_ = hmat(2,2) / 2.0;
562	>
563	>	selectionBstream << "select wrappedz > "
564	>	<< slabBCenter_ - 0.5*slabWidth_
565	>	<<  " && wrappedz < "
566	>	<< slabBCenter_ + 0.5*slabWidth_;
567	>	selectionB_ = selectionBstream.str();
568	>	} else {
569	>	if (hasSphereBRadius_) {
570	>	sphereBRadius_ = rnemdParams->getSphereBRadius();
571	>	selectionBstream << "select r > " << sphereBRadius_;
572	>	selectionB_ = selectionBstream.str();
573	>	} else {
574	>	selectionB_ = "select hull";
575	>	hasSelectionB_ = true;
576	>	}
577	>	}
578	>	}
579		}
580
581	<	#ifndef IS_MPI
582	<	if (simParams->haveSeed()) {
583	<	seedValue = simParams->getSeed();
584	<	randNumGen_ = new SeqRandNumGen(seedValue);
585	<	}else {
586	<	randNumGen_ = new SeqRandNumGen();
587	<	}
588	<	#else
589	<	if (simParams->haveSeed()) {
187	<	seedValue = simParams->getSeed();
188	<	randNumGen_ = new ParallelRandNumGen(seedValue);
189	<	}else {
190	<	randNumGen_ = new ParallelRandNumGen();
191	<	}
192	<	#endif
581	>	// object evaluator:
582	>	evaluator_.loadScriptString(rnemdObjectSelection_);
583	>	seleMan_.setSelectionSet(evaluator_.evaluate());
584	>	evaluatorA_.loadScriptString(selectionA_);
585	>	evaluatorB_.loadScriptString(selectionB_);
586	>	seleManA_.setSelectionSet(evaluatorA_.evaluate());
587	>	seleManB_.setSelectionSet(evaluatorB_.evaluate());
588	>	commonA_ = seleManA_ & seleMan_;
589	>	commonB_ = seleManB_ & seleMan_;
590		}
591
195	–	RNEMD::~RNEMD() {
196	–	delete randNumGen_;
592
593	+	RNEMD::~RNEMD() {
594	+	if (!doRNEMD_) return;
595		#ifdef IS_MPI
596		if (worldRank == 0) {
597		#endif
598	<	std::cerr << "total fail trials: " << failTrialCount_ << "\n";
599	<	rnemdLog_.close();
600	<	if (rnemdType_ == rnemdKineticScale || rnemdType_ == rnemdPxScale || rnemdType_ == rnemdPyScale)
601	<	std::cerr<< "total root-checking warnings: " << failRootCount_ << "\n";
602	<	if (rnemdType_ == rnemdPx || rnemdType_ == rnemdPxScale || rnemdType_ == rnemdPy || rnemdType_ == rnemdPyScale) {
206	<	xTempLog_.close();
207	<	yTempLog_.close();
208	<	zTempLog_.close();
209	<	}
598	>
599	>	writeOutputFile();
600	>
601	>	rnemdFile_.close();
602	>
603		#ifdef IS_MPI
604		}
605		#endif
606		}
607	+
608	+	void RNEMD::doSwap(SelectionManager& smanA, SelectionManager& smanB) {
609	+	if (!doRNEMD_) return;
610	+	int selei;
611	+	int selej;
612
215	–	void RNEMD::doSwap() {
216	–
613		Snapshot* currentSnap_ = info_->getSnapshotManager()->getCurrentSnapshot();
614		Mat3x3d hmat = currentSnap_->getHmat();
615
220	–	seleMan_.setSelectionSet(evaluator_.evaluate());
221	–
222	–	int selei;
616		StuntDouble* sd;
224	–	int idx;
617
618		RealType min_val;
619		bool min_found = false;
#	Line 231 | Line 623 | namespace OpenMD {
623		bool max_found = false;
624		StuntDouble* max_sd;
625
626	<	for (sd = seleMan_.beginSelected(selei); sd != NULL;
627	<	sd = seleMan_.nextSelected(selei)) {
626	>	for (sd = seleManA_.beginSelected(selei); sd != NULL;
627	>	sd = seleManA_.nextSelected(selei)) {
628
237	–	idx = sd->getLocalIndex();
238	–
629		Vector3d pos = sd->getPos();
630	<
630	>
631		// wrap the stuntdouble's position back into the box:
632	<
632	>
633		if (usePeriodicBoundaryConditions_)
634		currentSnap_->wrapVector(pos);
635	<
636	<	// which bin is this stuntdouble in?
637	<	// wrapped positions are in the range [-0.5*hmat(2,2), +0.5*hmat(2,2)]
638	<
639	<	int binNo = int(nBins_ * (pos.z() / hmat(2,2) + 0.5)) % nBins_;
640	<
641	<
252	<	// if we're in bin 0 or the middleBin
253	<	if (binNo == 0 || binNo == midBin_) {
635	>
636	>	RealType mass = sd->getMass();
637	>	Vector3d vel = sd->getVel();
638	>	RealType value;
639	>
640	>	switch(rnemdFluxType_) {
641	>	case rnemdKE :
642
643	<	RealType mass = sd->getMass();
644	<	Vector3d vel = sd->getVel();
645	<	RealType value;
646	<
647	<	switch(rnemdType_) {
260	<	case rnemdKineticSwap :
643	>	value = mass * vel.lengthSquare();
644	>
645	>	if (sd->isDirectional()) {
646	>	Vector3d angMom = sd->getJ();
647	>	Mat3x3d I = sd->getI();
648
649	<	value = mass * (vel[0]*vel[0] + vel[1]*vel[1] +
650	<	vel[2]*vel[2]);
651	<	if (sd->isDirectional()) {
652	<	Vector3d angMom = sd->getJ();
653	<	Mat3x3d I = sd->getI();
654	<
655	<	if (sd->isLinear()) {
656	<	int i = sd->linearAxis();
657	<	int j = (i + 1) % 3;
658	<	int k = (i + 2) % 3;
272	<	value += angMom[j] * angMom[j] / I(j, j) +
273	<	angMom[k] * angMom[k] / I(k, k);
274	<	} else {
275	<	value += angMom[0]*angMom[0]/I(0, 0)
276	<	+ angMom[1]*angMom[1]/I(1, 1)
277	<	+ angMom[2]*angMom[2]/I(2, 2);
278	<	}
649	>	if (sd->isLinear()) {
650	>	int i = sd->linearAxis();
651	>	int j = (i + 1) % 3;
652	>	int k = (i + 2) % 3;
653	>	value += angMom[j] * angMom[j] / I(j, j) +
654	>	angMom[k] * angMom[k] / I(k, k);
655	>	} else {
656	>	value += angMom[0]*angMom[0]/I(0, 0)
657	>	+ angMom[1]*angMom[1]/I(1, 1)
658	>	+ angMom[2]*angMom[2]/I(2, 2);
659		}
660	<	//make exchangeSum_ comparable between swap & scale
661	<	//temporarily without using energyConvert
662	<	//value = value * 0.5 / PhysicalConstants::energyConvert;
663	<	value *= 0.5;
664	<	break;
665	<	case rnemdPx :
666	<	value = mass * vel[0];
667	<	break;
668	<	case rnemdPy :
669	<	value = mass * vel[1];
670	<	break;
671	<	case rnemdPz :
672	<	value = mass * vel[2];
673	<	break;
674	<	default :
675	<	break;
660	>	} //angular momenta exchange enabled
661	>	value *= 0.5;
662	>	break;
663	>	case rnemdPx :
664	>	value = mass * vel[0];
665	>	break;
666	>	case rnemdPy :
667	>	value = mass * vel[1];
668	>	break;
669	>	case rnemdPz :
670	>	value = mass * vel[2];
671	>	break;
672	>	default :
673	>	break;
674	>	}
675	>	if (!max_found) {
676	>	max_val = value;
677	>	max_sd = sd;
678	>	max_found = true;
679	>	} else {
680	>	if (max_val < value) {
681	>	max_val = value;
682	>	max_sd = sd;
683		}
684	+	}
685	+	}
686
687	<	if (binNo == 0) {
688	<	if (!min_found) {
689	<	min_val = value;
690	<	min_sd = sd;
691	<	min_found = true;
692	<	} else {
693	<	if (min_val > value) {
694	<	min_val = value;
695	<	min_sd = sd;
696	<	}
697	<	}
698	<	} else { //midBin_
699	<	if (!max_found) {
700	<	max_val = value;
701	<	max_sd = sd;
702	<	max_found = true;
703	<	} else {
704	<	if (max_val < value) {
705	<	max_val = value;
706	<	max_sd = sd;
707	<	}
708	<	}
709	<	}
687	>	for (sd = seleManB_.beginSelected(selej); sd != NULL;
688	>	sd = seleManB_.nextSelected(selej)) {
689	>
690	>	Vector3d pos = sd->getPos();
691	>
692	>	// wrap the stuntdouble's position back into the box:
693	>
694	>	if (usePeriodicBoundaryConditions_)
695	>	currentSnap_->wrapVector(pos);
696	>
697	>	RealType mass = sd->getMass();
698	>	Vector3d vel = sd->getVel();
699	>	RealType value;
700	>
701	>	switch(rnemdFluxType_) {
702	>	case rnemdKE :
703	>
704	>	value = mass * vel.lengthSquare();
705	>
706	>	if (sd->isDirectional()) {
707	>	Vector3d angMom = sd->getJ();
708	>	Mat3x3d I = sd->getI();
709	>
710	>	if (sd->isLinear()) {
711	>	int i = sd->linearAxis();
712	>	int j = (i + 1) % 3;
713	>	int k = (i + 2) % 3;
714	>	value += angMom[j] * angMom[j] / I(j, j) +
715	>	angMom[k] * angMom[k] / I(k, k);
716	>	} else {
717	>	value += angMom[0]*angMom[0]/I(0, 0)
718	>	+ angMom[1]*angMom[1]/I(1, 1)
719	>	+ angMom[2]*angMom[2]/I(2, 2);
720	>	}
721	>	} //angular momenta exchange enabled
722	>	value *= 0.5;
723	>	break;
724	>	case rnemdPx :
725	>	value = mass * vel[0];
726	>	break;
727	>	case rnemdPy :
728	>	value = mass * vel[1];
729	>	break;
730	>	case rnemdPz :
731	>	value = mass * vel[2];
732	>	break;
733	>	default :
734	>	break;
735		}
736	+
737	+	if (!min_found) {
738	+	min_val = value;
739	+	min_sd = sd;
740	+	min_found = true;
741	+	} else {
742	+	if (min_val > value) {
743	+	min_val = value;
744	+	min_sd = sd;
745	+	}
746	+	}
747		}
748	<
749	<	#ifdef IS_MPI
750	<	int nProc, worldRank;
751	<
327	<	nProc = MPI::COMM_WORLD.Get_size();
328	<	worldRank = MPI::COMM_WORLD.Get_rank();
329	<
748	>
749	>	#ifdef IS_MPI
750	>	int worldRank = MPI::COMM_WORLD.Get_rank();
751	>
752		bool my_min_found = min_found;
753		bool my_max_found = max_found;
754
755		// Even if we didn't find a minimum, did someone else?
756	<	MPI::COMM_WORLD.Allreduce(&my_min_found, &min_found,
335	<	1, MPI::BOOL, MPI::LAND);
336	<
756	>	MPI::COMM_WORLD.Allreduce(&my_min_found, &min_found, 1, MPI::BOOL, MPI::LOR);
757		// Even if we didn't find a maximum, did someone else?
758	<	MPI::COMM_WORLD.Allreduce(&my_max_found, &max_found,
759	<	1, MPI::BOOL, MPI::LAND);
760	<
761	<	struct {
762	<	RealType val;
763	<	int rank;
764	<	} max_vals, min_vals;
765	<
766	<	if (min_found) {
767	<	if (my_min_found)
758	>	MPI::COMM_WORLD.Allreduce(&my_max_found, &max_found, 1, MPI::BOOL, MPI::LOR);
759	>	#endif
760	>
761	>	if (max_found && min_found) {
762	>
763	>	#ifdef IS_MPI
764	>	struct {
765	>	RealType val;
766	>	int rank;
767	>	} max_vals, min_vals;
768	>
769	>	if (my_min_found) {
770		min_vals.val = min_val;
771	<	else
771	>	} else {
772		min_vals.val = HONKING_LARGE_VALUE;
773	<
773	>	}
774		min_vals.rank = worldRank;
775
776		// Who had the minimum?
