ViewVC Help

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

root/OpenMD/trunk/src/rnemd/RNEMD.cpp

Comparing:
branches/development/src/integrators/RNEMD.cpp (file contents), Revision 1627 by gezelter, Tue Sep 13 22:05:04 2011 UTC vs.
trunk/src/rnemd/RNEMD.cpp (file contents), Revision 2026 by gezelter, Wed Oct 22 12:23:59 2014 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	+	#ifdef IS_MPI
42	+	#include <mpi.h>
43	+	#endif
44
45		#include <cmath>
46	<	#include "integrators/RNEMD.hpp"
46	>	#include <sstream>
47	>	#include <string>
48	>
49	>	#include "rnemd/RNEMD.hpp"
50		#include "math/Vector3.hpp"
51	+	#include "math/Vector.hpp"
52		#include "math/SquareMatrix3.hpp"
53		#include "math/Polynomial.hpp"
54		#include "primitives/Molecule.hpp"
55		#include "primitives/StuntDouble.hpp"
56		#include "utils/PhysicalConstants.hpp"
57		#include "utils/Tuple.hpp"
58	+	#include "brains/Thermo.hpp"
59	+	#include "math/ConvexHull.hpp"
60
61	<	#ifndef IS_MPI
62	<	#include "math/SeqRandNumGen.hpp"
63	<	#else
55	<	#include <mpi.h>
56	<	#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	>	hasData_(false), hasDividingArea_(false),
76	>	usePeriodicBoundaryConditions_(info->getSimParams()->getUsePeriodicBoundaryConditions()) {
77
78	+	trialCount_ = 0;
79		failTrialCount_ = 0;
80		failRootCount_ = 0;
81
82	<	int seedValue;
83	<	Globals * simParams = info->getSimParams();
82	>	Globals* simParams = info->getSimParams();
83	>	RNEMDParameters* rnemdParams = simParams->getRNEMDParameters();
84
85	<	stringToEnumMap_["KineticSwap"] = rnemdKineticSwap;
86	<	stringToEnumMap_["KineticScale"] = rnemdKineticScale;
73	<	stringToEnumMap_["PxScale"] = rnemdPxScale;
74	<	stringToEnumMap_["PyScale"] = rnemdPyScale;
75	<	stringToEnumMap_["PzScale"] = rnemdPzScale;
76	<	stringToEnumMap_["Px"] = rnemdPx;
77	<	stringToEnumMap_["Py"] = rnemdPy;
78	<	stringToEnumMap_["Pz"] = rnemdPz;
79	<	stringToEnumMap_["Unknown"] = rnemdUnknown;
85	>	doRNEMD_ = rnemdParams->getUseRNEMD();
86	>	if (!doRNEMD_) return;
87
88	<	rnemdObjectSelection_ = simParams->getRNEMD_objectSelection();
89	<	evaluator_.loadScriptString(rnemdObjectSelection_);
90	<	seleMan_.setSelectionSet(evaluator_.evaluate());
88	>	stringToMethod_["Swap"]  = rnemdSwap;
89	>	stringToMethod_["NIVS"]  = rnemdNIVS;
90	>	stringToMethod_["VSS"]   = rnemdVSS;
91
92	<	// do some sanity checking
92	>	stringToFluxType_["KE"]  = rnemdKE;
93	>	stringToFluxType_["Px"]  = rnemdPx;
94	>	stringToFluxType_["Py"]  = rnemdPy;
95	>	stringToFluxType_["Pz"]  = rnemdPz;
96	>	stringToFluxType_["Pvector"]  = rnemdPvector;
97	>	stringToFluxType_["Lx"]  = rnemdLx;
98	>	stringToFluxType_["Ly"]  = rnemdLy;
99	>	stringToFluxType_["Lz"]  = rnemdLz;
100	>	stringToFluxType_["Lvector"]  = rnemdLvector;
101	>	stringToFluxType_["KE+Px"]  = rnemdKePx;
102	>	stringToFluxType_["KE+Py"]  = rnemdKePy;
103	>	stringToFluxType_["KE+Pvector"]  = rnemdKePvector;
104	>	stringToFluxType_["KE+Lx"]  = rnemdKeLx;
105	>	stringToFluxType_["KE+Ly"]  = rnemdKeLy;
106	>	stringToFluxType_["KE+Lz"]  = rnemdKeLz;
107	>	stringToFluxType_["KE+Lvector"]  = rnemdKeLvector;
108
109	<	int selectionCount = seleMan_.getSelectionCount();
110	<	int nIntegrable = info->getNGlobalIntegrableObjects();
109	>	runTime_ = simParams->getRunTime();
110	>	statusTime_ = simParams->getStatusTime();
111
112	<	if (selectionCount > nIntegrable) {
112	>	const string methStr = rnemdParams->getMethod();
113	>	bool hasFluxType = rnemdParams->haveFluxType();
114	>
115	>	rnemdObjectSelection_ = rnemdParams->getObjectSelection();
116	>
117	>	string fluxStr;
118	>	if (hasFluxType) {
119	>	fluxStr = rnemdParams->getFluxType();
120	>	} else {
121		sprintf(painCave.errMsg,
122	<	"RNEMD warning: The current RNEMD_objectSelection,\n"
123	<	"\t\t%s\n"
124	<	"\thas resulted in %d selected objects.  However,\n"
125	<	"\tthe total number of integrable objects in the system\n"
126	<	"\tis only %d.  This is almost certainly not what you want\n"
127	<	"\tto do.  A likely cause of this is forgetting the _RB_0\n"
128	<	"\tselector in the selection script!\n",
99	<	rnemdObjectSelection_.c_str(),
100	<	selectionCount, nIntegrable);
101	<	painCave.isFatal = 0;
122	>	"RNEMD: No fluxType was set in the md file.  This parameter,\n"
123	>	"\twhich must be one of the following values:\n"
124	>	"\tKE, Px, Py, Pz, Pvector, Lx, Ly, Lz, Lvector,\n"
125	>	"\tKE+Px, KE+Py, KE+Pvector, KE+Lx, KE+Ly, KE+Lz, KE+Lvector\n"
126	>	"\tmust be set to use RNEMD\n");
127	>	painCave.isFatal = 1;
128	>	painCave.severity = OPENMD_ERROR;
129		simError();
103	–
130		}
131	+
132	+	bool hasKineticFlux = rnemdParams->haveKineticFlux();
133	+	bool hasMomentumFlux = rnemdParams->haveMomentumFlux();
134	+	bool hasMomentumFluxVector = rnemdParams->haveMomentumFluxVector();
135	+	bool hasAngularMomentumFlux = rnemdParams->haveAngularMomentumFlux();
136	+	bool hasAngularMomentumFluxVector = rnemdParams->haveAngularMomentumFluxVector();
137	+	hasSelectionA_ = rnemdParams->haveSelectionA();
138	+	hasSelectionB_ = rnemdParams->haveSelectionB();
139	+	bool hasSlabWidth = rnemdParams->haveSlabWidth();
140	+	bool hasSlabACenter = rnemdParams->haveSlabACenter();
141	+	bool hasSlabBCenter = rnemdParams->haveSlabBCenter();
142	+	bool hasSphereARadius = rnemdParams->haveSphereARadius();
143	+	hasSphereBRadius_ = rnemdParams->haveSphereBRadius();
144	+	bool hasCoordinateOrigin = rnemdParams->haveCoordinateOrigin();
145	+	bool hasOutputFileName = rnemdParams->haveOutputFileName();
146	+	bool hasOutputFields = rnemdParams->haveOutputFields();
147
148	<	const std::string st = simParams->getRNEMD_exchangeType();
148	>	map<string, RNEMDMethod>::iterator i;
149	>	i = stringToMethod_.find(methStr);
150	>	if (i != stringToMethod_.end())
151	>	rnemdMethod_ = i->second;
152	>	else {
153	>	sprintf(painCave.errMsg,
154	>	"RNEMD: The current method,\n"
155	>	"\t\t%s is not one of the recognized\n"
156	>	"\texchange methods: Swap, NIVS, or VSS\n",
157	>	methStr.c_str());
158	>	painCave.isFatal = 1;
159	>	painCave.severity = OPENMD_ERROR;
160	>	simError();
161	>	}
162
163	<	std::map<std::string, RNEMDTypeEnum>::iterator i;
164	<	i = stringToEnumMap_.find(st);
165	<	rnemdType_ = (i == stringToEnumMap_.end()) ? RNEMD::rnemdUnknown : i->second;
166	<	if (rnemdType_ == rnemdUnknown) {
167	<	std::cerr << "WARNING! RNEMD Type Unknown!\n";
163	>	map<string, RNEMDFluxType>::iterator j;
164	>	j = stringToFluxType_.find(fluxStr);
165	>	if (j != stringToFluxType_.end())
166	>	rnemdFluxType_ = j->second;
167	>	else {
168	>	sprintf(painCave.errMsg,
169	>	"RNEMD: The current fluxType,\n"
170	>	"\t\t%s\n"
171	>	"\tis not one of the recognized flux types.\n",
172	>	fluxStr.c_str());
173	>	painCave.isFatal = 1;
174	>	painCave.severity = OPENMD_ERROR;
175	>	simError();
176		}
177
178	<	#ifdef IS_MPI
179	<	if (worldRank == 0) {
180	<	#endif
181	<
182	<	std::string rnemdFileName;
183	<	std::string xTempFileName;
184	<	std::string yTempFileName;
122	<	std::string zTempFileName;
123	<	switch(rnemdType_) {
124	<	case rnemdKineticSwap :
125	<	case rnemdKineticScale :
126	<	rnemdFileName = "temperature.log";
178	>	bool methodFluxMismatch = false;
179	>	bool hasCorrectFlux = false;
180	>	switch(rnemdMethod_) {
181	>	case rnemdSwap:
182	>	switch (rnemdFluxType_) {
183	>	case rnemdKE:
184	>	hasCorrectFlux = hasKineticFlux;
185		break;
186	<	case rnemdPx :
187	<	case rnemdPxScale :
188	<	case rnemdPy :