777		MPI::COMM_WORLD.Allreduce(&min_vals, &min_vals,
778		1, MPI::REALTYPE_INT, MPI::MINLOC);
779		min_val = min_vals.val;
358	–	}
780
781	<	if (max_found) {
361	<	if (my_max_found)
781	>	if (my_max_found) {
782		max_vals.val = max_val;
783	<	else
783	>	} else {
784		max_vals.val = -HONKING_LARGE_VALUE;
785	<
785	>	}
786		max_vals.rank = worldRank;
787
788		// Who had the maximum?
789		MPI::COMM_WORLD.Allreduce(&max_vals, &max_vals,
790		1, MPI::REALTYPE_INT, MPI::MAXLOC);
791		max_val = max_vals.val;
372	–	}
792		#endif
793	<
794	<	if (max_found && min_found) {
795	<	if (min_val< max_val) {
377	<
793	>
794	>	if (min_val < max_val) {
795	>
796		#ifdef IS_MPI
797		if (max_vals.rank == worldRank && min_vals.rank == worldRank) {
798		// I have both maximum and minimum, so proceed like a single
799		// processor version:
800		#endif
801	<	// objects to be swapped: velocity & angular velocity
801	>
802		Vector3d min_vel = min_sd->getVel();
803		Vector3d max_vel = max_sd->getVel();
804		RealType temp_vel;
805
806	<	switch(rnemdType_) {
807	<	case rnemdKineticSwap :
806	>	switch(rnemdFluxType_) {
807	>	case rnemdKE :
808		min_sd->setVel(max_vel);
809		max_sd->setVel(min_vel);
810	<	if (min_sd->isDirectional() && max_sd->isDirectional()) {
810	>	if (min_sd->isDirectional() && max_sd->isDirectional()) {
811		Vector3d min_angMom = min_sd->getJ();
812		Vector3d max_angMom = max_sd->getJ();
813		min_sd->setJ(max_angMom);
814		max_sd->setJ(min_angMom);
815	<	}
815	>	}//angular momenta exchange enabled
816	>	//assumes same rigid body identity
817		break;
818		case rnemdPx :
819		temp_vel = min_vel.x();
#	Line 420 | Line 839 | namespace OpenMD {
839		default :
840		break;
841		}
842	+
843		#ifdef IS_MPI
844		// the rest of the cases only apply in parallel simulations:
845		} else if (max_vals.rank == worldRank) {
#	Line 435 | Line 855 | namespace OpenMD {
855		min_vel.getArrayPointer(), 3, MPI::REALTYPE,
856		min_vals.rank, 0, status);
857
858	<	switch(rnemdType_) {
859	<	case rnemdKineticSwap :
858	>	switch(rnemdFluxType_) {
859	>	case rnemdKE :
860		max_sd->setVel(min_vel);
861	<
861	>	//angular momenta exchange enabled
862		if (max_sd->isDirectional()) {
863		Vector3d min_angMom;
864		Vector3d max_angMom = max_sd->getJ();
865	<
865	>
866		// point-to-point swap of the angular momentum vector
867		MPI::COMM_WORLD.Sendrecv(max_angMom.getArrayPointer(), 3,
868		MPI::REALTYPE, min_vals.rank, 1,
869		min_angMom.getArrayPointer(), 3,
870		MPI::REALTYPE, min_vals.rank, 1,
871		status);
872	<
872	>
873		max_sd->setJ(min_angMom);
874	<	}
874	>	}
875		break;
876		case rnemdPx :
877		max_vel.x() = min_vel.x();
#	Line 481 | Line 901 | namespace OpenMD {
901		max_vel.getArrayPointer(), 3, MPI::REALTYPE,
902		max_vals.rank, 0, status);
903
904	<	switch(rnemdType_) {
905	<	case rnemdKineticSwap :
904	>	switch(rnemdFluxType_) {
905	>	case rnemdKE :
906		min_sd->setVel(max_vel);
907	<
907	>	//angular momenta exchange enabled
908		if (min_sd->isDirectional()) {
909		Vector3d min_angMom = min_sd->getJ();
910		Vector3d max_angMom;
911	<
911	>
912		// point-to-point swap of the angular momentum vector
913		MPI::COMM_WORLD.Sendrecv(min_angMom.getArrayPointer(), 3,
914		MPI::REALTYPE, max_vals.rank, 1,
915		max_angMom.getArrayPointer(), 3,
916		MPI::REALTYPE, max_vals.rank, 1,
917		status);
918	<
918	>
919		min_sd->setJ(max_angMom);
920		}
921		break;
#	Line 516 | Line 936 | namespace OpenMD {
936		}
937		}
938		#endif
939	<	exchangeSum_ += max_val - min_val;
940	<	} else {
941	<	std::cerr << "exchange NOT performed!\nmin_val > max_val.\n";
939	>
940	>	switch(rnemdFluxType_) {
941	>	case rnemdKE:
942	>	kineticExchange_ += max_val - min_val;
943	>	break;
944	>	case rnemdPx:
945	>	momentumExchange_.x() += max_val - min_val;
946	>	break;
947	>	case rnemdPy:
948	>	momentumExchange_.y() += max_val - min_val;
949	>	break;
950	>	case rnemdPz:
951	>	momentumExchange_.z() += max_val - min_val;
952	>	break;
953	>	default:
954	>	break;
955	>	}
956	>	} else {
957	>	sprintf(painCave.errMsg,
958	>	"RNEMD::doSwap exchange NOT performed because min_val > max_val\n");
959	>	painCave.isFatal = 0;
960	>	painCave.severity = OPENMD_INFO;
961	>	simError();
962		failTrialCount_++;
963		}
964		} else {
965	<	std::cerr << "exchange NOT performed!\n";
966	<	std::cerr << "at least one of the two slabs empty.\n";
965	>	sprintf(painCave.errMsg,
966	>	"RNEMD::doSwap exchange NOT performed because selected object\n"
967	>	"\twas not present in at least one of the two slabs.\n");
968	>	painCave.isFatal = 0;
969	>	painCave.severity = OPENMD_INFO;
970	>	simError();
971		failTrialCount_++;
972	<	}
529	<
972	>	}
973		}
974
975	<	void RNEMD::doScale() {
975	>	void RNEMD::doNIVS(SelectionManager& smanA, SelectionManager& smanB) {
976	>	if (!doRNEMD_) return;
977	>	int selei;
978	>	int selej;
979
980		Snapshot* currentSnap_ = info_->getSnapshotManager()->getCurrentSnapshot();
981	+	RealType time = currentSnap_->getTime();
982		Mat3x3d hmat = currentSnap_->getHmat();
983
537	–	seleMan_.setSelectionSet(evaluator_.evaluate());
538	–
539	–	int selei;
984		StuntDouble* sd;
541	–	int idx;
985
986	<	std::vector<StuntDouble*> hotBin, coldBin;
986	>	vector<StuntDouble*> hotBin, coldBin;
987
988		RealType Phx = 0.0;
989		RealType Phy = 0.0;
#	Line 548 | Line 991 | namespace OpenMD {
991		RealType Khx = 0.0;
992		RealType Khy = 0.0;
993		RealType Khz = 0.0;
994	+	RealType Khw = 0.0;
995		RealType Pcx = 0.0;
996		RealType Pcy = 0.0;
997		RealType Pcz = 0.0;
998		RealType Kcx = 0.0;
999		RealType Kcy = 0.0;
1000		RealType Kcz = 0.0;
1001	+	RealType Kcw = 0.0;
1002
1003	<	for (sd = seleMan_.beginSelected(selei); sd != NULL;
1004	<	sd = seleMan_.nextSelected(selei)) {
1003	>	for (sd = smanA.beginSelected(selei); sd != NULL;
1004	>	sd = smanA.nextSelected(selei)) {
1005
561	–	idx = sd->getLocalIndex();
562	–
1006		Vector3d pos = sd->getPos();
1007	<
1007	>
1008		// wrap the stuntdouble's position back into the box:
1009	<
1009	>
1010		if (usePeriodicBoundaryConditions_)
1011		currentSnap_->wrapVector(pos);
1012	+
1013	+
1014	+	RealType mass = sd->getMass();
1015	+	Vector3d vel = sd->getVel();
1016	+
1017	+	hotBin.push_back(sd);
1018	+	Phx += mass * vel.x();
1019	+	Phy += mass * vel.y();
1020	+	Phz += mass * vel.z();
1021	+	Khx += mass * vel.x() * vel.x();
1022	+	Khy += mass * vel.y() * vel.y();
1023	+	Khz += mass * vel.z() * vel.z();
1024	+	if (sd->isDirectional()) {
1025	+	Vector3d angMom = sd->getJ();
1026	+	Mat3x3d I = sd->getI();
1027	+	if (sd->isLinear()) {
1028	+	int i = sd->linearAxis();
1029	+	int j = (i + 1) % 3;
1030	+	int k = (i + 2) % 3;
1031	+	Khw += angMom[j] * angMom[j] / I(j, j) +
1032	+	angMom[k] * angMom[k] / I(k, k);
1033	+	} else {
1034	+	Khw += angMom[0]*angMom[0]/I(0, 0)
1035	+	+ angMom[1]*angMom[1]/I(1, 1)
1036	+	+ angMom[2]*angMom[2]/I(2, 2);
1037	+	}
1038	+	}
1039	+	}
1040	+	for (sd = smanB.beginSelected(selej); sd != NULL;
1041	+	sd = smanB.nextSelected(selej)) {
1042	+	Vector3d pos = sd->getPos();
1043	+
1044	+	// wrap the stuntdouble's position back into the box:
1045	+
1046	+	if (usePeriodicBoundaryConditions_)
1047	+	currentSnap_->wrapVector(pos);
1048	+
1049	+	RealType mass = sd->getMass();
1050	+	Vector3d vel = sd->getVel();
1051
1052	<	// which bin is this stuntdouble in?