189	<	case rnemdPyScale :
132	<	rnemdFileName = "momemtum.log";
133	<	xTempFileName = "temperatureX.log";
134	<	yTempFileName = "temperatureY.log";
135	<	zTempFileName = "temperatureZ.log";
136	<	xTempLog_.open(xTempFileName.c_str());
137	<	yTempLog_.open(yTempFileName.c_str());
138	<	zTempLog_.open(zTempFileName.c_str());
186	>	case rnemdPx:
187	>	case rnemdPy:
188	>	case rnemdPz:
189	>	hasCorrectFlux = hasMomentumFlux;
190		break;
140	–	case rnemdPz :
141	–	case rnemdPzScale :
142	–	case rnemdUnknown :
191		default :
192	<	rnemdFileName = "rnemd.log";
192	>	methodFluxMismatch = true;
193		break;
194		}
195	<	rnemdLog_.open(rnemdFileName.c_str());
196	<
197	<	#ifdef IS_MPI
195	>	break;
196	>	case rnemdNIVS:
197	>	switch (rnemdFluxType_) {
198	>	case rnemdKE:
199	>	case rnemdRotKE:
200	>	case rnemdFullKE:
201	>	hasCorrectFlux = hasKineticFlux;
202	>	break;
203	>	case rnemdPx:
204	>	case rnemdPy:
205	>	case rnemdPz:
206	>	hasCorrectFlux = hasMomentumFlux;
207	>	break;
208	>	case rnemdKePx:
209	>	case rnemdKePy:
210	>	hasCorrectFlux = hasMomentumFlux && hasKineticFlux;
211	>	break;
212	>	default:
213	>	methodFluxMismatch = true;
214	>	break;
215	>	}
216	>	break;
217	>	case rnemdVSS:
218	>	switch (rnemdFluxType_) {
219	>	case rnemdKE:
220	>	case rnemdRotKE:
221	>	case rnemdFullKE:
222	>	hasCorrectFlux = hasKineticFlux;
223	>	break;
224	>	case rnemdPx:
225	>	case rnemdPy:
226	>	case rnemdPz:
227	>	hasCorrectFlux = hasMomentumFlux;
228	>	break;
229	>	case rnemdLx:
230	>	case rnemdLy:
231	>	case rnemdLz:
232	>	hasCorrectFlux = hasAngularMomentumFlux;
233	>	break;
234	>	case rnemdPvector:
235	>	hasCorrectFlux = hasMomentumFluxVector;
236	>	break;
237	>	case rnemdLvector:
238	>	hasCorrectFlux = hasAngularMomentumFluxVector;
239	>	break;
240	>	case rnemdKePx:
241	>	case rnemdKePy:
242	>	hasCorrectFlux = hasMomentumFlux && hasKineticFlux;
243	>	break;
244	>	case rnemdKeLx:
245	>	case rnemdKeLy:
246	>	case rnemdKeLz:
247	>	hasCorrectFlux = hasAngularMomentumFlux && hasKineticFlux;
248	>	break;
249	>	case rnemdKePvector:
250	>	hasCorrectFlux = hasMomentumFluxVector && hasKineticFlux;
251	>	break;
252	>	case rnemdKeLvector:
253	>	hasCorrectFlux = hasAngularMomentumFluxVector && hasKineticFlux;
254	>	break;
255	>	default:
256	>	methodFluxMismatch = true;
257	>	break;
258	>	}
259	>	default:
260	>	break;
261		}
151	–	#endif
262
263	<	set_RNEMD_exchange_time(simParams->getRNEMD_exchangeTime());
264	<	set_RNEMD_nBins(simParams->getRNEMD_nBins());
265	<	midBin_ = nBins_ / 2;
266	<	if (simParams->haveRNEMD_logWidth()) {
267	<	rnemdLogWidth_ = simParams->getRNEMD_logWidth();
268	<	if (rnemdLogWidth_ != nBins_ && rnemdLogWidth_ != midBin_ + 1) {
269	<	std::cerr << "WARNING! RNEMD_logWidth has abnormal value!\n";
270	<	std::cerr << "Automaically set back to default.\n";
271	<	rnemdLogWidth_ = nBins_;
272	<	}
263	>	if (methodFluxMismatch) {
264	>	sprintf(painCave.errMsg,
265	>	"RNEMD: The current method,\n"
266	>	"\t\t%s\n"
267	>	"\tcannot be used with the current flux type, %s\n",
268	>	methStr.c_str(), fluxStr.c_str());
269	>	painCave.isFatal = 1;
270	>	painCave.severity = OPENMD_ERROR;
271	>	simError();
272	>	}
273	>	if (!hasCorrectFlux) {
274	>	sprintf(painCave.errMsg,
275	>	"RNEMD: The current method, %s, and flux type, %s,\n"
276	>	"\tdid not have the correct flux value specified. Options\n"
277	>	"\tinclude: kineticFlux, momentumFlux, angularMomentumFlux,\n"
278	>	"\tmomentumFluxVector, and angularMomentumFluxVector.\n",
279	>	methStr.c_str(), fluxStr.c_str());
280	>	painCave.isFatal = 1;
281	>	painCave.severity = OPENMD_ERROR;
282	>	simError();
283	>	}
284	>
285	>	if (hasKineticFlux) {
286	>	// convert the kcal / mol / Angstroms^2 / fs values in the md file
287	>	// into  amu / fs^3:
288	>	kineticFlux_ = rnemdParams->getKineticFlux()
289	>	* PhysicalConstants::energyConvert;
290		} else {
291	<	rnemdLogWidth_ = nBins_;
291	>	kineticFlux_ = 0.0;
292		}
293	<	valueHist_.resize(rnemdLogWidth_, 0.0);
294	<	valueCount_.resize(rnemdLogWidth_, 0);
295	<	xTempHist_.resize(rnemdLogWidth_, 0.0);
296	<	yTempHist_.resize(rnemdLogWidth_, 0.0);
297	<	zTempHist_.resize(rnemdLogWidth_, 0.0);
298	<
299	<	set_RNEMD_exchange_total(0.0);
300	<	if (simParams->haveRNEMD_targetFlux()) {
301	<	set_RNEMD_target_flux(simParams->getRNEMD_targetFlux());
293	>	if (hasMomentumFluxVector) {
294	>	std::vector<RealType> mf = rnemdParams->getMomentumFluxVector();
295	>	if (mf.size() != 3) {
296	>	sprintf(painCave.errMsg,
297	>	"RNEMD: Incorrect number of parameters specified for momentumFluxVector.\n"
298	>	"\tthere should be 3 parameters, but %lu were specified.\n",
299	>	mf.size());
300	>	painCave.isFatal = 1;
301	>	simError();
302	>	}
303	>	momentumFluxVector_.x() = mf[0];
304	>	momentumFluxVector_.y() = mf[1];
305	>	momentumFluxVector_.z() = mf[2];
306		} else {
307	<	set_RNEMD_target_flux(0.0);
307	>	momentumFluxVector_ = V3Zero;
308	>	if (hasMomentumFlux) {
309	>	RealType momentumFlux = rnemdParams->getMomentumFlux();
310	>	switch (rnemdFluxType_) {
311	>	case rnemdPx:
312	>	momentumFluxVector_.x() = momentumFlux;
313	>	break;
314	>	case rnemdPy:
315	>	momentumFluxVector_.y() = momentumFlux;
316	>	break;
317	>	case rnemdPz:
318	>	momentumFluxVector_.z() = momentumFlux;
319	>	break;
320	>	case rnemdKePx:
321	>	momentumFluxVector_.x() = momentumFlux;
322	>	break;
323	>	case rnemdKePy:
324	>	momentumFluxVector_.y() = momentumFlux;
325	>	break;
326	>	default:
327	>	break;
328	>	}
329	>	}
330	>	if (hasAngularMomentumFluxVector) {
331	>	std::vector<RealType> amf = rnemdParams->getAngularMomentumFluxVector();
332	>	if (amf.size() != 3) {
333	>	sprintf(painCave.errMsg,
334	>	"RNEMD: Incorrect number of parameters specified for angularMomentumFluxVector.\n"
335	>	"\tthere should be 3 parameters, but %lu were specified.\n",
336	>	amf.size());
337	>	painCave.isFatal = 1;
338	>	simError();
339	>	}
340	>	angularMomentumFluxVector_.x() = amf[0];
341	>	angularMomentumFluxVector_.y() = amf[1];
342	>	angularMomentumFluxVector_.z() = amf[2];
343	>	} else {
344	>	angularMomentumFluxVector_ = V3Zero;
345	>	if (hasAngularMomentumFlux) {
346	>	RealType angularMomentumFlux = rnemdParams->getAngularMomentumFlux();
347	>	switch (rnemdFluxType_) {
348	>	case rnemdLx:
349	>	angularMomentumFluxVector_.x() = angularMomentumFlux;
350	>	break;
351	>	case rnemdLy:
352	>	angularMomentumFluxVector_.y() = angularMomentumFlux;
353	>	break;
354	>	case rnemdLz:
355	>	angularMomentumFluxVector_.z() = angularMomentumFlux;
356	>	break;
357	>	case rnemdKeLx:
358	>	angularMomentumFluxVector_.x() = angularMomentumFlux;
359	>	break;
360	>	case rnemdKeLy:
361	>	angularMomentumFluxVector_.y() = angularMomentumFlux;
362	>	break;
363	>	case rnemdKeLz:
364	>	angularMomentumFluxVector_.z() = angularMomentumFlux;
365	>	break;
366	>	default:
367	>	break;
368	>	}
369	>	}
370	>	}
371	>
372	>	if (hasCoordinateOrigin) {
373	>	std::vector<RealType> co = rnemdParams->getCoordinateOrigin();
374	>	if (co.size() != 3) {
375	>	sprintf(painCave.errMsg,
376	>	"RNEMD: Incorrect number of parameters specified for coordinateOrigin.\n"
377	>	"\tthere should be 3 parameters, but %lu were specified.\n",
378	>	co.size());
379	>	painCave.isFatal = 1;
380	>	simError();
381	>	}
382	>	coordinateOrigin_.x() = co[0];
383	>	coordinateOrigin_.y() = co[1];
384	>	coordinateOrigin_.z() = co[2];
385	>	} else {
386	>	coordinateOrigin_ = V3Zero;
387	>	}
388	>
389	>	// do some sanity checking