1053	<	// wrapped positions are in the range [-0.5*hmat(2,2), +0.5*hmat(2,2)]
1054	<
1055	<	int binNo = int(nBins_ * (pos.z() / hmat(2,2) + 0.5)) % nBins_;
1056	<
1057	<	// if we're in bin 0 or the middleBin
1058	<	if (binNo == 0 || binNo == midBin_) {
1059	<
1060	<	RealType mass = sd->getMass();
1061	<	Vector3d vel = sd->getVel();
1062	<
1063	<	if (binNo == 0) {
1064	<	hotBin.push_back(sd);
1065	<	Phx += mass * vel.x();
1066	<	Phy += mass * vel.y();
1067	<	Phz += mass * vel.z();
1068	<	Khx += mass * vel.x() * vel.x();
1069	<	Khy += mass * vel.y() * vel.y();
1070	<	Khz += mass * vel.z() * vel.z();
1071	<	} else { //midBin_
1072	<	coldBin.push_back(sd);
591	<	Pcx += mass * vel.x();
592	<	Pcy += mass * vel.y();
593	<	Pcz += mass * vel.z();
594	<	Kcx += mass * vel.x() * vel.x();
595	<	Kcy += mass * vel.y() * vel.y();
596	<	Kcz += mass * vel.z() * vel.z();
597	<	}
1052	>	coldBin.push_back(sd);
1053	>	Pcx += mass * vel.x();
1054	>	Pcy += mass * vel.y();
1055	>	Pcz += mass * vel.z();
1056	>	Kcx += mass * vel.x() * vel.x();
1057	>	Kcy += mass * vel.y() * vel.y();
1058	>	Kcz += mass * vel.z() * vel.z();
1059	>	if (sd->isDirectional()) {
1060	>	Vector3d angMom = sd->getJ();
1061	>	Mat3x3d I = sd->getI();
1062	>	if (sd->isLinear()) {
1063	>	int i = sd->linearAxis();
1064	>	int j = (i + 1) % 3;
1065	>	int k = (i + 2) % 3;
1066	>	Kcw += angMom[j] * angMom[j] / I(j, j) +
1067	>	angMom[k] * angMom[k] / I(k, k);
1068	>	} else {
1069	>	Kcw += angMom[0]*angMom[0]/I(0, 0)
1070	>	+ angMom[1]*angMom[1]/I(1, 1)
1071	>	+ angMom[2]*angMom[2]/I(2, 2);
1072	>	}
1073		}
1074		}
1075	<
1075	>
1076		Khx *= 0.5;
1077		Khy *= 0.5;
1078		Khz *= 0.5;
1079	+	Khw *= 0.5;
1080		Kcx *= 0.5;
1081		Kcy *= 0.5;
1082		Kcz *= 0.5;
1083	+	Kcw *= 0.5;
1084
1085		#ifdef IS_MPI
1086		MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Phx, 1, MPI::REALTYPE, MPI::SUM);
#	Line 616 | Line 1093 | namespace OpenMD {
1093		MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Khx, 1, MPI::REALTYPE, MPI::SUM);
1094		MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Khy, 1, MPI::REALTYPE, MPI::SUM);
1095		MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Khz, 1, MPI::REALTYPE, MPI::SUM);
1096	+	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Khw, 1, MPI::REALTYPE, MPI::SUM);
1097	+
1098		MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Kcx, 1, MPI::REALTYPE, MPI::SUM);
1099		MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Kcy, 1, MPI::REALTYPE, MPI::SUM);
1100		MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Kcz, 1, MPI::REALTYPE, MPI::SUM);
1101	+	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Kcw, 1, MPI::REALTYPE, MPI::SUM);
1102		#endif
1103
1104	<	//use coldBin coeff's
1104	>	//solve coldBin coeff's first
1105		RealType px = Pcx / Phx;
1106		RealType py = Pcy / Phy;
1107		RealType pz = Pcz / Phz;
1108	+	RealType c, x, y, z;
1109	+	bool successfulScale = false;
1110	+	if ((rnemdFluxType_ == rnemdFullKE) ||
1111	+	(rnemdFluxType_ == rnemdRotKE)) {
1112	+	//may need sanity check Khw & Kcw > 0
1113
1114	<	RealType a000, a110, c0, a001, a111, b01, b11, c1, c;
1115	<	switch(rnemdType_) {
1116	<	case rnemdKineticScale :
1117	<	/*used hotBin coeff's & only scale x & y dimensions
1118	<	RealType px = Phx / Pcx;
634	<	RealType py = Phy / Pcy;
635	<	a110 = Khy;
636	<	c0 = - Khx - Khy - targetFlux_;
637	<	a000 = Khx;
638	<	a111 = Kcy * py * py
639	<	b11 = -2.0 * Kcy * py * (1.0 + py);
640	<	c1 = Kcy * py * (2.0 + py) + Kcx * px * ( 2.0 + px) + targetFlux_;
641	<	b01 = -2.0 * Kcx * px * (1.0 + px);
642	<	a001 = Kcx * px * px;
643	<	*/
1114	>	if (rnemdFluxType_ == rnemdFullKE) {
1115	>	c = 1.0 - kineticTarget_ / (Kcx + Kcy + Kcz + Kcw);
1116	>	} else {
1117	>	c = 1.0 - kineticTarget_ / Kcw;
1118	>	}
1119
1120	<	//scale all three dimensions, let c_x = c_y
1121	<	a000 = Kcx + Kcy;
1122	<	a110 = Kcz;
1123	<	c0 = targetFlux_ - Kcx - Kcy - Kcz;
1124	<	a001 = Khx * px * px + Khy * py * py;
1125	<	a111 = Khz * pz * pz;
1126	<	b01 = -2.0 * (Khx * px * (1.0 + px) + Khy * py * (1.0 + py));
1127	<	b11 = -2.0 * Khz * pz * (1.0 + pz);
1128	<	c1 = Khx * px * (2.0 + px) + Khy * py * (2.0 + py)
1129	<	+ Khz * pz * (2.0 + pz) - targetFlux_;
1130	<	break;
1131	<	case rnemdPxScale :
1132	<	c = 1 - targetFlux_ / Pcx;
1133	<	a000 = Kcy;
1134	<	a110 = Kcz;
1135	<	c0 = Kcx * c * c - Kcx - Kcy - Kcz;
1136	<	a001 = py * py * Khy;
1137	<	a111 = pz * pz * Khz;
1138	<	b01 = -2.0 * Khy * py * (1.0 + py);
1139	<	b11 = -2.0 * Khz * pz * (1.0 + pz);
1140	<	c1 = Khy * py * (2.0 + py) + Khz * pz * (2.0 + pz)
1141	<	+ Khx * (fastpow(c * px - px - 1.0, 2) - 1.0);
1142	<	break;
1143	<	case rnemdPyScale :
1144	<	c = 1 - targetFlux_ / Pcy;
1145	<	a000 = Kcx;
1146	<	a110 = Kcz;
1147	<	c0 = Kcy * c * c - Kcx - Kcy - Kcz;
1148	<	a001 = px * px * Khx;
1149	<	a111 = pz * pz * Khz;
1150	<	b01 = -2.0 * Khx * px * (1.0 + px);
1151	<	b11 = -2.0 * Khz * pz * (1.0 + pz);
1152	<	c1 = Khx * px * (2.0 + px) + Khz * pz * (2.0 + pz)
1153	<	+ Khy * (fastpow(c * py - py - 1.0, 2) - 1.0);
1120	>	if ((c > 0.81) && (c < 1.21)) {//restrict scaling coefficients
1121	>	c = sqrt(c);
1122	>
1123	>	RealType w = 0.0;
1124	>	if (rnemdFluxType_ ==  rnemdFullKE) {
1125	>	x = 1.0 + px * (1.0 - c);
1126	>	y = 1.0 + py * (1.0 - c);
1127	>	z = 1.0 + pz * (1.0 - c);
1128	>	/* more complicated way
1129	>	w = 1.0 + (Kcw - Kcw * c * c - (c * c * (Kcx + Kcy + Kcz
1130	>	+ Khx * px * px + Khy * py * py + Khz * pz * pz)
1131	>	- 2.0 * c * (Khx * px * (1.0 + px) + Khy * py * (1.0 + py)
1132	>	+ Khz * pz * (1.0 + pz)) + Khx * px * (2.0 + px)
1133	>	+ Khy * py * (2.0 + py) + Khz * pz * (2.0 + pz)
1134	>	- Kcx - Kcy - Kcz)) / Khw; the following is simpler
1135	>	*/
1136	>	if ((fabs(x - 1.0) < 0.1) && (fabs(y - 1.0) < 0.1) &&
1137	>	(fabs(z - 1.0) < 0.1)) {
1138	>	w = 1.0 + (kineticTarget_
1139	>	+ Khx * (1.0 - x * x) + Khy * (1.0 - y * y)
1140	>	+ Khz * (1.0 - z * z)) / Khw;
1141	>	}//no need to calculate w if x, y or z is out of range
1142	>	} else {
1143	>	w = 1.0 + kineticTarget_ / Khw;
1144	>	}
1145	>	if ((w > 0.81) && (w < 1.21)) {//restrict scaling coefficients
1146	>	//if w is in the right range, so should be x, y, z.
1147	>	vector<StuntDouble*>::iterator sdi;
1148	>	Vector3d vel;
1149	>	for (sdi = coldBin.begin(); sdi != coldBin.end(); sdi++) {
1150	>	if (rnemdFluxType_ == rnemdFullKE) {
1151	>	vel = (*sdi)->getVel() * c;
1152	>	(*sdi)->setVel(vel);
1153	>	}
1154	>	if ((*sdi)->isDirectional()) {
1155	>	Vector3d angMom = (*sdi)->getJ() * c;
1156	>	(*sdi)->setJ(angMom);
1157	>	}
1158	>	}
1159	>	w = sqrt(w);
1160	>	for (sdi = hotBin.begin(); sdi != hotBin.end(); sdi++) {
1161	>	if (rnemdFluxType_ == rnemdFullKE) {
1162	>	vel = (*sdi)->getVel();
1163	>	vel.x() *= x;
1164	>	vel.y() *= y;
1165	>	vel.z() *= z;
1166	>	(*sdi)->setVel(vel);
1167	>	}
1168	>	if ((*sdi)->isDirectional()) {
1169	>	Vector3d angMom = (*sdi)->getJ() * w;
1170	>	(*sdi)->setJ(angMom);
1171	>	}
1172	>	}
1173	>	successfulScale = true;
1174	>	kineticExchange_ += kineticTarget_;
1175	>	}
1176	>	}
1177	>	} else {
1178	>	RealType a000, a110, c0, a001, a111, b01, b11, c1;
1179	>	switch(rnemdFluxType_) {
1180	>	case rnemdKE :
1181	>	/* used hotBin coeff's & only scale x & y dimensions
1182	>	RealType px = Phx / Pcx;
1183	>	RealType py = Phy / Pcy;
1184	>	a110 = Khy;
1185	>	c0 = - Khx - Khy - kineticTarget_;
1186	>	a000 = Khx;
1187	>	a111 = Kcy * py * py;
1188	>	b11 = -2.0 * Kcy * py * (1.0 + py);
1189	>	c1 = Kcy * py * (2.0 + py) + Kcx * px * ( 2.0 + px) + kineticTarget_;
1190	>	b01 = -2.0 * Kcx * px * (1.0 + px);
1191	>	a001 = Kcx * px * px;
1192	>	*/
1193	>	//scale all three dimensions, let c_x = c_y
1194	>	a000 = Kcx + Kcy;
1195	>	a110 = Kcz;
1196	>	c0 = kineticTarget_ - Kcx - Kcy - Kcz;
1197	>	a001 = Khx * px * px + Khy * py * py;
1198	>	a111 = Khz * pz * pz;
1199	>	b01 = -2.0 * (Khx * px * (1.0 + px) + Khy * py * (1.0 + py));
1200	>	b11 = -2.0 * Khz * pz * (1.0 + pz);
1201	>	c1 = Khx * px * (2.0 + px) + Khy * py * (2.0 + py)
1202	>	+ Khz * pz * (2.0 + pz) - kineticTarget_;
1203	>	break;
1204	>	case rnemdPx :
1205	>	c = 1 - momentumTarget_.x() / Pcx;
1206	>	a000 = Kcy;
1207	>	a110 = Kcz;
1208	>	c0 = Kcx * c * c - Kcx - Kcy - Kcz;
1209	>	a001 = py * py * Khy;
1210	>	a111 = pz * pz * Khz;
1211	>	b01 = -2.0 * Khy * py * (1.0 + py);
1212	>	b11 = -2.0 * Khz * pz * (1.0 + pz);
1213	>	c1 = Khy * py * (2.0 + py) + Khz * pz * (2.0 + pz)
1214	>	+ Khx * (fastpow(c * px - px - 1.0, 2) - 1.0);
1215	>	break;
1216	>	case rnemdPy :
1217	>	c = 1 - momentumTarget_.y() / Pcy;
1218	>	a000 = Kcx;
1219	>	a110 = Kcz;
1220	>	c0 = Kcy * c * c - Kcx - Kcy - Kcz;
1221	>	a001 = px * px * Khx;
1222	>	a111 = pz * pz * Khz;
1223	>	b01 = -2.0 * Khx * px * (1.0 + px);
1224	>	b11 = -2.0 * Khz * pz * (1.0 + pz);
1225	>	c1 = Khx * px * (2.0 + px) + Khz * pz * (2.0 + pz)
1226	>	+ Khy * (fastpow(c * py - py - 1.0, 2) - 1.0);
1227	>	break;
1228	>	case rnemdPz ://we don't really do this, do we?
1229	>	c = 1 - momentumTarget_.z() / Pcz;
1230	>	a000 = Kcx;
1231	>	a110 = Kcy;
1232	>	c0 = Kcz * c * c - Kcx - Kcy - Kcz;
1233	>	a001 = px * px * Khx;
1234	>	a111 = py * py * Khy;
1235	>	b01 = -2.0 * Khx * px * (1.0 + px);
1236	>	b11 = -2.0 * Khy * py * (1.0 + py);
1237	>	c1 = Khx * px * (2.0 + px) + Khy * py * (2.0 + py)
1238	>	+ Khz * (fastpow(c * pz - pz - 1.0, 2) - 1.0);
1239	>	break;
1240	>	default :
1241	>	break;
1242	>	}
1243	>
1244	>	RealType v1 = a000 * a111 - a001 * a110;
1245	>	RealType v2 = a000 * b01;
1246	>	RealType v3 = a000 * b11;
1247	>	RealType v4 = a000 * c1 - a001 * c0;
1248	>	RealType v8 = a110 * b01;
1249	>	RealType v10 = - b01 * c0;
1250	>
1251	>	RealType u0 = v2 * v10 - v4 * v4;
1252	>	RealType u1 = -2.0 * v3 * v4;
1253	>	RealType u2 = -v2 * v8 - v3 * v3 - 2.0 * v1 * v4;
1254	>	RealType u3 = -2.0 * v1 * v3;
1255	>	RealType u4 = - v1 * v1;
1256	>	//rescale coefficients
1257	>	RealType maxAbs = fabs(u0);
1258	>	if (maxAbs < fabs(u1)) maxAbs = fabs(u1);
1259	>	if (maxAbs < fabs(u2)) maxAbs = fabs(u2);
1260	>	if (maxAbs < fabs(u3)) maxAbs = fabs(u3);
1261	>	if (maxAbs < fabs(u4)) maxAbs = fabs(u4);
1262	>	u0 /= maxAbs;
1263	>	u1 /= maxAbs;
1264	>	u2 /= maxAbs;
1265	>	u3 /= maxAbs;
1266	>	u4 /= maxAbs;
1267	>	//max_element(start, end) is also available.
1268	>	Polynomial<RealType> poly; //same as DoublePolynomial poly;
1269	>	poly.setCoefficient(4, u4);
1270	>	poly.setCoefficient(3, u3);
1271	>	poly.setCoefficient(2, u2);
1272	>	poly.setCoefficient(1, u1);
1273	>	poly.setCoefficient(0, u0);
1274	>	vector<RealType> realRoots = poly.FindRealRoots();
1275	>
1276	>	vector<RealType>::iterator ri;
1277	>	RealType r1, r2, alpha0;
1278	>	vector<pair<RealType,RealType> > rps;
1279	>	for (ri = realRoots.begin(); ri !=realRoots.end(); ri++) {
1280	>	r2 = *ri;
1281	>	//check if FindRealRoots() give the right answer
1282	>	if ( fabs(u0 + r2 * (u1 + r2 * (u2 + r2 * (u3 + r2 * u4)))) > 1e-6 ) {
1283	>	sprintf(painCave.errMsg,
1284	>	"RNEMD Warning: polynomial solve seems to have an error!");
1285	>	painCave.isFatal = 0;
1286	>	simError();
1287	>	failRootCount_++;
1288	>	}
1289	>	//might not be useful w/o rescaling coefficients
1290	>	alpha0 = -c0 - a110 * r2 * r2;
1291	>	if (alpha0 >= 0.0) {
1292	>	r1 = sqrt(alpha0 / a000);
1293	>	if (fabs(c1 + r1 * (b01 + r1 * a001) + r2 * (b11 + r2 * a111))
1294	>	< 1e-6)
1295	>	{ rps.push_back(make_pair(r1, r2)); }
1296	>	if (r1 > 1e-6) { //r1 non-negative
1297	>	r1 = -r1;
1298	>	if (fabs(c1 + r1 * (b01 + r1 * a001) + r2 * (b11 + r2 * a111))
1299	>	< 1e-6)
1300	>	{ rps.push_back(make_pair(r1, r2)); }
1301	>	}
1302	>	}
1303	>	}
1304	>	// Consider combining together the solving pair part w/ the searching
1305	>	// best solution part so that we don't need the pairs vector
1306	>	if (!rps.empty()) {
1307	>	RealType smallestDiff = HONKING_LARGE_VALUE;
1308	>	RealType diff;
1309	>	pair<RealType,RealType> bestPair = make_pair(1.0, 1.0);
1310	>	vector<pair<RealType,RealType> >::iterator rpi;
1311	>	for (rpi = rps.begin(); rpi != rps.end(); rpi++) {
1312	>	r1 = (*rpi).first;
1313	>	r2 = (*rpi).second;
1314	>	switch(rnemdFluxType_) {
1315	>	case rnemdKE :
1316	>	diff = fastpow(1.0 - r1, 2) + fastpow(1.0 - r2, 2)
1317	>	+ fastpow(r1 * r1 / r2 / r2 - Kcz/Kcx, 2)
1318	>	+ fastpow(r1 * r1 / r2 / r2 - Kcz/Kcy, 2);
1319	>	break;
1320	>	case rnemdPx :
1321	>	diff = fastpow(1.0 - r1, 2) + fastpow(1.0 - r2, 2)
1322	>	+ fastpow(r1 * r1 / r2 / r2 - Kcz/Kcy, 2);
1323	>	break;
1324	>	case rnemdPy :
1325	>	diff = fastpow(1.0 - r1, 2) + fastpow(1.0 - r2, 2)
1326	>	+ fastpow(r1 * r1 / r2 / r2 - Kcz/Kcx, 2);
1327	>	break;
1328	>	case rnemdPz :
1329	>	diff = fastpow(1.0 - r1, 2) + fastpow(1.0 - r2, 2)
1330	>	+ fastpow(r1 * r1 / r2 / r2 - Kcy/Kcx, 2);
1331	>	default :
1332	>	break;
1333	>	}
1334	>	if (diff < smallestDiff) {
1335	>	smallestDiff = diff;
1336	>	bestPair = *rpi;
1337	>	}
1338	>	}
1339	>	#ifdef IS_MPI
1340	>	if (worldRank == 0) {
1341	>	#endif
1342	>	// sprintf(painCave.errMsg,
1343	>	//         "RNEMD: roots r1= %lf\tr2 = %lf\n",
1344	>	//         bestPair.first, bestPair.second);
1345	>	// painCave.isFatal = 0;
1346	>	// painCave.severity = OPENMD_INFO;
1347	>	// simError();
1348	>	#ifdef IS_MPI
1349	>	}
1350	>	#endif
1351	>
1352	>	switch(rnemdFluxType_) {
1353	>	case rnemdKE :
1354	>	x = bestPair.first;
1355	>	y = bestPair.first;
1356	>	z = bestPair.second;
1357	>	break;
1358	>	case rnemdPx :
1359	>	x = c;
1360	>	y = bestPair.first;
1361	>	z = bestPair.second;
1362	>	break;
1363	>	case rnemdPy :
1364	>	x = bestPair.first;
1365	>	y = c;
1366	>	z = bestPair.second;
1367	>	break;
1368	>	case rnemdPz :
1369	>	x = bestPair.first;
1370	>	y = bestPair.second;
1371	>	z = c;
1372	>	break;
1373	>	default :
1374	>	break;
1375	>	}
1376	>	vector<StuntDouble*>::iterator sdi;
1377	>	Vector3d vel;
1378	>	for (sdi = coldBin.begin(); sdi != coldBin.end(); sdi++) {
1379	>	vel = (*sdi)->getVel();
1380	>	vel.x() *= x;
1381	>	vel.y() *= y;
1382	>	vel.z() *= z;
1383	>	(*sdi)->setVel(vel);
1384	>	}
1385	>	//convert to hotBin coefficient
1386	>	x = 1.0 + px * (1.0 - x);
1387	>	y = 1.0 + py * (1.0 - y);
1388	>	z = 1.0 + pz * (1.0 - z);
1389	>	for (sdi = hotBin.begin(); sdi != hotBin.end(); sdi++) {
1390	>	vel = (*sdi)->getVel();
1391	>	vel.x() *= x;
1392	>	vel.y() *= y;
1393	>	vel.z() *= z;
1394	>	(*sdi)->setVel(vel);
1395	>	}
1396	>	successfulScale = true;
1397	>	switch(rnemdFluxType_) {
1398	>	case rnemdKE :
1399	>	kineticExchange_ += kineticTarget_;
1400	>	break;
1401	>	case rnemdPx :
1402	>	case rnemdPy :
1403	>	case rnemdPz :
1404	>	momentumExchange_ += momentumTarget_;
1405	>	break;
1406	>	default :
1407	>	break;
1408	>	}
1409	>	}
1410	>	}
1411	>	if (successfulScale != true) {
1412	>	sprintf(painCave.errMsg,
1413	>	"RNEMD::doNIVS exchange NOT performed - roots that solve\n"
1414	>	"\tthe constraint equations may not exist or there may be\n"
1415	>	"\tno selected objects in one or both slabs.\n");
1416	>	painCave.isFatal = 0;
1417	>	painCave.severity = OPENMD_INFO;
1418	>	simError();
1419	>	failTrialCount_++;
1420	>	}
1421	>	}
1422	>
1423	>	void RNEMD::doVSS(SelectionManager& smanA, SelectionManager& smanB) {
1424	>	if (!doRNEMD_) return;
1425	>	int selei;
1426	>	int selej;
1427	>
1428	>	Snapshot* currentSnap_ = info_->getSnapshotManager()->getCurrentSnapshot();
1429	>	RealType time = currentSnap_->getTime();
1430	>	Mat3x3d hmat = currentSnap_->getHmat();
1431	>
1432	>	StuntDouble* sd;
1433	>
1434	>	vector<StuntDouble*> hotBin, coldBin;
1435	>
1436	>	Vector3d Ph(V3Zero);
1437	>	Vector3d Lh(V3Zero);
1438	>	RealType Mh = 0.0;
1439	>	Mat3x3d Ih(0.0);
1440	>	RealType Kh = 0.0;
1441	>	Vector3d Pc(V3Zero);
1442	>	Vector3d Lc(V3Zero);
1443	>	RealType Mc = 0.0;
1444	>	Mat3x3d Ic(0.0);
1445	>	RealType Kc = 0.0;
1446	>
1447	>	// Constraints can be on only the linear or angular momentum, but
1448	>	// not both.  Usually, the user will specify which they want, but
1449	>	// in case they don't, the use of periodic boundaries should make
1450	>	// the choice for us.
1451	>	bool doLinearPart = false;
1452	>	bool doAngularPart = false;
1453	>
1454	>	switch (rnemdFluxType_) {
1455	>	case rnemdPx:
1456	>	case rnemdPy:
1457	>	case rnemdPz:
1458	>	case rnemdPvector:
1459	>	case rnemdKePx:
1460	>	case rnemdKePy:
1461	>	case rnemdKePvector:
1462	>	doLinearPart = true;
1463		break;
1464	<	case rnemdPzScale ://we don't really do this, do we?
1465	<	c = 1 - targetFlux_ / Pcz;
1466	<	a000 = Kcx;
1467	<	a110 = Kcy;
1468	<	c0 = Kcz * c * c - Kcx - Kcy - Kcz;
1469	<	a001 = px * px * Khx;
1470	<	a111 = py * py * Khy;
1471	<	b01 = -2.0 * Khx * px * (1.0 + px);
1472	<	b11 = -2.0 * Khy * py * (1.0 + py);
689	<	c1 = Khx * px * (2.0 + px) + Khy * py * (2.0 + py)
690	<	+ Khz * (fastpow(c * pz - pz - 1.0, 2) - 1.0);
691	<	break;
692	<	default :
1464	>	case rnemdLx:
1465	>	case rnemdLy:
1466	>	case rnemdLz:
1467	>	case rnemdLvector:
1468	>	case rnemdKeLx:
1469	>	case rnemdKeLy:
1470	>	case rnemdKeLz:
1471	>	case rnemdKeLvector:
1472	>	doAngularPart = true;
1473		break;
1474	+	case rnemdKE:
1475	+	case rnemdRotKE:
1476	+	case rnemdFullKE:
1477	+	default:
1478	+	if (usePeriodicBoundaryConditions_)
1479	+	doLinearPart = true;
1480	+	else
1481	+	doAngularPart = true;
1482	+	break;
1483		}
1484	+
1485	+	for (sd = smanA.beginSelected(selei); sd != NULL;
1486	+	sd = smanA.nextSelected(selei)) {
1487
1488	<	RealType v1 = a000 * a111 - a001 * a110;
697	<	RealType v2 = a000 * b01;
698	<	RealType v3 = a000 * b11;
699	<	RealType v4 = a000 * c1 - a001 * c0;
700	<	RealType v8 = a110 * b01;
701	<	RealType v10 = - b01 * c0;
1488	>	Vector3d pos = sd->getPos();
1489
1490	<	RealType u0 = v2 * v10 - v4 * v4;
1491	<	RealType u1 = -2.0 * v3 * v4;
1492	<	RealType u2 = -v2 * v8 - v3 * v3 - 2.0 * v1 * v4;
1493	<	RealType u3 = -2.0 * v1 * v3;
1494	<	RealType u4 = - v1 * v1;
1495	<	//rescale coefficients
1496	<	RealType maxAbs = fabs(u0);
1497	<	if (maxAbs < fabs(u1)) maxAbs = fabs(u1);
1498	<	if (maxAbs < fabs(u2)) maxAbs = fabs(u2);
1499	<	if (maxAbs < fabs(u3)) maxAbs = fabs(u3);
1500	<	if (maxAbs < fabs(u4)) maxAbs = fabs(u4);
1501	<	u0 /= maxAbs;
1502	<	u1 /= maxAbs;
1503	<	u2 /= maxAbs;
1504	<	u3 /= maxAbs;
1505	<	u4 /= maxAbs;
1506	<	//max_element(start, end) is also available.