390	>
391	>	int selectionCount = seleMan_.getSelectionCount();
392	>
393	>	int nIntegrable = info->getNGlobalIntegrableObjects();
394	>
395	>	if (selectionCount > nIntegrable) {
396	>	sprintf(painCave.errMsg,
397	>	"RNEMD: The current objectSelection,\n"
398	>	"\t\t%s\n"
399	>	"\thas resulted in %d selected objects.  However,\n"
400	>	"\tthe total number of integrable objects in the system\n"
401	>	"\tis only %d.  This is almost certainly not what you want\n"
402	>	"\tto do.  A likely cause of this is forgetting the _RB_0\n"
403	>	"\tselector in the selection script!\n",
404	>	rnemdObjectSelection_.c_str(),
405	>	selectionCount, nIntegrable);
406	>	painCave.isFatal = 0;
407	>	painCave.severity = OPENMD_WARNING;
408	>	simError();
409	>	}
410	>
411	>	areaAccumulator_ = new Accumulator();
412	>
413	>	nBins_ = rnemdParams->getOutputBins();
414	>	binWidth_ = rnemdParams->getOutputBinWidth();
415	>
416	>	data_.resize(RNEMD::ENDINDEX);
417	>	OutputData z;
418	>	z.units =  "Angstroms";
419	>	z.title =  "Z";
420	>	z.dataType = "RealType";
421	>	z.accumulator.reserve(nBins_);
422	>	for (int i = 0; i < nBins_; i++)
423	>	z.accumulator.push_back( new Accumulator() );
424	>	data_[Z] = z;
425	>	outputMap_["Z"] =  Z;
426	>
427	>	OutputData r;
428	>	r.units =  "Angstroms";
429	>	r.title =  "R";
430	>	r.dataType = "RealType";
431	>	r.accumulator.reserve(nBins_);
432	>	for (int i = 0; i < nBins_; i++)
433	>	r.accumulator.push_back( new Accumulator() );
434	>	data_[R] = r;
435	>	outputMap_["R"] =  R;
436	>
437	>	OutputData temperature;
438	>	temperature.units =  "K";
439	>	temperature.title =  "Temperature";
440	>	temperature.dataType = "RealType";
441	>	temperature.accumulator.reserve(nBins_);
442	>	for (int i = 0; i < nBins_; i++)
443	>	temperature.accumulator.push_back( new Accumulator() );
444	>	data_[TEMPERATURE] = temperature;
445	>	outputMap_["TEMPERATURE"] =  TEMPERATURE;
446	>
447	>	OutputData velocity;
448	>	velocity.units = "angstroms/fs";
449	>	velocity.title =  "Velocity";
450	>	velocity.dataType = "Vector3d";
451	>	velocity.accumulator.reserve(nBins_);
452	>	for (int i = 0; i < nBins_; i++)
453	>	velocity.accumulator.push_back( new VectorAccumulator() );
454	>	data_[VELOCITY] = velocity;
455	>	outputMap_["VELOCITY"] = VELOCITY;
456	>
457	>	OutputData angularVelocity;
458	>	angularVelocity.units = "angstroms^2/fs";
459	>	angularVelocity.title =  "AngularVelocity";
460	>	angularVelocity.dataType = "Vector3d";
461	>	angularVelocity.accumulator.reserve(nBins_);
462	>	for (int i = 0; i < nBins_; i++)
463	>	angularVelocity.accumulator.push_back( new VectorAccumulator() );
464	>	data_[ANGULARVELOCITY] = angularVelocity;
465	>	outputMap_["ANGULARVELOCITY"] = ANGULARVELOCITY;
466	>
467	>	OutputData density;
468	>	density.units =  "g cm^-3";
469	>	density.title =  "Density";
470	>	density.dataType = "RealType";
471	>	density.accumulator.reserve(nBins_);
472	>	for (int i = 0; i < nBins_; i++)
473	>	density.accumulator.push_back( new Accumulator() );
474	>	data_[DENSITY] = density;
475	>	outputMap_["DENSITY"] =  DENSITY;
476	>
477	>	if (hasOutputFields) {
478	>	parseOutputFileFormat(rnemdParams->getOutputFields());
479	>	} else {
480	>	if (usePeriodicBoundaryConditions_)
481	>	outputMask_.set(Z);
482	>	else
483	>	outputMask_.set(R);
484	>	switch (rnemdFluxType_) {
485	>	case rnemdKE:
486	>	case rnemdRotKE:
487	>	case rnemdFullKE:
488	>	outputMask_.set(TEMPERATURE);
489	>	break;
490	>	case rnemdPx:
491	>	case rnemdPy:
492	>	outputMask_.set(VELOCITY);
493	>	break;
494	>	case rnemdPz:
495	>	case rnemdPvector:
496	>	outputMask_.set(VELOCITY);
497	>	outputMask_.set(DENSITY);
498	>	break;
499	>	case rnemdLx:
500	>	case rnemdLy:
501	>	case rnemdLz:
502	>	case rnemdLvector:
503	>	outputMask_.set(ANGULARVELOCITY);
504	>	break;
505	>	case rnemdKeLx:
506	>	case rnemdKeLy:
507	>	case rnemdKeLz:
508	>	case rnemdKeLvector:
509	>	outputMask_.set(TEMPERATURE);
510	>	outputMask_.set(ANGULARVELOCITY);
511	>	break;
512	>	case rnemdKePx:
513	>	case rnemdKePy:
514	>	outputMask_.set(TEMPERATURE);
515	>	outputMask_.set(VELOCITY);
516	>	break;
517	>	case rnemdKePvector:
518	>	outputMask_.set(TEMPERATURE);
519	>	outputMask_.set(VELOCITY);
520	>	outputMask_.set(DENSITY);
521	>	break;
522	>	default:
523	>	break;
524	>	}
525	>	}
526	>
527	>	if (hasOutputFileName) {
528	>	rnemdFileName_ = rnemdParams->getOutputFileName();
529	>	} else {
530	>	rnemdFileName_ = getPrefix(info->getFinalConfigFileName()) + ".rnemd";
531	>	}
532	>
533	>	exchangeTime_ = rnemdParams->getExchangeTime();
534	>
535	>	Snapshot* currentSnap_ = info->getSnapshotManager()->getCurrentSnapshot();
536	>	// total exchange sums are zeroed out at the beginning:
537	>
538	>	kineticExchange_ = 0.0;
539	>	momentumExchange_ = V3Zero;
540	>	angularMomentumExchange_ = V3Zero;
541	>
542	>	std::ostringstream selectionAstream;
543	>	std::ostringstream selectionBstream;
544	>
545	>	if (hasSelectionA_) {
546	>	selectionA_ = rnemdParams->getSelectionA();
547	>	} else {
548	>	if (usePeriodicBoundaryConditions_) {
549	>	Mat3x3d hmat = currentSnap_->getHmat();
550	>
551	>	if (hasSlabWidth)
552	>	slabWidth_ = rnemdParams->getSlabWidth();
553	>	else
554	>	slabWidth_ = hmat(2,2) / 10.0;
555	>
556	>	if (hasSlabACenter)
557	>	slabACenter_ = rnemdParams->getSlabACenter();
558	>	else
559	>	slabACenter_ = 0.0;
560	>
561	>	selectionAstream << "select wrappedz > "
562	>	<< slabACenter_ - 0.5*slabWidth_
563	>	<<  " && wrappedz < "
564	>	<< slabACenter_ + 0.5*slabWidth_;
565	>	selectionA_ = selectionAstream.str();
566	>	} else {
567	>	if (hasSphereARadius)
568	>	sphereARadius_ = rnemdParams->getSphereARadius();
569	>	else {
570	>	// use an initial guess to the size of the inner slab to be 1/10 the
571	>	// radius of an approximately spherical hull:
572	>	Thermo thermo(info);
573	>	RealType hVol = thermo.getHullVolume();
574	>	sphereARadius_ = 0.1 * pow((3.0 * hVol / (4.0 * M_PI)), 1.0/3.0);
575	>	}
576	>	selectionAstream << "select r < " << sphereARadius_;
577	>	selectionA_ = selectionAstream.str();
578	>	}
579	>	}
580	>
581	>	if (hasSelectionB_) {
582	>	selectionB_ = rnemdParams->getSelectionB();
583	>
584	>	} else {
585	>	if (usePeriodicBoundaryConditions_) {
586	>	Mat3x3d hmat = currentSnap_->getHmat();
587	>
588	>	if (hasSlabWidth)
589	>	slabWidth_ = rnemdParams->getSlabWidth();
590	>	else
591	>	slabWidth_ = hmat(2,2) / 10.0;
592	>
593	>	if (hasSlabBCenter)
594	>	slabBCenter_ = rnemdParams->getSlabBCenter();
595	>	else
596	>	slabBCenter_ = hmat(2,2) / 2.0;
597	>
598	>	selectionBstream << "select wrappedz > "
599	>	<< slabBCenter_ - 0.5*slabWidth_
600	>	<<  " && wrappedz < "
601	>	<< slabBCenter_ + 0.5*slabWidth_;
602	>	selectionB_ = selectionBstream.str();
603	>	} else {
604	>	if (hasSphereBRadius_) {
605	>	sphereBRadius_ = rnemdParams->getSphereBRadius();
606	>	selectionBstream << "select r > " << sphereBRadius_;
607	>	selectionB_ = selectionBstream.str();
608	>	} else {
609	>	selectionB_ = "select hull";
610	>	BisHull_ = true;
611	>	hasSelectionB_ = true;
612	>	}
613	>	}
614	>	}
615		}
616
617	<	#ifndef IS_MPI
618	<	if (simParams->haveSeed()) {
619	<	seedValue = simParams->getSeed();
620	<	randNumGen_ = new SeqRandNumGen(seedValue);
621	<	}else {
622	<	randNumGen_ = new SeqRandNumGen();
623	<	}
624	<	#else
625	<	if (simParams->haveSeed()) {
188	<	seedValue = simParams->getSeed();
189	<	randNumGen_ = new ParallelRandNumGen(seedValue);
190	<	}else {