1507	<	Polynomial<RealType> poly; //same as DoublePolynomial poly;
1508	<	poly.setCoefficient(4, u4);
1509	<	poly.setCoefficient(3, u3);
1510	<	poly.setCoefficient(2, u2);
1511	<	poly.setCoefficient(1, u1);
1512	<	poly.setCoefficient(0, u0);
1513	<	std::vector<RealType> realRoots = poly.FindRealRoots();
1514	<
1515	<	std::vector<RealType>::iterator ri;
1516	<	RealType r1, r2, alpha0;
1517	<	std::vector<std::pair<RealType,RealType> > rps;
1518	<	for (ri = realRoots.begin(); ri !=realRoots.end(); ri++) {
1519	<	r2 = *ri;
1520	<	//check if FindRealRoots() give the right answer
1521	<	if ( fabs(u0 + r2 * (u1 + r2 * (u2 + r2 * (u3 + r2 * u4)))) > 1e-6 ) {
1522	<	sprintf(painCave.errMsg,
1523	<	"RNEMD Warning: polynomial solve seems to have an error!");
1524	<	painCave.isFatal = 0;
1525	<	simError();
739	<	failRootCount_++;
740	<	}
741	<	//might not be useful w/o rescaling coefficients
742	<	alpha0 = -c0 - a110 * r2 * r2;
743	<	if (alpha0 >= 0.0) {
744	<	r1 = sqrt(alpha0 / a000);
745	<	if (fabs(c1 + r1 * (b01 + r1 * a001) + r2 * (b11 + r2 * a111)) < 1e-6)
746	<	{ rps.push_back(std::make_pair(r1, r2)); }
747	<	if (r1 > 1e-6) { //r1 non-negative
748	<	r1 = -r1;
749	<	if (fabs(c1 + r1 * (b01 + r1 * a001) + r2 * (b11 + r2 * a111)) <1e-6)
750	<	{ rps.push_back(std::make_pair(r1, r2)); }
1490	>	// wrap the stuntdouble's position back into the box:
1491	>
1492	>	if (usePeriodicBoundaryConditions_)
1493	>	currentSnap_->wrapVector(pos);
1494	>
1495	>	RealType mass = sd->getMass();
1496	>	Vector3d vel = sd->getVel();
1497	>	Vector3d rPos = sd->getPos() - coordinateOrigin_;
1498	>	RealType r2;
1499	>
1500	>	hotBin.push_back(sd);
1501	>	Ph += mass * vel;
1502	>	Mh += mass;
1503	>	Kh += mass * vel.lengthSquare();
1504	>	Lh += mass * cross(rPos, vel);
1505	>	Ih -= outProduct(rPos, rPos) * mass;
1506	>	r2 = rPos.lengthSquare();
1507	>	Ih(0, 0) += mass * r2;
1508	>	Ih(1, 1) += mass * r2;
1509	>	Ih(2, 2) += mass * r2;
1510	>
1511	>	if (rnemdFluxType_ == rnemdFullKE) {
1512	>	if (sd->isDirectional()) {
1513	>	Vector3d angMom = sd->getJ();
1514	>	Mat3x3d I = sd->getI();
1515	>	if (sd->isLinear()) {
1516	>	int i = sd->linearAxis();
1517	>	int j = (i + 1) % 3;
1518	>	int k = (i + 2) % 3;
1519	>	Kh += angMom[j] * angMom[j] / I(j, j) +
1520	>	angMom[k] * angMom[k] / I(k, k);
1521	>	} else {
1522	>	Kh += angMom[0] * angMom[0] / I(0, 0) +
1523	>	angMom[1] * angMom[1] / I(1, 1) +
1524	>	angMom[2] * angMom[2] / I(2, 2);
1525	>	}
1526		}
1527		}
1528		}
1529	<	// Consider combininig together the solving pair part w/ the searching
1530	<	// best solution part so that we don't need the pairs vector
1531	<	if (!rps.empty()) {
1532	<	RealType smallestDiff = HONKING_LARGE_VALUE;
1533	<	RealType diff;
1534	<	std::pair<RealType,RealType> bestPair = std::make_pair(1.0, 1.0);
1535	<	std::vector<std::pair<RealType,RealType> >::iterator rpi;
1536	<	for (rpi = rps.begin(); rpi != rps.end(); rpi++) {
1537	<	r1 = (*rpi).first;
1538	<	r2 = (*rpi).second;
1539	<	switch(rnemdType_) {
1540	<	case rnemdKineticScale :
1541	<	diff = fastpow(1.0 - r1, 2) + fastpow(1.0 - r2, 2)
1542	<	+ fastpow(r1 * r1 / r2 / r2 - Kcz/Kcx, 2)
1543	<	+ fastpow(r1 * r1 / r2 / r2 - Kcz/Kcy, 2);
1544	<	break;
1545	<	case rnemdPxScale :
1546	<	diff = fastpow(1.0 - r1, 2) + fastpow(1.0 - r2, 2)
1547	<	+ fastpow(r1 * r1 / r2 / r2 - Kcz/Kcy, 2);
1548	<	break;
1549	<	case rnemdPyScale :
1550	<	diff = fastpow(1.0 - r1, 2) + fastpow(1.0 - r2, 2)
1551	<	+ fastpow(r1 * r1 / r2 / r2 - Kcz/Kcx, 2);
1552	<	break;
1553	<	case rnemdPzScale :
1554	<	default :
1555	<	break;
1529	>	for (sd = smanB.beginSelected(selej); sd != NULL;
1530	>	sd = smanB.nextSelected(selej)) {
1531	>
1532	>	Vector3d pos = sd->getPos();
1533	>
1534	>	// wrap the stuntdouble's position back into the box:
1535	>
1536	>	if (usePeriodicBoundaryConditions_)
1537	>	currentSnap_->wrapVector(pos);
1538	>
1539	>	RealType mass = sd->getMass();
1540	>	Vector3d vel = sd->getVel();
1541	>	Vector3d rPos = sd->getPos() - coordinateOrigin_;
1542	>	RealType r2;
1543	>
1544	>	coldBin.push_back(sd);
1545	>	Pc += mass * vel;
1546	>	Mc += mass;
1547	>	Kc += mass * vel.lengthSquare();
1548	>	Lc += mass * cross(rPos, vel);
1549	>	Ic -= outProduct(rPos, rPos) * mass;
1550	>	r2 = rPos.lengthSquare();
1551	>	Ic(0, 0) += mass * r2;
1552	>	Ic(1, 1) += mass * r2;
1553	>	Ic(2, 2) += mass * r2;
1554	>
1555	>	if (rnemdFluxType_ == rnemdFullKE) {
1556	>	if (sd->isDirectional()) {
1557	>	Vector3d angMom = sd->getJ();
1558	>	Mat3x3d I = sd->getI();
1559	>	if (sd->isLinear()) {
1560	>	int i = sd->linearAxis();
1561	>	int j = (i + 1) % 3;
1562	>	int k = (i + 2) % 3;
1563	>	Kc += angMom[j] * angMom[j] / I(j, j) +
1564	>	angMom[k] * angMom[k] / I(k, k);
1565	>	} else {
1566	>	Kc += angMom[0] * angMom[0] / I(0, 0) +
1567	>	angMom[1] * angMom[1] / I(1, 1) +
1568	>	angMom[2] * angMom[2] / I(2, 2);
1569	>	}
1570		}
782	–	if (diff < smallestDiff) {
783	–	smallestDiff = diff;
784	–	bestPair = *rpi;
785	–	}
1571		}
1572	+	}
1573	+
1574	+	Kh *= 0.5;
1575	+	Kc *= 0.5;
1576	+
1577		#ifdef IS_MPI
1578	<	if (worldRank == 0) {
1578	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Ph[0], 3, MPI::REALTYPE, MPI::SUM);
1579	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Pc[0], 3, MPI::REALTYPE, MPI::SUM);
1580	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Lh[0], 3, MPI::REALTYPE, MPI::SUM);
1581	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Lc[0], 3, MPI::REALTYPE, MPI::SUM);
1582	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Mh, 1, MPI::REALTYPE, MPI::SUM);
1583	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Kh, 1, MPI::REALTYPE, MPI::SUM);
1584	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Mc, 1, MPI::REALTYPE, MPI::SUM);
1585	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Kc, 1, MPI::REALTYPE, MPI::SUM);
1586	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, Ih.getArrayPointer(), 9,
1587	>	MPI::REALTYPE, MPI::SUM);
1588	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, Ic.getArrayPointer(), 9,
1589	>	MPI::REALTYPE, MPI::SUM);
1590		#endif
1591	<	std::cerr << "we choose r1 = " << bestPair.first
791	<	<< " and r2 = " << bestPair.second << "\n";
792	<	#ifdef IS_MPI
793	<	}
794	<	#endif
1591	>
1592
1593	<	RealType x, y, z;
1594	<	switch(rnemdType_) {
1595	<	case rnemdKineticScale :
1596	<	x = bestPair.first;
1597	<	y = bestPair.first;
1598	<	z = bestPair.second;
1599	<	break;
1600	<	case rnemdPxScale :
1601	<	x = c;
1602	<	y = bestPair.first;
1603	<	z = bestPair.second;
1604	<	break;
1605	<	case rnemdPyScale :
1606	<	x = bestPair.first;
1607	<	y = c;
1608	<	z = bestPair.second;
1609	<	break;
1610	<	case rnemdPzScale :
1611	<	x = bestPair.first;
1612	<	y = bestPair.second;
1613	<	z = c;
1614	<	break;
1615	<	default :
1616	<	break;
1617	<	}
1618	<	std::vector<StuntDouble*>::iterator sdi;
1619	<	Vector3d vel;
1620	<	for (sdi = coldBin.begin(); sdi != coldBin.end(); sdi++) {
1621	<	vel = (*sdi)->getVel();
1622	<	vel.x() *= x;
1623	<	vel.y() *= y;
1624	<	vel.z() *= z;
1625	<	(*sdi)->setVel(vel);
1626	<	}
1627	<	//convert to hotBin coefficient
1628	<	x = 1.0 + px * (1.0 - x);
1629	<	y = 1.0 + py * (1.0 - y);
1630	<	z = 1.0 + pz * (1.0 - z);
1631	<	for (sdi = hotBin.begin(); sdi != hotBin.end(); sdi++) {
1632	<	vel = (*sdi)->getVel();
1633	<	vel.x() *= x;
1634	<	vel.y() *= y;
1635	<	vel.z() *= z;
1636	<	(*sdi)->setVel(vel);
1593	>	Vector3d ac, acrec, bc, bcrec;
1594	>	Vector3d ah, ahrec, bh, bhrec;
1595	>	RealType cNumerator, cDenominator;
1596	>	RealType hNumerator, hDenominator;
1597	>
1598	>
1599	>	bool successfulExchange = false;
1600	>	if ((Mh > 0.0) && (Mc > 0.0)) {//both slabs are not empty
1601	>	Vector3d vc = Pc / Mc;
1602	>	ac = -momentumTarget_ / Mc + vc;
1603	>	acrec = -momentumTarget_ / Mc;
1604	>
1605	>	// We now need the inverse of the inertia tensor to calculate the
1606	>	// angular velocity of the cold slab;
1607	>	Mat3x3d Ici = Ic.inverse();
1608	>	Vector3d omegac = Ici * Lc;
1609	>	bc  = -(Ici * angularMomentumTarget_) + omegac;
1610	>	bcrec = bc - omegac;
1611	>
1612	>	cNumerator = Kc - kineticTarget_;
1613	>	if (doLinearPart)
1614	>	cNumerator -= 0.5 * Mc * ac.lengthSquare();
1615	>
1616	>	if (doAngularPart)
1617	>	cNumerator -= 0.5 * ( dot(bc, Ic * bc));
1618	>
1619	>	if (cNumerator > 0.0) {
1620	>
1621	>	cDenominator = Kc;
1622	>
1623	>	if (doLinearPart)
1624	>	cDenominator -= 0.5 * Mc * vc.lengthSquare();
1625	>
1626	>	if (doAngularPart)
1627	>	cDenominator -= 0.5*(dot(omegac, Ic * omegac));
1628	>
1629	>	if (cDenominator > 0.0) {
1630	>	RealType c = sqrt(cNumerator / cDenominator);
1631	>	if ((c > 0.9) && (c < 1.1)) {//restrict scaling coefficients
1632	>
1633	>	Vector3d vh = Ph / Mh;
1634	>	ah = momentumTarget_ / Mh + vh;
1635	>	ahrec = momentumTarget_ / Mh;
1636	>
1637	>	// We now need the inverse of the inertia tensor to
1638	>	// calculate the angular velocity of the hot slab;
1639	>	Mat3x3d Ihi = Ih.inverse();
1640	>	Vector3d omegah = Ihi * Lh;
1641	>	bh  = (Ihi * angularMomentumTarget_) + omegah;
1642	>	bhrec = bh - omegah;
1643	>
1644	>	hNumerator = Kh + kineticTarget_;
1645	>	if (doLinearPart)
1646	>	hNumerator -= 0.5 * Mh * ah.lengthSquare();
1647	>
1648	>	if (doAngularPart)
1649	>	hNumerator -= 0.5 * ( dot(bh, Ih * bh));
1650	>
1651	>	if (hNumerator > 0.0) {
1652	>
1653	>	hDenominator = Kh;
1654	>	if (doLinearPart)
1655	>	hDenominator -= 0.5 * Mh * vh.lengthSquare();
1656	>	if (doAngularPart)
1657	>	hDenominator -= 0.5*(dot(omegah, Ih * omegah));
1658	>
1659	>	if (hDenominator > 0.0) {
1660	>	RealType h = sqrt(hNumerator / hDenominator);
1661	>	if ((h > 0.9) && (h < 1.1)) {
1662	>
1663	>	vector<StuntDouble*>::iterator sdi;
1664	>	Vector3d vel;
1665	>	Vector3d rPos;
1666	>
1667	>	for (sdi = coldBin.begin(); sdi != coldBin.end(); sdi++) {
1668	>	//vel = (*sdi)->getVel();
1669	>	rPos = (*sdi)->getPos() - coordinateOrigin_;
1670	>	if (doLinearPart)
1671	>	vel = ((*sdi)->getVel() - vc) * c + ac;
1672	>	if (doAngularPart)
1673	>	vel = ((*sdi)->getVel() - cross(omegac, rPos)) * c + cross(bc, rPos);
1674	>
1675	>	(*sdi)->setVel(vel);
1676	>	if (rnemdFluxType_ == rnemdFullKE) {
1677	>	if ((*sdi)->isDirectional()) {
1678	>	Vector3d angMom = (*sdi)->getJ() * c;
1679	>	(*sdi)->setJ(angMom);
1680	>	}
1681	>	}
1682	>	}
1683	>	for (sdi = hotBin.begin(); sdi != hotBin.end(); sdi++) {
1684	>	//vel = (*sdi)->getVel();
1685	>	rPos = (*sdi)->getPos() - coordinateOrigin_;
1686	>	if (doLinearPart)
1687	>	vel = ((*sdi)->getVel() - vh) * h + ah;
1688	>	if (doAngularPart)
1689	>	vel = ((*sdi)->getVel() - cross(omegah, rPos)) * h + cross(bh, rPos);
1690	>
1691	>	(*sdi)->setVel(vel);
1692	>	if (rnemdFluxType_ == rnemdFullKE) {
1693	>	if ((*sdi)->isDirectional()) {
1694	>	Vector3d angMom = (*sdi)->getJ() * h;
1695	>	(*sdi)->setJ(angMom);
1696	>	}
1697	>	}
1698	>	}
1699	>	successfulExchange = true;
1700	>	kineticExchange_ += kineticTarget_;
1701	>	momentumExchange_ += momentumTarget_;
1702	>	angularMomentumExchange_ += angularMomentumTarget_;
1703	>	}
1704	>	}
1705	>	}
1706	>	}
1707	>	}
1708		}
1709	<	exchangeSum_ += targetFlux_;
1710	<	//we may want to check whether the exchange has been successful
1711	<	} else {
1712	<	std::cerr << "exchange NOT performed!\n";//MPI incompatible
1709	>	}
1710	>	if (successfulExchange != true) {
1711	>	sprintf(painCave.errMsg,
1712	>	"RNEMD::doVSS exchange NOT performed - roots that solve\n"
1713	>	"\tthe constraint equations may not exist or there may be\n"
1714	>	"\tno selected objects in one or both slabs.\n");
1715	>	painCave.isFatal = 0;
1716	>	painCave.severity = OPENMD_INFO;
1717	>	simError();
1718		failTrialCount_++;
1719		}
847	–
1720		}
1721
1722	+	RealType RNEMD::getDividingArea() {
1723	+
1724	+	if (hasDividingArea_) return dividingArea_;
1725	+
1726	+	RealType areaA, areaB;
1727	+	Snapshot* snap = info_->getSnapshotManager()->getCurrentSnapshot();
1728	+
1729	+	if (hasSelectionA_) {
1730	+	int isd;
1731	+	StuntDouble* sd;
1732	+	vector<StuntDouble*> aSites;
1733	+	ConvexHull* surfaceMeshA = new ConvexHull();
1734	+	seleManA_.setSelectionSet(evaluatorA_.evaluate());
1735	+	for (sd = seleManA_.beginSelected(isd); sd != NULL;
1736	+	sd = seleManA_.nextSelected(isd)) {
1737	+	aSites.push_back(sd);
1738	+	}
1739	+
1740	+	surfaceMeshA->computeHull(aSites);
1741	+	areaA = surfaceMeshA->getArea();
1742	+	} else {
1743	+	if (usePeriodicBoundaryConditions_) {
1744	+	// in periodic boundaries, the surface area is twice the x-y
1745	+	// area of the current box:
1746	+	areaA = 2.0 * snap->getXYarea();
1747	+	} else {
1748	+	// in non-periodic simulations, without explicitly setting
1749	+	// selections, the sphere radius sets the surface area of the
1750	+	// dividing surface:
1751	+	areaA = 4.0 * M_PI * pow(sphereARadius_, 2);
1752	+	}
1753	+	}
1754	+
1755	+
1756	+
1757	+	if (hasSelectionB_) {
1758	+	int isd;
1759	+	StuntDouble* sd;
1760	+	vector<StuntDouble*> bSites;
1761	+
1762	+	seleManB_.setSelectionSet(evaluatorB_.evaluate());
1763	+
1764	+	for (sd = seleManB_.beginSelected(isd); sd != NULL;
1765	+	sd = seleManB_.nextSelected(isd)) {
1766	+	bSites.push_back(sd);
1767	+	}
1768	+
1769	+	ConvexHull* surfaceMeshB = new ConvexHull();
1770	+
1771	+	surfaceMeshB->computeHull(bSites);
1772	+	areaB = surfaceMeshB->getArea();
1773	+	} else {
1774	+	if (usePeriodicBoundaryConditions_) {
1775	+	// in periodic boundaries, the surface area is twice the x-y
1776	+	// area of the current box:
1777	+	areaB = 2.0 * snap->getXYarea();
1778	+	} else {
1779	+	// in non-periodic simulations, without explicitly setting
1780	+	// selections, but if a sphereBradius has been set, just use that:
1781	+	areaB = 4.0 * M_PI * pow(sphereBRadius_, 2);
1782	+	}
1783	+	}
1784	+
1785	+	dividingArea_ = min(areaA, areaB);
1786	+	hasDividingArea_ = true;
1787	+	return dividingArea_;
1788	+	}
1789	+
1790		void RNEMD::doRNEMD() {
1791	+	if (!doRNEMD_) return;
1792	+	trialCount_++;
1793
1794	<	switch(rnemdType_) {
1795	<	case rnemdKineticScale :
1796	<	case rnemdPxScale :
1797	<	case rnemdPyScale :
1798	<	case rnemdPzScale :
1799	<	doScale();
1794	>	// object evaluator:
1795	>	evaluator_.loadScriptString(rnemdObjectSelection_);
1796	>	seleMan_.setSelectionSet(evaluator_.evaluate());
1797	>
1798	>	evaluatorA_.loadScriptString(selectionA_);
1799	>	evaluatorB_.loadScriptString(selectionB_);
1800	>
1801	>	seleManA_.setSelectionSet(evaluatorA_.evaluate());
1802	>	seleManB_.setSelectionSet(evaluatorB_.evaluate());
1803	>
1804	>	commonA_ = seleManA_ & seleMan_;
1805	>	commonB_ = seleManB_ & seleMan_;
1806	>
1807	>	// Target exchange quantities (in each exchange) = dividingArea * dt * flux
1808	>	// dt = exchange time interval
1809	>	// flux = target flux
1810	>	// dividingArea = smallest dividing surface between the two regions
1811	>
1812	>	hasDividingArea_ = false;
1813	>	RealType area = getDividingArea();
1814	>
1815	>	kineticTarget_ = kineticFlux_ * exchangeTime_ * area;
1816	>	momentumTarget_ = momentumFluxVector_ * exchangeTime_ * area;
1817	>	angularMomentumTarget_ = angularMomentumFluxVector_ * exchangeTime_ * area;
1818	>
1819	>	switch(rnemdMethod_) {
1820	>	case rnemdSwap:
1821	>	doSwap(commonA_, commonB_);
1822		break;
1823	<	case rnemdKineticSwap :
1824	<	case rnemdPx :
861	<	case rnemdPy :
862	<	case rnemdPz :
863	<	doSwap();
1823	>	case rnemdNIVS:
1824	>	doNIVS(commonA_, commonB_);
1825		break;
1826	<	case rnemdUnknown :
1826	>	case rnemdVSS:
1827	>	doVSS(commonA_, commonB_);
1828	>	break;
1829	>	case rnemdUnkownMethod:
1830		default :
1831		break;
1832		}
1833		}
1834
1835		void RNEMD::collectData() {
1836	<
1836	>	if (!doRNEMD_) return;
1837		Snapshot* currentSnap_ = info_->getSnapshotManager()->getCurrentSnapshot();
1838	+
1839	+	// collectData can be called more frequently than the doRNEMD, so use the
1840	+	// computed area from the last exchange time:
1841	+	RealType area = getDividingArea();
1842	+	areaAccumulator_->add(area);
1843		Mat3x3d hmat = currentSnap_->getHmat();
875	–
1844		seleMan_.setSelectionSet(evaluator_.evaluate());
1845
1846	<	int selei;
1846	>	int selei(0);
1847		StuntDouble* sd;
1848	<	int idx;
1848	>	int binNo;
1849
1850	+	vector<RealType> binMass(nBins_, 0.0);
1851	+	vector<RealType> binPx(nBins_, 0.0);
1852	+	vector<RealType> binPy(nBins_, 0.0);
1853	+	vector<RealType> binPz(nBins_, 0.0);
1854	+	vector<RealType> binOmegax(nBins_, 0.0);
1855	+	vector<RealType> binOmegay(nBins_, 0.0);
1856	+	vector<RealType> binOmegaz(nBins_, 0.0);
1857	+	vector<RealType> binKE(nBins_, 0.0);
1858	+	vector<int> binDOF(nBins_, 0);
1859	+	vector<int> binCount(nBins_, 0);
1860	+
1861	+	// alternative approach, track all molecules instead of only those
1862	+	// selected for scaling/swapping:
1863	+	/*
1864	+	SimInfo::MoleculeIterator miter;
1865	+	vector<StuntDouble*>::iterator iiter;
1866	+	Molecule* mol;
1867	+	StuntDouble* sd;
1868	+	for (mol = info_->beginMolecule(miter); mol != NULL;
1869	+	mol = info_->nextMolecule(miter))
1870	+	sd is essentially sd
1871	+	for (sd = mol->beginIntegrableObject(iiter);
1872	+	sd != NULL;
1873	+	sd = mol->nextIntegrableObject(iiter))
1874	+	*/
1875	+
1876		for (sd = seleMan_.beginSelected(selei); sd != NULL;
1877	<	sd = seleMan_.nextSelected(selei)) {
1878	<
885	<	idx = sd->getLocalIndex();
886	<
1877	>	sd = seleMan_.nextSelected(selei)) {
1878	>
1879		Vector3d pos = sd->getPos();
1880
1881		// wrap the stuntdouble's position back into the box:
1882
1883	<	if (usePeriodicBoundaryConditions_)
1883	>	if (usePeriodicBoundaryConditions_) {
1884		currentSnap_->wrapVector(pos);
1885	<
1886	<	// which bin is this stuntdouble in?
1887	<	// wrapped positions are in the range [-0.5*hmat(2,2), +0.5*hmat(2,2)]
1888	<
1889	<	int binNo = int(nBins_ * (pos.z() / hmat(2,2) + 0.5)) % nBins_;
1885	>	// which bin is this stuntdouble in?
1886	>	// wrapped positions are in the range [-0.5*hmat(2,2), +0.5*hmat(2,2)]
1887	>	// Shift molecules by half a box to have bins start at 0
1888	>	// The modulo operator is used to wrap the case when we are
1889	>	// beyond the end of the bins back to the beginning.