191	<	randNumGen_ = new ParallelRandNumGen();
192	<	}
193	<	#endif
617	>	// object evaluator:
618	>	evaluator_.loadScriptString(rnemdObjectSelection_);
619	>	seleMan_.setSelectionSet(evaluator_.evaluate());
620	>	evaluatorA_.loadScriptString(selectionA_);
621	>	evaluatorB_.loadScriptString(selectionB_);
622	>	seleManA_.setSelectionSet(evaluatorA_.evaluate());
623	>	seleManB_.setSelectionSet(evaluatorB_.evaluate());
624	>	commonA_ = seleManA_ & seleMan_;
625	>	commonB_ = seleManB_ & seleMan_;
626		}
627
196	–	RNEMD::~RNEMD() {
197	–	delete randNumGen_;
628
629	+	RNEMD::~RNEMD() {
630	+	if (!doRNEMD_) return;
631		#ifdef IS_MPI
632		if (worldRank == 0) {
633		#endif
634	<	std::cerr << "total fail trials: " << failTrialCount_ << "\n";
635	<	rnemdLog_.close();
636	<	if (rnemdType_ == rnemdKineticScale || rnemdType_ == rnemdPxScale || rnemdType_ == rnemdPyScale)
637	<	std::cerr<< "total root-checking warnings: " << failRootCount_ << "\n";
638	<	if (rnemdType_ == rnemdPx || rnemdType_ == rnemdPxScale || rnemdType_ == rnemdPy || rnemdType_ == rnemdPyScale) {
207	<	xTempLog_.close();
208	<	yTempLog_.close();
209	<	zTempLog_.close();
210	<	}
634	>
635	>	writeOutputFile();
636	>
637	>	rnemdFile_.close();
638	>
639		#ifdef IS_MPI
640		}
641		#endif
642	+
643	+	// delete all of the objects we created:
644	+	delete areaAccumulator_;
645	+	data_.clear();
646		}
647	+
648	+	void RNEMD::doSwap(SelectionManager& smanA, SelectionManager& smanB) {
649	+	if (!doRNEMD_) return;
650	+	int selei;
651	+	int selej;
652
216	–	void RNEMD::doSwap() {
217	–
653		Snapshot* currentSnap_ = info_->getSnapshotManager()->getCurrentSnapshot();
654		Mat3x3d hmat = currentSnap_->getHmat();
655
221	–	seleMan_.setSelectionSet(evaluator_.evaluate());
222	–
223	–	int selei;
656		StuntDouble* sd;
225	–	int idx;
657
658		RealType min_val;
659	<	bool min_found = false;
659	>	int min_found = 0;
660		StuntDouble* min_sd;
661
662		RealType max_val;
663	<	bool max_found = false;
663	>	int max_found = 0;
664		StuntDouble* max_sd;
665
666	<	for (sd = seleMan_.beginSelected(selei); sd != NULL;
667	<	sd = seleMan_.nextSelected(selei)) {
666	>	for (sd = seleManA_.beginSelected(selei); sd != NULL;
667	>	sd = seleManA_.nextSelected(selei)) {
668
238	–	idx = sd->getLocalIndex();
239	–
669		Vector3d pos = sd->getPos();
670	<
670	>
671		// wrap the stuntdouble's position back into the box:
672	<
672	>
673		if (usePeriodicBoundaryConditions_)
674		currentSnap_->wrapVector(pos);
675	<
676	<	// which bin is this stuntdouble in?
677	<	// wrapped positions are in the range [-0.5*hmat(2,2), +0.5*hmat(2,2)]
678	<
679	<	int binNo = int(nBins_ * (pos.z() / hmat(2,2) + 0.5)) % nBins_;
680	<
681	<
253	<	// if we're in bin 0 or the middleBin
254	<	if (binNo == 0 || binNo == midBin_) {
675	>
676	>	RealType mass = sd->getMass();
677	>	Vector3d vel = sd->getVel();
678	>	RealType value;
679	>
680	>	switch(rnemdFluxType_) {
681	>	case rnemdKE :
682
683	<	RealType mass = sd->getMass();
684	<	Vector3d vel = sd->getVel();
685	<	RealType value;
686	<
687	<	switch(rnemdType_) {
261	<	case rnemdKineticSwap :
683	>	value = mass * vel.lengthSquare();
684	>
685	>	if (sd->isDirectional()) {
686	>	Vector3d angMom = sd->getJ();
687	>	Mat3x3d I = sd->getI();
688
689	<	value = mass * (vel[0]*vel[0] + vel[1]*vel[1] +
690	<	vel[2]*vel[2]);
691	<	if (sd->isDirectional()) {
692	<	Vector3d angMom = sd->getJ();
693	<	Mat3x3d I = sd->getI();
694	<
695	<	if (sd->isLinear()) {
696	<	int i = sd->linearAxis();
697	<	int j = (i + 1) % 3;
698	<	int k = (i + 2) % 3;
273	<	value += angMom[j] * angMom[j] / I(j, j) +
274	<	angMom[k] * angMom[k] / I(k, k);
275	<	} else {
276	<	value += angMom[0]*angMom[0]/I(0, 0)
277	<	+ angMom[1]*angMom[1]/I(1, 1)
278	<	+ angMom[2]*angMom[2]/I(2, 2);
279	<	}
689	>	if (sd->isLinear()) {
690	>	int i = sd->linearAxis();
691	>	int j = (i + 1) % 3;
692	>	int k = (i + 2) % 3;
693	>	value += angMom[j] * angMom[j] / I(j, j) +
694	>	angMom[k] * angMom[k] / I(k, k);
695	>	} else {
696	>	value += angMom[0]*angMom[0]/I(0, 0)
697	>	+ angMom[1]*angMom[1]/I(1, 1)
698	>	+ angMom[2]*angMom[2]/I(2, 2);
699		}
700	<	//make exchangeSum_ comparable between swap & scale
701	<	//temporarily without using energyConvert
702	<	//value = value * 0.5 / PhysicalConstants::energyConvert;
703	<	value *= 0.5;
704	<	break;
705	<	case rnemdPx :
706	<	value = mass * vel[0];
707	<	break;
708	<	case rnemdPy :
709	<	value = mass * vel[1];
710	<	break;
711	<	case rnemdPz :
712	<	value = mass * vel[2];
713	<	break;
714	<	default :
715	<	break;
700	>	} //angular momenta exchange enabled
701	>	value *= 0.5;
702	>	break;
703	>	case rnemdPx :
704	>	value = mass * vel[0];
705	>	break;
706	>	case rnemdPy :
707	>	value = mass * vel[1];
708	>	break;
709	>	case rnemdPz :
710	>	value = mass * vel[2];
711	>	break;
712	>	default :
713	>	break;
714	>	}
715	>	if (!max_found) {
716	>	max_val = value;
717	>	max_sd = sd;
718	>	max_found = 1;
719	>	} else {
720	>	if (max_val < value) {
721	>	max_val = value;
722	>	max_sd = sd;
723		}
724	+	}
725	+	}
726
727	<	if (binNo == 0) {
728	<	if (!min_found) {
729	<	min_val = value;
730	<	min_sd = sd;
731	<	min_found = true;
732	<	} else {
733	<	if (min_val > value) {
734	<	min_val = value;
735	<	min_sd = sd;
736	<	}
737	<	}
738	<	} else { //midBin_
739	<	if (!max_found) {
740	<	max_val = value;
741	<	max_sd = sd;
742	<	max_found = true;
743	<	} else {
744	<	if (max_val < value) {
745	<	max_val = value;
746	<	max_sd = sd;
747	<	}
748	<	}
749	<	}
727	>	for (sd = seleManB_.beginSelected(selej); sd != NULL;
728	>	sd = seleManB_.nextSelected(selej)) {
729	>
730	>	Vector3d pos = sd->getPos();
731	>
732	>	// wrap the stuntdouble's position back into the box:
733	>
734	>	if (usePeriodicBoundaryConditions_)
735	>	currentSnap_->wrapVector(pos);
736	>
737	>	RealType mass = sd->getMass();
738	>	Vector3d vel = sd->getVel();
739	>	RealType value;
740	>
741	>	switch(rnemdFluxType_) {
742	>	case rnemdKE :
743	>
744	>	value = mass * vel.lengthSquare();
745	>
746	>	if (sd->isDirectional()) {
747	>	Vector3d angMom = sd->getJ();
748	>	Mat3x3d I = sd->getI();
749	>
750	>	if (sd->isLinear()) {
751	>	int i = sd->linearAxis();
752	>	int j = (i + 1) % 3;
753	>	int k = (i + 2) % 3;
754	>	value += angMom[j] * angMom[j] / I(j, j) +
755	>	angMom[k] * angMom[k] / I(k, k);
756	>	} else {
757	>	value += angMom[0]*angMom[0]/I(0, 0)
758	>	+ angMom[1]*angMom[1]/I(1, 1)
759	>	+ angMom[2]*angMom[2]/I(2, 2);
760	>	}
761	>	} //angular momenta exchange enabled
762	>	value *= 0.5;
763	>	break;
764	>	case rnemdPx :
765	>	value = mass * vel[0];
766	>	break;
767	>	case rnemdPy :
768	>	value = mass * vel[1];
769	>	break;
770	>	case rnemdPz :
771	>	value = mass * vel[2];
772	>	break;
773	>	default :
774	>	break;
775		}
776	+
777	+	if (!min_found) {
778	+	min_val = value;
779	+	min_sd = sd;
780	+	min_found = 1;
781	+	} else {
782	+	if (min_val > value) {
783	+	min_val = value;
784	+	min_sd = sd;
785	+	}
786	+	}
787		}
788	+
789	+	#ifdef IS_MPI
790	+	int worldRank;
791	+	MPI_Comm_rank( MPI_COMM_WORLD, &worldRank);
792	+
793	+	int my_min_found = min_found;
794	+	int my_max_found = max_found;
795
325	–	#ifdef IS_MPI
326	–	int nProc, worldRank;
327	–
328	–	nProc = MPI::COMM_WORLD.Get_size();
329	–	worldRank = MPI::COMM_WORLD.Get_rank();
330	–
331	–	bool my_min_found = min_found;
332	–	bool my_max_found = max_found;
333	–
796		// Even if we didn't find a minimum, did someone else?
797	<	MPI::COMM_WORLD.Allreduce(&my_min_found, &min_found,
798	<	1, MPI::BOOL, MPI::LAND);
337	<
797	>	MPI_Allreduce(&my_min_found, &min_found, 1, MPI_INT, MPI_LOR,
798	>	MPI_COMM_WORLD);
799		// Even if we didn't find a maximum, did someone else?
800	<	MPI::COMM_WORLD.Allreduce(&my_max_found, &max_found,
801	<	1, MPI::BOOL, MPI::LAND);
802	<
803	<	struct {
804	<	RealType val;
805	<	int rank;
806	<	} max_vals, min_vals;
807	<
808	<	if (min_found) {
809	<	if (my_min_found)
800	>	MPI_Allreduce(&my_max_found, &max_found, 1, MPI_INT, MPI_LOR,
801	>	MPI_COMM_WORLD);
802	>	#endif
803	>
804	>	if (max_found && min_found) {
805	>
806	>	#ifdef IS_MPI
807	>	struct {
808	>	RealType val;
809	>	int rank;
810	>	} max_vals, min_vals;
811	>
812	>	if (my_min_found) {
813		min_vals.val = min_val;
814	<	else
814	>	} else {
815		min_vals.val = HONKING_LARGE_VALUE;
816	<
816	>	}
817		min_vals.rank = worldRank;
818
819		// Who had the minimum?
820	<	MPI::COMM_WORLD.Allreduce(&min_vals, &min_vals,
821	<	1, MPI::REALTYPE_INT, MPI::MINLOC);
820	>	MPI_Allreduce(&min_vals, &min_vals,
821	>	1, MPI_REALTYPE_INT, MPI_MINLOC, MPI_COMM_WORLD);
822		min_val = min_vals.val;
359	–	}
823
824	<	if (max_found) {
362	<	if (my_max_found)
824	>	if (my_max_found) {
825		max_vals.val = max_val;
826	<	else
826	>	} else {
827		max_vals.val = -HONKING_LARGE_VALUE;
828	<
828	>	}
829		max_vals.rank = worldRank;
830
831		// Who had the maximum?
832	<	MPI::COMM_WORLD.Allreduce(&max_vals, &max_vals,
833	<	1, MPI::REALTYPE_INT, MPI::MAXLOC);
832	>	MPI_Allreduce(&max_vals, &max_vals,
833	>	1, MPI_REALTYPE_INT, MPI_MAXLOC, MPI_COMM_WORLD);
834		max_val = max_vals.val;
373	–	}
835		#endif
836	<
837	<	if (max_found && min_found) {
838	<	if (min_val< max_val) {
378	<
836	>
837	>	if (min_val < max_val) {
838	>
839		#ifdef IS_MPI
840		if (max_vals.rank == worldRank && min_vals.rank == worldRank) {
841		// I have both maximum and minimum, so proceed like a single
842		// processor version:
843		#endif
844	<	// objects to be swapped: velocity & angular velocity
844	>
845		Vector3d min_vel = min_sd->getVel();
846		Vector3d max_vel = max_sd->getVel();
847		RealType temp_vel;
848
849	<	switch(rnemdType_) {
850	<	case rnemdKineticSwap :
849	>	switch(rnemdFluxType_) {
850	>	case rnemdKE :
851		min_sd->setVel(max_vel);
852		max_sd->setVel(min_vel);
853	<	if (min_sd->isDirectional() && max_sd->isDirectional()) {
853	>	if (min_sd->isDirectional() && max_sd->isDirectional()) {
854		Vector3d min_angMom = min_sd->getJ();
855		Vector3d max_angMom = max_sd->getJ();
856		min_sd->setJ(max_angMom);
857		max_sd->setJ(min_angMom);
858	<	}
858	>	}//angular momenta exchange enabled
859	>	//assumes same rigid body identity
860		break;
861		case rnemdPx :
862		temp_vel = min_vel.x();
#	Line 421 | Line 882 | namespace OpenMD {
882		default :
883		break;
884		}
885	+
886		#ifdef IS_MPI
887		// the rest of the cases only apply in parallel simulations:
888		} else if (max_vals.rank == worldRank) {
#	Line 428 | Line 890 | namespace OpenMD {
890
891		Vector3d min_vel;
892		Vector3d max_vel = max_sd->getVel();
893	<	MPI::Status status;
893	>	MPI_Status status;
894
895		// point-to-point swap of the velocity vector
896	<	MPI::COMM_WORLD.Sendrecv(max_vel.getArrayPointer(), 3, MPI::REALTYPE,
897	<	min_vals.rank, 0,
898	<	min_vel.getArrayPointer(), 3, MPI::REALTYPE,
899	<	min_vals.rank, 0, status);
896	>	MPI_Sendrecv(max_vel.getArrayPointer(), 3, MPI_REALTYPE,
897	>	min_vals.rank, 0,
898	>	min_vel.getArrayPointer(), 3, MPI_REALTYPE,
899	>	min_vals.rank, 0, MPI_COMM_WORLD, &status);
900
901	<	switch(rnemdType_) {
902	<	case rnemdKineticSwap :
901	>	switch(rnemdFluxType_) {
902	>	case rnemdKE :
903		max_sd->setVel(min_vel);
904	<
904	>	//angular momenta exchange enabled
905		if (max_sd->isDirectional()) {
906		Vector3d min_angMom;
907		Vector3d max_angMom = max_sd->getJ();
908	<
908	>
909		// point-to-point swap of the angular momentum vector
910	<	MPI::COMM_WORLD.Sendrecv(max_angMom.getArrayPointer(), 3,
911	<	MPI::REALTYPE, min_vals.rank, 1,
912	<	min_angMom.getArrayPointer(), 3,
913	<	MPI::REALTYPE, min_vals.rank, 1,
914	<	status);
915	<
910	>	MPI_Sendrecv(max_angMom.getArrayPointer(), 3,
911	>	MPI_REALTYPE, min_vals.rank, 1,
912	>	min_angMom.getArrayPointer(), 3,
913	>	MPI_REALTYPE, min_vals.rank, 1,
914	>	MPI_COMM_WORLD, &status);
915	>
916		max_sd->setJ(min_angMom);
917	<	}
917	>	}
918		break;
919		case rnemdPx :
920		max_vel.x() = min_vel.x();
#	Line 474 | Line 936 | namespace OpenMD {
936
937		Vector3d max_vel;
938		Vector3d min_vel = min_sd->getVel();
939	<	MPI::Status status;
939	>	MPI_Status status;
940
941		// point-to-point swap of the velocity vector
942	<	MPI::COMM_WORLD.Sendrecv(min_vel.getArrayPointer(), 3, MPI::REALTYPE,
943	<	max_vals.rank, 0,
944	<	max_vel.getArrayPointer(), 3, MPI::REALTYPE,
945	<	max_vals.rank, 0, status);
942	>	MPI_Sendrecv(min_vel.getArrayPointer(), 3, MPI_REALTYPE,
943	>	max_vals.rank, 0,
944	>	max_vel.getArrayPointer(), 3, MPI_REALTYPE,
945	>	max_vals.rank, 0, MPI_COMM_WORLD, &status);
946
947	<	switch(rnemdType_) {
948	<	case rnemdKineticSwap :
947	>	switch(rnemdFluxType_) {
948	>	case rnemdKE :
949		min_sd->setVel(max_vel);
950	<
950	>	//angular momenta exchange enabled
951		if (min_sd->isDirectional()) {
952		Vector3d min_angMom = min_sd->getJ();
953		Vector3d max_angMom;
954	<
954	>
955		// point-to-point swap of the angular momentum vector
956	<	MPI::COMM_WORLD.Sendrecv(min_angMom.getArrayPointer(), 3,
957	<	MPI::REALTYPE, max_vals.rank, 1,
958	<	max_angMom.getArrayPointer(), 3,
959	<	MPI::REALTYPE, max_vals.rank, 1,
960	<	status);
961	<
956	>	MPI_Sendrecv(min_angMom.getArrayPointer(), 3,
957	>	MPI_REALTYPE, max_vals.rank, 1,
958	>	max_angMom.getArrayPointer(), 3,
959	>	MPI_REALTYPE, max_vals.rank, 1,