1890	>	binNo = int(nBins_ * (pos.z() / hmat(2,2) + 0.5)) % nBins_;
1891	>	} else {
1892	>	Vector3d rPos = pos - coordinateOrigin_;
1893	>	binNo = int(rPos.length() / binWidth_);
1894	>	}
1895
899	–	if (rnemdLogWidth_ == midBin_ + 1)
900	–	if (binNo > midBin_)
901	–	binNo = nBins_ - binNo;
902	–
1896		RealType mass = sd->getMass();
1897		Vector3d vel = sd->getVel();
1898	<	RealType value;
1899	<	RealType xVal, yVal, zVal;
1898	>	Vector3d rPos = sd->getPos() - coordinateOrigin_;
1899	>	Vector3d aVel = cross(rPos, vel);
1900	>
1901	>	if (binNo >= 0 && binNo < nBins_)  {
1902	>	binCount[binNo]++;
1903	>	binMass[binNo] += mass;
1904	>	binPx[binNo] += mass*vel.x();
1905	>	binPy[binNo] += mass*vel.y();
1906	>	binPz[binNo] += mass*vel.z();
1907	>	binOmegax[binNo] += aVel.x();
1908	>	binOmegay[binNo] += aVel.y();
1909	>	binOmegaz[binNo] += aVel.z();
1910	>	binKE[binNo] += 0.5 * (mass * vel.lengthSquare());
1911	>	binDOF[binNo] += 3;
1912	>
1913	>	if (sd->isDirectional()) {
1914	>	Vector3d angMom = sd->getJ();
1915	>	Mat3x3d I = sd->getI();
1916	>	if (sd->isLinear()) {
1917	>	int i = sd->linearAxis();
1918	>	int j = (i + 1) % 3;
1919	>	int k = (i + 2) % 3;
1920	>	binKE[binNo] += 0.5 * (angMom[j] * angMom[j] / I(j, j) +
1921	>	angMom[k] * angMom[k] / I(k, k));
1922	>	binDOF[binNo] += 2;
1923	>	} else {
1924	>	binKE[binNo] += 0.5 * (angMom[0] * angMom[0] / I(0, 0) +
1925	>	angMom[1] * angMom[1] / I(1, 1) +
1926	>	angMom[2] * angMom[2] / I(2, 2));
1927	>	binDOF[binNo] += 3;
1928	>	}
1929	>	}
1930	>	}
1931	>	}
1932	>
1933	>	#ifdef IS_MPI
1934	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &binCount[0],
1935	>	nBins_, MPI::INT, MPI::SUM);
1936	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &binMass[0],
1937	>	nBins_, MPI::REALTYPE, MPI::SUM);
1938	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &binPx[0],
1939	>	nBins_, MPI::REALTYPE, MPI::SUM);
1940	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &binPy[0],
1941	>	nBins_, MPI::REALTYPE, MPI::SUM);
1942	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &binPz[0],
1943	>	nBins_, MPI::REALTYPE, MPI::SUM);
1944	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &binOmegax[0],
1945	>	nBins_, MPI::REALTYPE, MPI::SUM);
1946	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &binOmegay[0],
1947	>	nBins_, MPI::REALTYPE, MPI::SUM);
1948	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &binOmegaz[0],
1949	>	nBins_, MPI::REALTYPE, MPI::SUM);
1950	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &binKE[0],
1951	>	nBins_, MPI::REALTYPE, MPI::SUM);
1952	>	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &binDOF[0],
1953	>	nBins_, MPI::INT, MPI::SUM);
1954	>	#endif
1955
1956	<	switch(rnemdType_) {
1957	<	case rnemdKineticSwap :
1958	<	case rnemdKineticScale :
1959	<
1960	<	value = mass * (vel[0]*vel[0] + vel[1]*vel[1] +
1961	<	vel[2]*vel[2]);
1962	<
1963	<	valueCount_[binNo] += 3;
1964	<	if (sd->isDirectional()) {
1965	<	Vector3d angMom = sd->getJ();
1966	<	Mat3x3d I = sd->getI();
1967	<
1968	<	if (sd->isLinear()) {
1969	<	int i = sd->linearAxis();
1970	<	int j = (i + 1) % 3;
1971	<	int k = (i + 2) % 3;
1972	<	value += angMom[j] * angMom[j] / I(j, j) +
1973	<	angMom[k] * angMom[k] / I(k, k);
1956	>	Vector3d vel;
1957	>	Vector3d aVel;
1958	>	RealType den;
1959	>	RealType temp;
1960	>	RealType z;
1961	>	RealType r;
1962	>	for (int i = 0; i < nBins_; i++) {
1963	>	if (usePeriodicBoundaryConditions_) {
1964	>	z = (((RealType)i + 0.5) / (RealType)nBins_) * hmat(2,2);
1965	>	den = binMass[i] * nBins_ * PhysicalConstants::densityConvert
1966	>	/ currentSnap_->getVolume() ;
1967	>	} else {
1968	>	r = (((RealType)i + 0.5) * binWidth_);
1969	>	RealType rinner = (RealType)i * binWidth_;
1970	>	RealType router = (RealType)(i+1) * binWidth_;
1971	>	den = binMass[i] * 3.0 * PhysicalConstants::densityConvert
1972	>	/ (4.0 * M_PI * (pow(router,3) - pow(rinner,3)));
1973	>	}
1974	>	vel.x() = binPx[i] / binMass[i];
1975	>	vel.y() = binPy[i] / binMass[i];
1976	>	vel.z() = binPz[i] / binMass[i];
1977	>	aVel.x() = binOmegax[i] / binCount[i];
1978	>	aVel.y() = binOmegay[i] / binCount[i];
1979	>	aVel.z() = binOmegaz[i] / binCount[i];
1980
1981	<	valueCount_[binNo] +=2;
1982	<
1983	<	} else {
1984	<	value += angMom[0]*angMom[0]/I(0, 0)
1985	<	+ angMom[1]*angMom[1]/I(1, 1)
1986	<	+ angMom[2]*angMom[2]/I(2, 2);
1987	<	valueCount_[binNo] +=3;
1988	<	}
1989	<	}
1990	<	value = value / PhysicalConstants::energyConvert / PhysicalConstants::kb;
1991	<
1992	<	break;
1993	<	case rnemdPx :
1994	<	case rnemdPxScale :
1995	<	value = mass * vel[0];
1996	<	valueCount_[binNo]++;
1997	<	xVal = mass * vel.x() * vel.x() / PhysicalConstants::energyConvert
1998	<	/ PhysicalConstants::kb;
1999	<	yVal = mass * vel.y() * vel.y() / PhysicalConstants::energyConvert
2000	<	/ PhysicalConstants::kb;
2001	<	zVal = mass * vel.z() * vel.z() / PhysicalConstants::energyConvert
2002	<	/ PhysicalConstants::kb;
2003	<	xTempHist_[binNo] += xVal;
2004	<	yTempHist_[binNo] += yVal;
2005	<	zTempHist_[binNo] += zVal;
2006	<	break;
2007	<	case rnemdPy :
2008	<	case rnemdPyScale :
2009	<	value = mass * vel[1];
956	<	valueCount_[binNo]++;
957	<	break;
958	<	case rnemdPz :
959	<	case rnemdPzScale :
960	<	value = mass * vel[2];
961	<	valueCount_[binNo]++;
962	<	break;
963	<	case rnemdUnknown :
964	<	default :
965	<	break;
1981	>	if (binCount[i] > 0) {
1982	>	// only add values if there are things to add
1983	>	temp = 2.0 * binKE[i] / (binDOF[i] * PhysicalConstants::kb *
1984	>	PhysicalConstants::energyConvert);
1985	>
1986	>	for (unsigned int j = 0; j < outputMask_.size(); ++j) {
1987	>	if(outputMask_[j]) {
1988	>	switch(j) {
1989	>	case Z:
1990	>	dynamic_cast<Accumulator *>(data_[j].accumulator[i])->add(z);
1991	>	break;
1992	>	case R:
1993	>	dynamic_cast<Accumulator *>(data_[j].accumulator[i])->add(r);
1994	>	break;
1995	>	case TEMPERATURE:
1996	>	dynamic_cast<Accumulator *>(data_[j].accumulator[i])->add(temp);
1997	>	break;
1998	>	case VELOCITY:
1999	>	dynamic_cast<VectorAccumulator *>(data_[j].accumulator[i])->add(vel);
2000	>	break;
2001	>	case ANGULARVELOCITY:
2002	>	dynamic_cast<VectorAccumulator *>(data_[j].accumulator[i])->add(aVel);
2003	>	break;
2004	>	case DENSITY:
2005	>	dynamic_cast<Accumulator *>(data_[j].accumulator[i])->add(den);
2006	>	break;
2007	>	}
2008	>	}
2009	>	}
2010		}
967	–	valueHist_[binNo] += value;
2011		}
969	–
2012		}
2013
2014		void RNEMD::getStarted() {
2015	<	Snapshot* currentSnap_ = info_->getSnapshotManager()->getCurrentSnapshot();
2016	<	Stats& stat = currentSnap_->statData;
2017	<	stat[Stats::RNEMD_EXCHANGE_TOTAL] = exchangeSum_;
2015	>	if (!doRNEMD_) return;
2016	>	hasDividingArea_ = false;
2017	>	collectData();
2018	>	writeOutputFile();
2019		}
2020
2021	<	void RNEMD::getStatus() {
2022	<
2023	<	Snapshot* currentSnap_ = info_->getSnapshotManager()->getCurrentSnapshot();
2024	<	Stats& stat = currentSnap_->statData;
2025	<	RealType time = currentSnap_->getTime();
2026	<
2027	<	stat[Stats::RNEMD_EXCHANGE_TOTAL] = exchangeSum_;
2028	<	//or to be more meaningful, define another item as exchangeSum_ / time
2029	<	int j;
2030	<
2021	>	void RNEMD::parseOutputFileFormat(const std::string& format) {
2022	>	if (!doRNEMD_) return;
2023	>	StringTokenizer tokenizer(format, " ,;|\t\n\r");
2024	>
2025	>	while(tokenizer.hasMoreTokens()) {
2026	>	std::string token(tokenizer.nextToken());
2027	>	toUpper(token);
2028	>	OutputMapType::iterator i = outputMap_.find(token);
2029	>	if (i != outputMap_.end()) {
2030	>	outputMask_.set(i->second);
2031	>	} else {
2032	>	sprintf( painCave.errMsg,
2033	>	"RNEMD::parseOutputFileFormat: %s is not a recognized\n"
2034	>	"\toutputFileFormat keyword.\n", token.c_str() );
2035	>	painCave.isFatal = 0;
2036	>	painCave.severity = OPENMD_ERROR;
2037	>	simError();
2038	>	}
2039	>	}
2040	>	}
2041	>
2042	>	void RNEMD::writeOutputFile() {
2043	>	if (!doRNEMD_) return;
2044	>
2045		#ifdef IS_MPI
989	–
990	–	// all processors have the same number of bins, and STL vectors pack their
991	–	// arrays, so in theory, this should be safe:
992	–
993	–	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &valueHist_[0],
994	–	rnemdLogWidth_, MPI::REALTYPE, MPI::SUM);
995	–	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &valueCount_[0],
996	–	rnemdLogWidth_, MPI::INT, MPI::SUM);
997	–	if (rnemdType_ == rnemdPx || rnemdType_ == rnemdPxScale) {
998	–	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &xTempHist_[0],
999	–	rnemdLogWidth_, MPI::REALTYPE, MPI::SUM);
1000	–	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &yTempHist_[0],
1001	–	rnemdLogWidth_, MPI::REALTYPE, MPI::SUM);
1002	–	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &zTempHist_[0],
1003	–	rnemdLogWidth_, MPI::REALTYPE, MPI::SUM);
1004	–	}
2046		// If we're the root node, should we print out the results
2047		int worldRank = MPI::COMM_WORLD.Get_rank();
2048		if (worldRank == 0) {
2049		#endif
2050	<	rnemdLog_ << time;
2051	<	for (j = 0; j < rnemdLogWidth_; j++) {
2052	<	rnemdLog_ << "\t" << valueHist_[j] / (RealType)valueCount_[j];
2050	>	rnemdFile_.open(rnemdFileName_.c_str(), std::ios::out | std::ios::trunc );
2051	>
2052	>	if( !rnemdFile_ ){
2053	>	sprintf( painCave.errMsg,
2054	>	"Could not open \"%s\" for RNEMD output.\n",
2055	>	rnemdFileName_.c_str());
2056	>	painCave.isFatal = 1;
2057	>	simError();
2058		}
2059	<	rnemdLog_ << "\n";
2060	<	if (rnemdType_ == rnemdPx || rnemdType_ == rnemdPxScale ) {