960	>	MPI_COMM_WORLD, &status);
961	>
962		min_sd->setJ(max_angMom);
963		}
964		break;
#	Line 517 | Line 979 | namespace OpenMD {
979		}
980		}
981		#endif
982	<	exchangeSum_ += max_val - min_val;
983	<	} else {
984	<	std::cerr << "exchange NOT performed!\nmin_val > max_val.\n";
982	>
983	>	switch(rnemdFluxType_) {
984	>	case rnemdKE:
985	>	kineticExchange_ += max_val - min_val;
986	>	break;
987	>	case rnemdPx:
988	>	momentumExchange_.x() += max_val - min_val;
989	>	break;
990	>	case rnemdPy:
991	>	momentumExchange_.y() += max_val - min_val;
992	>	break;
993	>	case rnemdPz:
994	>	momentumExchange_.z() += max_val - min_val;
995	>	break;
996	>	default:
997	>	break;
998	>	}
999	>	} else {
1000	>	sprintf(painCave.errMsg,
1001	>	"RNEMD::doSwap exchange NOT performed because min_val > max_val\n");
1002	>	painCave.isFatal = 0;
1003	>	painCave.severity = OPENMD_INFO;
1004	>	simError();
1005		failTrialCount_++;
1006		}
1007		} else {
1008	<	std::cerr << "exchange NOT performed!\n";
1009	<	std::cerr << "at least one of the two slabs empty.\n";
1008	>	sprintf(painCave.errMsg,
1009	>	"RNEMD::doSwap exchange NOT performed because selected object\n"
1010	>	"\twas not present in at least one of the two slabs.\n");
1011	>	painCave.isFatal = 0;
1012	>	painCave.severity = OPENMD_INFO;
1013	>	simError();
1014		failTrialCount_++;
1015	<	}
530	<
1015	>	}
1016		}
1017
1018	<	void RNEMD::doScale() {
1018	>	void RNEMD::doNIVS(SelectionManager& smanA, SelectionManager& smanB) {
1019	>	if (!doRNEMD_) return;
1020	>	int selei;
1021	>	int selej;
1022
1023		Snapshot* currentSnap_ = info_->getSnapshotManager()->getCurrentSnapshot();
1024	+	RealType time = currentSnap_->getTime();
1025		Mat3x3d hmat = currentSnap_->getHmat();
1026
538	–	seleMan_.setSelectionSet(evaluator_.evaluate());
539	–
540	–	int selei;
1027		StuntDouble* sd;
542	–	int idx;
1028
1029	<	std::vector<StuntDouble*> hotBin, coldBin;
1029	>	vector<StuntDouble*> hotBin, coldBin;
1030
1031		RealType Phx = 0.0;
1032		RealType Phy = 0.0;
#	Line 549 | Line 1034 | namespace OpenMD {
1034		RealType Khx = 0.0;
1035		RealType Khy = 0.0;
1036		RealType Khz = 0.0;
1037	+	RealType Khw = 0.0;
1038		RealType Pcx = 0.0;
1039		RealType Pcy = 0.0;
1040		RealType Pcz = 0.0;
1041		RealType Kcx = 0.0;
1042		RealType Kcy = 0.0;
1043		RealType Kcz = 0.0;
1044	+	RealType Kcw = 0.0;
1045
1046	<	for (sd = seleMan_.beginSelected(selei); sd != NULL;
1047	<	sd = seleMan_.nextSelected(selei)) {
1046	>	for (sd = smanA.beginSelected(selei); sd != NULL;
1047	>	sd = smanA.nextSelected(selei)) {
1048
562	–	idx = sd->getLocalIndex();
563	–
1049		Vector3d pos = sd->getPos();
1050	<
1050	>
1051		// wrap the stuntdouble's position back into the box:
1052	<
1052	>
1053		if (usePeriodicBoundaryConditions_)
1054		currentSnap_->wrapVector(pos);
1055	<
1056	<	// which bin is this stuntdouble in?
1057	<	// wrapped positions are in the range [-0.5*hmat(2,2), +0.5*hmat(2,2)]
1058	<
1059	<	int binNo = int(nBins_ * (pos.z() / hmat(2,2) + 0.5)) % nBins_;
1060	<
1061	<	// if we're in bin 0 or the middleBin
1062	<	if (binNo == 0 || binNo == midBin_) {
1063	<
1064	<	RealType mass = sd->getMass();
1065	<	Vector3d vel = sd->getVel();
1066	<
1067	<	if (binNo == 0) {
1068	<	hotBin.push_back(sd);
1069	<	Phx += mass * vel.x();
1070	<	Phy += mass * vel.y();
1071	<	Phz += mass * vel.z();
1072	<	Khx += mass * vel.x() * vel.x();
1073	<	Khy += mass * vel.y() * vel.y();
1074	<	Khz += mass * vel.z() * vel.z();
1075	<	} else { //midBin_
1076	<	coldBin.push_back(sd);
1077	<	Pcx += mass * vel.x();
1078	<	Pcy += mass * vel.y();
1079	<	Pcz += mass * vel.z();
1080	<	Kcx += mass * vel.x() * vel.x();
596	<	Kcy += mass * vel.y() * vel.y();
597	<	Kcz += mass * vel.z() * vel.z();
598	<	}
1055	>
1056	>
1057	>	RealType mass = sd->getMass();
1058	>	Vector3d vel = sd->getVel();
1059	>
1060	>	hotBin.push_back(sd);
1061	>	Phx += mass * vel.x();
1062	>	Phy += mass * vel.y();
1063	>	Phz += mass * vel.z();
1064	>	Khx += mass * vel.x() * vel.x();
1065	>	Khy += mass * vel.y() * vel.y();
1066	>	Khz += mass * vel.z() * vel.z();
1067	>	if (sd->isDirectional()) {
1068	>	Vector3d angMom = sd->getJ();
1069	>	Mat3x3d I = sd->getI();
1070	>	if (sd->isLinear()) {
1071	>	int i = sd->linearAxis();
1072	>	int j = (i + 1) % 3;
1073	>	int k = (i + 2) % 3;
1074	>	Khw += angMom[j] * angMom[j] / I(j, j) +
1075	>	angMom[k] * angMom[k] / I(k, k);
1076	>	} else {
1077	>	Khw += angMom[0]*angMom[0]/I(0, 0)
1078	>	+ angMom[1]*angMom[1]/I(1, 1)
1079	>	+ angMom[2]*angMom[2]/I(2, 2);
1080	>	}
1081		}
1082		}
1083	+	for (sd = smanB.beginSelected(selej); sd != NULL;
1084	+	sd = smanB.nextSelected(selej)) {
1085	+	Vector3d pos = sd->getPos();
1086	+
1087	+	// wrap the stuntdouble's position back into the box:
1088	+
1089	+	if (usePeriodicBoundaryConditions_)
1090	+	currentSnap_->wrapVector(pos);
1091	+
1092	+	RealType mass = sd->getMass();
1093	+	Vector3d vel = sd->getVel();
1094
1095	+	coldBin.push_back(sd);
1096	+	Pcx += mass * vel.x();
1097	+	Pcy += mass * vel.y();
1098	+	Pcz += mass * vel.z();
1099	+	Kcx += mass * vel.x() * vel.x();
1100	+	Kcy += mass * vel.y() * vel.y();
1101	+	Kcz += mass * vel.z() * vel.z();
1102	+	if (sd->isDirectional()) {
1103	+	Vector3d angMom = sd->getJ();
1104	+	Mat3x3d I = sd->getI();
1105	+	if (sd->isLinear()) {
1106	+	int i = sd->linearAxis();
1107	+	int j = (i + 1) % 3;
1108	+	int k = (i + 2) % 3;
1109	+	Kcw += angMom[j] * angMom[j] / I(j, j) +
1110	+	angMom[k] * angMom[k] / I(k, k);
1111	+	} else {
1112	+	Kcw += angMom[0]*angMom[0]/I(0, 0)
1113	+	+ angMom[1]*angMom[1]/I(1, 1)
1114	+	+ angMom[2]*angMom[2]/I(2, 2);
1115	+	}
1116	+	}
1117	+	}
1118	+
1119		Khx *= 0.5;
1120		Khy *= 0.5;
1121		Khz *= 0.5;
1122	+	Khw *= 0.5;
1123		Kcx *= 0.5;
1124		Kcy *= 0.5;
1125		Kcz *= 0.5;
1126	+	Kcw *= 0.5;
1127
1128		#ifdef IS_MPI
1129	<	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Phx, 1, MPI::REALTYPE, MPI::SUM);
1130	<	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Phy, 1, MPI::REALTYPE, MPI::SUM);
1131	<	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Phz, 1, MPI::REALTYPE, MPI::SUM);
1132	<	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Pcx, 1, MPI::REALTYPE, MPI::SUM);
1133	<	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Pcy, 1, MPI::REALTYPE, MPI::SUM);
1134	<	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Pcz, 1, MPI::REALTYPE, MPI::SUM);
1129	>	MPI_Allreduce(MPI_IN_PLACE, &Phx, 1, MPI_REALTYPE, MPI_SUM, MPI_COMM_WORLD);
1130	>	MPI_Allreduce(MPI_IN_PLACE, &Phy, 1, MPI_REALTYPE, MPI_SUM, MPI_COMM_WORLD);
1131	>	MPI_Allreduce(MPI_IN_PLACE, &Phz, 1, MPI_REALTYPE, MPI_SUM, MPI_COMM_WORLD);
1132	>	MPI_Allreduce(MPI_IN_PLACE, &Pcx, 1, MPI_REALTYPE, MPI_SUM, MPI_COMM_WORLD);
1133	>	MPI_Allreduce(MPI_IN_PLACE, &Pcy, 1, MPI_REALTYPE, MPI_SUM, MPI_COMM_WORLD);
1134	>	MPI_Allreduce(MPI_IN_PLACE, &Pcz, 1, MPI_REALTYPE, MPI_SUM, MPI_COMM_WORLD);
1135
1136	<	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Khx, 1, MPI::REALTYPE, MPI::SUM);
1137	<	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Khy, 1, MPI::REALTYPE, MPI::SUM);
1138	<	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Khz, 1, MPI::REALTYPE, MPI::SUM);
1139	<	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Kcx, 1, MPI::REALTYPE, MPI::SUM);
1140	<	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Kcy, 1, MPI::REALTYPE, MPI::SUM);
1141	<	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &Kcz, 1, MPI::REALTYPE, MPI::SUM);
1136	>	MPI_Allreduce(MPI_IN_PLACE, &Khx, 1, MPI_REALTYPE, MPI_SUM, MPI_COMM_WORLD);
1137	>	MPI_Allreduce(MPI_IN_PLACE, &Khy, 1, MPI_REALTYPE, MPI_SUM, MPI_COMM_WORLD);
1138	>	MPI_Allreduce(MPI_IN_PLACE, &Khz, 1, MPI_REALTYPE, MPI_SUM, MPI_COMM_WORLD);
1139	>	MPI_Allreduce(MPI_IN_PLACE, &Khw, 1, MPI_REALTYPE, MPI_SUM, MPI_COMM_WORLD);
1140	>
1141	>	MPI_Allreduce(MPI_IN_PLACE, &Kcx, 1, MPI_REALTYPE, MPI_SUM, MPI_COMM_WORLD);
1142	>	MPI_Allreduce(MPI_IN_PLACE, &Kcy, 1, MPI_REALTYPE, MPI_SUM, MPI_COMM_WORLD);
1143	>	MPI_Allreduce(MPI_IN_PLACE, &Kcz, 1, MPI_REALTYPE, MPI_SUM, MPI_COMM_WORLD);
1144	>	MPI_Allreduce(MPI_IN_PLACE, &Kcw, 1, MPI_REALTYPE, MPI_SUM, MPI_COMM_WORLD);
1145		#endif
1146
1147	<	//use coldBin coeff's
1147	>	//solve coldBin coeff's first
1148		RealType px = Pcx / Phx;
1149		RealType py = Pcy / Phy;
1150		RealType pz = Pcz / Phz;
1151	+	RealType c, x, y, z;
1152	+	bool successfulScale = false;
1153	+	if ((rnemdFluxType_ == rnemdFullKE) ||
1154	+	(rnemdFluxType_ == rnemdRotKE)) {
1155	+	//may need sanity check Khw & Kcw > 0
1156
1157	<	RealType a000, a110, c0, a001, a111, b01, b11, c1, c;
1158	<	switch(rnemdType_) {
1159	<	case rnemdKineticScale :
1160	<	/*used hotBin coeff's & only scale x & y dimensions
1161	<	RealType px = Phx / Pcx;
635	<	RealType py = Phy / Pcy;
636	<	a110 = Khy;
637	<	c0 = - Khx - Khy - targetFlux_;
638	<	a000 = Khx;
639	<	a111 = Kcy * py * py
640	<	b11 = -2.0 * Kcy * py * (1.0 + py);
641	<	c1 = Kcy * py * (2.0 + py) + Kcx * px * ( 2.0 + px) + targetFlux_;
642	<	b01 = -2.0 * Kcx * px * (1.0 + px);
643	<	a001 = Kcx * px * px;
644	<	*/
1157	>	if (rnemdFluxType_ == rnemdFullKE) {
1158	>	c = 1.0 - kineticTarget_ / (Kcx + Kcy + Kcz + Kcw);
1159	>	} else {
1160	>	c = 1.0 - kineticTarget_ / Kcw;
1161	>	}
1162
1163	<	//scale all three dimensions, let c_x = c_y
1164	<	a000 = Kcx + Kcy;
1165	<	a110 = Kcz;
1166	<	c0 = targetFlux_ - Kcx - Kcy - Kcz;
1167	<	a001 = Khx * px * px + Khy * py * py;
1168	<	a111 = Khz * pz * pz;
1169	<	b01 = -2.0 * (Khx * px * (1.0 + px) + Khy * py * (1.0 + py));
1170	<	b11 = -2.0 * Khz * pz * (1.0 + pz);
1171	<	c1 = Khx * px * (2.0 + px) + Khy * py * (2.0 + py)
1172	<	+ Khz * pz * (2.0 + pz) - targetFlux_;
1173	<	break;
1174	<	case rnemdPxScale :
1175	<	c = 1 - targetFlux_ / Pcx;
1176	<	a000 = Kcy;
1177	<	a110 = Kcz;
1178	<	c0 = Kcx * c * c - Kcx - Kcy - Kcz;
1179	<	a001 = py * py * Khy;
1180	<	a111 = pz * pz * Khz;
1181	<	b01 = -2.0 * Khy * py * (1.0 + py);
1182	<	b11 = -2.0 * Khz * pz * (1.0 + pz);
1183	<	c1 = Khy * py * (2.0 + py) + Khz * pz * (2.0 + pz)
1184	<	+ Khx * (fastpow(c * px - px - 1.0, 2) - 1.0);
1185	<	break;
1186	<	case rnemdPyScale :
1187	<	c = 1 - targetFlux_ / Pcy;
1188	<	a000 = Kcx;
1189	<	a110 = Kcz;
1190	<	c0 = Kcy * c * c - Kcx - Kcy - Kcz;
1191	<	a001 = px * px * Khx;
1192	<	a111 = pz * pz * Khz;
1193	<	b01 = -2.0 * Khx * px * (1.0 + px);
1194	<	b11 = -2.0 * Khz * pz * (1.0 + pz);
1195	<	c1 = Khx * px * (2.0 + px) + Khz * pz * (2.0 + pz)
1196	<	+ Khy * (fastpow(c * py - py - 1.0, 2) - 1.0);
1163	>	if ((c > 0.81) && (c < 1.21)) {//restrict scaling coefficients
1164	>	c = sqrt(c);
1165	>
1166	>	RealType w = 0.0;
1167	>	if (rnemdFluxType_ ==  rnemdFullKE) {
1168	>	x = 1.0 + px * (1.0 - c);
1169	>	y = 1.0 + py * (1.0 - c);
1170	>	z = 1.0 + pz * (1.0 - c);
1171	>	/* more complicated way
1172	>	w = 1.0 + (Kcw - Kcw * c * c - (c * c * (Kcx + Kcy + Kcz
1173	>	+ Khx * px * px + Khy * py * py + Khz * pz * pz)
1174	>	- 2.0 * c * (Khx * px * (1.0 + px) + Khy * py * (1.0 + py)
1175	>	+ Khz * pz * (1.0 + pz)) + Khx * px * (2.0 + px)
1176	>	+ Khy * py * (2.0 + py) + Khz * pz * (2.0 + pz)
1177	>	- Kcx - Kcy - Kcz)) / Khw; the following is simpler
1178	>	*/
1179	>	if ((fabs(x - 1.0) < 0.1) && (fabs(y - 1.0) < 0.1) &&
1180	>	(fabs(z - 1.0) < 0.1)) {
1181	>	w = 1.0 + (kineticTarget_
1182	>	+ Khx * (1.0 - x * x) + Khy * (1.0 - y * y)
1183	>	+ Khz * (1.0 - z * z)) / Khw;
1184	>	}//no need to calculate w if x, y or z is out of range
1185	>	} else {
1186	>	w = 1.0 + kineticTarget_ / Khw;
1187	>	}
1188	>	if ((w > 0.81) && (w < 1.21)) {//restrict scaling coefficients
1189	>	//if w is in the right range, so should be x, y, z.