2061	<	xTempLog_ << time;
2062	<	for (j = 0; j < rnemdLogWidth_; j++) {
2063	<	xTempLog_ << "\t" << xTempHist_[j] / (RealType)valueCount_[j];
2059	>
2060	>	Snapshot* currentSnap_ = info_->getSnapshotManager()->getCurrentSnapshot();
2061	>
2062	>	RealType time = currentSnap_->getTime();
2063	>	RealType avgArea;
2064	>	areaAccumulator_->getAverage(avgArea);
2065	>	RealType Jz = kineticExchange_ / (time * avgArea)
2066	>	/ PhysicalConstants::energyConvert;
2067	>	Vector3d JzP = momentumExchange_ / (time * avgArea);
2068	>	Vector3d JzL = angularMomentumExchange_ / (time * avgArea);
2069	>
2070	>	rnemdFile_ << "#######################################################\n";
2071	>	rnemdFile_ << "# RNEMD {\n";
2072	>
2073	>	map<string, RNEMDMethod>::iterator mi;
2074	>	for(mi = stringToMethod_.begin(); mi != stringToMethod_.end(); ++mi) {
2075	>	if ( (*mi).second == rnemdMethod_)
2076	>	rnemdFile_ << "#    exchangeMethod  = \"" << (*mi).first << "\";\n";
2077	>	}
2078	>	map<string, RNEMDFluxType>::iterator fi;
2079	>	for(fi = stringToFluxType_.begin(); fi != stringToFluxType_.end(); ++fi) {
2080	>	if ( (*fi).second == rnemdFluxType_)
2081	>	rnemdFile_ << "#    fluxType  = \"" << (*fi).first << "\";\n";
2082	>	}
2083	>
2084	>	rnemdFile_ << "#    exchangeTime = " << exchangeTime_ << ";\n";
2085	>
2086	>	rnemdFile_ << "#    objectSelection = \""
2087	>	<< rnemdObjectSelection_ << "\";\n";
2088	>	rnemdFile_ << "#    selectionA = \"" << selectionA_ << "\";\n";
2089	>	rnemdFile_ << "#    selectionB = \"" << selectionB_ << "\";\n";
2090	>	rnemdFile_ << "# }\n";
2091	>	rnemdFile_ << "#######################################################\n";
2092	>	rnemdFile_ << "# RNEMD report:\n";
2093	>	rnemdFile_ << "#      running time = " << time << " fs\n";
2094	>	rnemdFile_ << "# Target flux:\n";
2095	>	rnemdFile_ << "#           kinetic = "
2096	>	<< kineticFlux_ / PhysicalConstants::energyConvert
2097	>	<< " (kcal/mol/A^2/fs)\n";
2098	>	rnemdFile_ << "#          momentum = " << momentumFluxVector_
2099	>	<< " (amu/A/fs^2)\n";
2100	>	rnemdFile_ << "#  angular momentum = " << angularMomentumFluxVector_
2101	>	<< " (amu/A^2/fs^2)\n";
2102	>	rnemdFile_ << "# Target one-time exchanges:\n";
2103	>	rnemdFile_ << "#          kinetic = "
2104	>	<< kineticTarget_ / PhysicalConstants::energyConvert
2105	>	<< " (kcal/mol)\n";
2106	>	rnemdFile_ << "#          momentum = " << momentumTarget_
2107	>	<< " (amu*A/fs)\n";
2108	>	rnemdFile_ << "#  angular momentum = " << angularMomentumTarget_
2109	>	<< " (amu*A^2/fs)\n";
2110	>	rnemdFile_ << "# Actual exchange totals:\n";
2111	>	rnemdFile_ << "#          kinetic = "
2112	>	<< kineticExchange_ / PhysicalConstants::energyConvert
2113	>	<< " (kcal/mol)\n";
2114	>	rnemdFile_ << "#          momentum = " << momentumExchange_
2115	>	<< " (amu*A/fs)\n";
2116	>	rnemdFile_ << "#  angular momentum = " << angularMomentumExchange_
2117	>	<< " (amu*A^2/fs)\n";
2118	>	rnemdFile_ << "# Actual flux:\n";
2119	>	rnemdFile_ << "#          kinetic = " << Jz
2120	>	<< " (kcal/mol/A^2/fs)\n";
2121	>	rnemdFile_ << "#          momentum = " << JzP
2122	>	<< " (amu/A/fs^2)\n";
2123	>	rnemdFile_ << "#  angular momentum = " << JzL
2124	>	<< " (amu/A^2/fs^2)\n";
2125	>	rnemdFile_ << "# Exchange statistics:\n";
2126	>	rnemdFile_ << "#               attempted = " << trialCount_ << "\n";
2127	>	rnemdFile_ << "#                  failed = " << failTrialCount_ << "\n";
2128	>	if (rnemdMethod_ == rnemdNIVS) {
2129	>	rnemdFile_ << "#  NIVS root-check errors = "
2130	>	<< failRootCount_ << "\n";
2131	>	}
2132	>	rnemdFile_ << "#######################################################\n";
2133	>
2134	>
2135	>
2136	>	//write title
2137	>	rnemdFile_ << "#";
2138	>	for (unsigned int i = 0; i < outputMask_.size(); ++i) {
2139	>	if (outputMask_[i]) {
2140	>	rnemdFile_ << "\t" << data_[i].title <<
2141	>	"(" << data_[i].units << ")";
2142	>	// add some extra tabs for column alignment
2143	>	if (data_[i].dataType == "Vector3d") rnemdFile_ << "\t\t";
2144		}
2145	<	xTempLog_ << "\n";
2146	<	yTempLog_ << time;
2147	<	for (j = 0; j < rnemdLogWidth_; j++) {
2148	<	yTempLog_ << "\t" << yTempHist_[j] / (RealType)valueCount_[j];
2145	>	}
2146	>	rnemdFile_ << std::endl;
2147	>
2148	>	rnemdFile_.precision(8);
2149	>
2150	>	for (int j = 0; j < nBins_; j++) {
2151	>
2152	>	for (unsigned int i = 0; i < outputMask_.size(); ++i) {
2153	>	if (outputMask_[i]) {
2154	>	if (data_[i].dataType == "RealType")
2155	>	writeReal(i,j);
2156	>	else if (data_[i].dataType == "Vector3d")
2157	>	writeVector(i,j);
2158	>	else {
2159	>	sprintf( painCave.errMsg,
2160	>	"RNEMD found an unknown data type for: %s ",
2161	>	data_[i].title.c_str());
2162	>	painCave.isFatal = 1;
2163	>	simError();
2164	>	}
2165	>	}
2166		}
2167	<	yTempLog_ << "\n";
2168	<	zTempLog_ << time;
2169	<	for (j = 0; j < rnemdLogWidth_; j++) {
2170	<	zTempLog_ << "\t" << zTempHist_[j] / (RealType)valueCount_[j];
2167	>	rnemdFile_ << std::endl;
2168	>
2169	>	}
2170	>
2171	>	rnemdFile_ << "#######################################################\n";
2172	>	rnemdFile_ << "# Standard Deviations in those quantities follow:\n";
2173	>	rnemdFile_ << "#######################################################\n";
2174	>
2175	>
2176	>	for (int j = 0; j < nBins_; j++) {
2177	>	rnemdFile_ << "#";
2178	>	for (unsigned int i = 0; i < outputMask_.size(); ++i) {
2179	>	if (outputMask_[i]) {
2180	>	if (data_[i].dataType == "RealType")
2181	>	writeRealStdDev(i,j);
2182	>	else if (data_[i].dataType == "Vector3d")
2183	>	writeVectorStdDev(i,j);
2184	>	else {
2185	>	sprintf( painCave.errMsg,
2186	>	"RNEMD found an unknown data type for: %s ",
2187	>	data_[i].title.c_str());
2188	>	painCave.isFatal = 1;
2189	>	simError();
2190	>	}
2191	>	}
2192		}
2193	<	zTempLog_ << "\n";
2194	<	}
2193	>	rnemdFile_ << std::endl;
2194	>
2195	>	}
2196	>
2197	>	rnemdFile_.flush();
2198	>	rnemdFile_.close();
2199	>
2200		#ifdef IS_MPI
2201		}
2202		#endif
2203	<	for (j = 0; j < rnemdLogWidth_; j++) {
2204	<	valueCount_[j] = 0;
2205	<	valueHist_[j] = 0.0;
2203	>
2204	>	}
2205	>
2206	>	void RNEMD::writeReal(int index, unsigned int bin) {
2207	>	if (!doRNEMD_) return;
2208	>	assert(index >=0 && index < ENDINDEX);
2209	>	assert(int(bin) < nBins_);
2210	>	RealType s;
2211	>	int count;
2212	>
2213	>	count = data_[index].accumulator[bin]->count();
2214	>	if (count == 0) return;
2215	>
2216	>	dynamic_cast<Accumulator *>(data_[index].accumulator[bin])->getAverage(s);
2217	>
2218	>	if (! isinf(s) && ! isnan(s)) {
2219	>	rnemdFile_ << "\t" << s;
2220	>	} else{
2221	>	sprintf( painCave.errMsg,
2222	>	"RNEMD detected a numerical error writing: %s for bin %d",
2223	>	data_[index].title.c_str(), bin);
2224	>	painCave.isFatal = 1;
2225	>	simError();
2226	>	}
2227	>	}
2228	>
2229	>	void RNEMD::writeVector(int index, unsigned int bin) {
2230	>	if (!doRNEMD_) return;
2231	>	assert(index >=0 && index < ENDINDEX);
2232	>	assert(int(bin) < nBins_);
2233	>	Vector3d s;
2234	>	int count;
2235	>
2236	>	count = data_[index].accumulator[bin]->count();
2237	>
2238	>	if (count == 0) return;
2239	>
2240	>	dynamic_cast<VectorAccumulator*>(data_[index].accumulator[bin])->getAverage(s);
2241	>	if (isinf(s[0]) || isnan(s[0]) ||
2242	>	isinf(s[1]) || isnan(s[1]) ||
2243	>	isinf(s[2]) || isnan(s[2]) ) {
2244	>	sprintf( painCave.errMsg,
2245	>	"RNEMD detected a numerical error writing: %s for bin %d",
2246	>	data_[index].title.c_str(), bin);
2247	>	painCave.isFatal = 1;
2248	>	simError();
2249	>	} else {
2250	>	rnemdFile_ << "\t" << s[0] << "\t" << s[1] << "\t" << s[2];
2251		}
2252	<	if (rnemdType_ == rnemdPx || rnemdType_ == rnemdPxScale)
2253	<	for (j = 0; j < rnemdLogWidth_; j++) {
2254	<	xTempHist_[j] = 0.0;
2255	<	yTempHist_[j] = 0.0;
2256	<	zTempHist_[j] = 0.0;
2257	<	}
2252	>	}
2253	>
2254	>	void RNEMD::writeRealStdDev(int index, unsigned int bin) {
2255	>	if (!doRNEMD_) return;
2256	>	assert(index >=0 && index < ENDINDEX);
2257	>	assert(int(bin) < nBins_);
2258	>	RealType s;
2259	>	int count;
2260	>
2261	>	count = data_[index].accumulator[bin]->count();
2262	>	if (count == 0) return;
2263	>
2264	>	dynamic_cast<Accumulator *>(data_[index].accumulator[bin])->getStdDev(s);
2265	>
2266	>	if (! isinf(s) && ! isnan(s)) {
2267	>	rnemdFile_ << "\t" << s;
2268	>	} else{
2269	>	sprintf( painCave.errMsg,
2270	>	"RNEMD detected a numerical error writing: %s std. dev. for bin %d",
2271	>	data_[index].title.c_str(), bin);
2272	>	painCave.isFatal = 1;
2273	>	simError();
2274	>	}
2275		}
2276	+
2277	+	void RNEMD::writeVectorStdDev(int index, unsigned int bin) {
2278	+	if (!doRNEMD_) return;
2279	+	assert(index >=0 && index < ENDINDEX);
2280	+	assert(int(bin) < nBins_);
2281	+	Vector3d s;
2282	+	int count;
2283	+
2284	+	count = data_[index].accumulator[bin]->count();
2285	+	if (count == 0) return;
2286	+
2287	+	dynamic_cast<VectorAccumulator*>(data_[index].accumulator[bin])->getStdDev(s);
2288	+	if (isinf(s[0]) || isnan(s[0]) ||
2289	+	isinf(s[1]) || isnan(s[1]) ||
2290	+	isinf(s[2]) || isnan(s[2]) ) {
2291	+	sprintf( painCave.errMsg,
2292	+	"RNEMD detected a numerical error writing: %s std. dev. for bin %d",
2293	+	data_[index].title.c_str(), bin);
2294	+	painCave.isFatal = 1;
2295	+	simError();
2296	+	} else {
2297	+	rnemdFile_ << "\t" << s[0] << "\t" << s[1] << "\t" << s[2];
2298	+	}
2299	+	}
2300		}
2301	+

Diff Legend

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