1190	>	vector<StuntDouble*>::iterator sdi;
1191	>	Vector3d vel;
1192	>	for (sdi = coldBin.begin(); sdi != coldBin.end(); ++sdi) {
1193	>	if (rnemdFluxType_ == rnemdFullKE) {
1194	>	vel = (*sdi)->getVel() * c;
1195	>	(*sdi)->setVel(vel);
1196	>	}
1197	>	if ((*sdi)->isDirectional()) {
1198	>	Vector3d angMom = (*sdi)->getJ() * c;
1199	>	(*sdi)->setJ(angMom);
1200	>	}
1201	>	}
1202	>	w = sqrt(w);
1203	>	for (sdi = hotBin.begin(); sdi != hotBin.end(); ++sdi) {
1204	>	if (rnemdFluxType_ == rnemdFullKE) {
1205	>	vel = (*sdi)->getVel();
1206	>	vel.x() *= x;
1207	>	vel.y() *= y;
1208	>	vel.z() *= z;
1209	>	(*sdi)->setVel(vel);
1210	>	}
1211	>	if ((*sdi)->isDirectional()) {
1212	>	Vector3d angMom = (*sdi)->getJ() * w;
1213	>	(*sdi)->setJ(angMom);
1214	>	}
1215	>	}
1216	>	successfulScale = true;
1217	>	kineticExchange_ += kineticTarget_;
1218	>	}
1219	>	}
1220	>	} else {
1221	>	RealType a000, a110, c0, a001, a111, b01, b11, c1;
1222	>	switch(rnemdFluxType_) {
1223	>	case rnemdKE :
1224	>	/* used hotBin coeff's & only scale x & y dimensions
1225	>	RealType px = Phx / Pcx;
1226	>	RealType py = Phy / Pcy;
1227	>	a110 = Khy;
1228	>	c0 = - Khx - Khy - kineticTarget_;
1229	>	a000 = Khx;
1230	>	a111 = Kcy * py * py;
1231	>	b11 = -2.0 * Kcy * py * (1.0 + py);
1232	>	c1 = Kcy * py * (2.0 + py) + Kcx * px * ( 2.0 + px) + kineticTarget_;
1233	>	b01 = -2.0 * Kcx * px * (1.0 + px);
1234	>	a001 = Kcx * px * px;
1235	>	*/
1236	>	//scale all three dimensions, let c_x = c_y
1237	>	a000 = Kcx + Kcy;
1238	>	a110 = Kcz;
1239	>	c0 = kineticTarget_ - Kcx - Kcy - Kcz;
1240	>	a001 = Khx * px * px + Khy * py * py;
1241	>	a111 = Khz * pz * pz;
1242	>	b01 = -2.0 * (Khx * px * (1.0 + px) + Khy * py * (1.0 + py));
1243	>	b11 = -2.0 * Khz * pz * (1.0 + pz);
1244	>	c1 = Khx * px * (2.0 + px) + Khy * py * (2.0 + py)
1245	>	+ Khz * pz * (2.0 + pz) - kineticTarget_;
1246	>	break;
1247	>	case rnemdPx :
1248	>	c = 1 - momentumTarget_.x() / Pcx;
1249	>	a000 = Kcy;
1250	>	a110 = Kcz;
1251	>	c0 = Kcx * c * c - Kcx - Kcy - Kcz;
1252	>	a001 = py * py * Khy;
1253	>	a111 = pz * pz * Khz;
1254	>	b01 = -2.0 * Khy * py * (1.0 + py);
1255	>	b11 = -2.0 * Khz * pz * (1.0 + pz);
1256	>	c1 = Khy * py * (2.0 + py) + Khz * pz * (2.0 + pz)
1257	>	+ Khx * (fastpow(c * px - px - 1.0, 2) - 1.0);
1258	>	break;
1259	>	case rnemdPy :
1260	>	c = 1 - momentumTarget_.y() / Pcy;
1261	>	a000 = Kcx;
1262	>	a110 = Kcz;
1263	>	c0 = Kcy * c * c - Kcx - Kcy - Kcz;
1264	>	a001 = px * px * Khx;
1265	>	a111 = pz * pz * Khz;
1266	>	b01 = -2.0 * Khx * px * (1.0 + px);
1267	>	b11 = -2.0 * Khz * pz * (1.0 + pz);
1268	>	c1 = Khx * px * (2.0 + px) + Khz * pz * (2.0 + pz)
1269	>	+ Khy * (fastpow(c * py - py - 1.0, 2) - 1.0);
1270	>	break;
1271	>	case rnemdPz ://we don't really do this, do we?
1272	>	c = 1 - momentumTarget_.z() / Pcz;
1273	>	a000 = Kcx;
1274	>	a110 = Kcy;
1275	>	c0 = Kcz * c * c - Kcx - Kcy - Kcz;
1276	>	a001 = px * px * Khx;
1277	>	a111 = py * py * Khy;
1278	>	b01 = -2.0 * Khx * px * (1.0 + px);
1279	>	b11 = -2.0 * Khy * py * (1.0 + py);
1280	>	c1 = Khx * px * (2.0 + px) + Khy * py * (2.0 + py)
1281	>	+ Khz * (fastpow(c * pz - pz - 1.0, 2) - 1.0);
1282	>	break;
1283	>	default :
1284	>	break;
1285	>	}
1286	>
1287	>	RealType v1 = a000 * a111 - a001 * a110;
1288	>	RealType v2 = a000 * b01;
1289	>	RealType v3 = a000 * b11;
1290	>	RealType v4 = a000 * c1 - a001 * c0;
1291	>	RealType v8 = a110 * b01;
1292	>	RealType v10 = - b01 * c0;
1293	>
1294	>	RealType u0 = v2 * v10 - v4 * v4;
1295	>	RealType u1 = -2.0 * v3 * v4;
1296	>	RealType u2 = -v2 * v8 - v3 * v3 - 2.0 * v1 * v4;
1297	>	RealType u3 = -2.0 * v1 * v3;
1298	>	RealType u4 = - v1 * v1;
1299	>	//rescale coefficients
1300	>	RealType maxAbs = fabs(u0);
1301	>	if (maxAbs < fabs(u1)) maxAbs = fabs(u1);
1302	>	if (maxAbs < fabs(u2)) maxAbs = fabs(u2);
1303	>	if (maxAbs < fabs(u3)) maxAbs = fabs(u3);
1304	>	if (maxAbs < fabs(u4)) maxAbs = fabs(u4);
1305	>	u0 /= maxAbs;
1306	>	u1 /= maxAbs;
1307	>	u2 /= maxAbs;
1308	>	u3 /= maxAbs;
1309	>	u4 /= maxAbs;
1310	>	//max_element(start, end) is also available.
1311	>	Polynomial<RealType> poly; //same as DoublePolynomial poly;
1312	>	poly.setCoefficient(4, u4);
1313	>	poly.setCoefficient(3, u3);
1314	>	poly.setCoefficient(2, u2);
1315	>	poly.setCoefficient(1, u1);
1316	>	poly.setCoefficient(0, u0);
1317	>	vector<RealType> realRoots = poly.FindRealRoots();
1318	>
1319	>	vector<RealType>::iterator ri;
1320	>	RealType r1, r2, alpha0;
1321	>	vector<pair<RealType,RealType> > rps;
1322	>	for (ri = realRoots.begin(); ri !=realRoots.end(); ++ri) {
1323	>	r2 = *ri;
1324	>	//check if FindRealRoots() give the right answer
1325	>	if ( fabs(u0 + r2 * (u1 + r2 * (u2 + r2 * (u3 + r2 * u4)))) > 1e-6 ) {
1326	>	sprintf(painCave.errMsg,
1327	>	"RNEMD Warning: polynomial solve seems to have an error!");
1328	>	painCave.isFatal = 0;
1329	>	simError();
1330	>	failRootCount_++;
1331	>	}
1332	>	//might not be useful w/o rescaling coefficients
1333	>	alpha0 = -c0 - a110 * r2 * r2;
1334	>	if (alpha0 >= 0.0) {
1335	>	r1 = sqrt(alpha0 / a000);
1336	>	if (fabs(c1 + r1 * (b01 + r1 * a001) + r2 * (b11 + r2 * a111))
1337	>	< 1e-6)
1338	>	{ rps.push_back(make_pair(r1, r2)); }
1339	>	if (r1 > 1e-6) { //r1 non-negative
1340	>	r1 = -r1;
1341	>	if (fabs(c1 + r1 * (b01 + r1 * a001) + r2 * (b11 + r2 * a111))
1342	>	< 1e-6)
1343	>	{ rps.push_back(make_pair(r1, r2)); }
1344	>	}
1345	>	}
1346	>	}
1347	>	// Consider combining together the solving pair part w/ the searching
1348	>	// best solution part so that we don't need the pairs vector
1349	>	if (!rps.empty()) {
1350	>	RealType smallestDiff = HONKING_LARGE_VALUE;
1351	>	RealType diff;
1352	>	pair<RealType,RealType> bestPair = make_pair(1.0, 1.0);
1353	>	vector<pair<RealType,RealType> >::iterator rpi;
1354	>	for (rpi = rps.begin(); rpi != rps.end(); ++rpi) {
1355	>	r1 = (*rpi).first;
1356	>	r2 = (*rpi).second;
1357	>	switch(rnemdFluxType_) {
1358	>	case rnemdKE :
1359	>	diff = fastpow(1.0 - r1, 2) + fastpow(1.0 - r2, 2)
1360	>	+ fastpow(r1 * r1 / r2 / r2 - Kcz/Kcx, 2)
1361	>	+ fastpow(r1 * r1 / r2 / r2 - Kcz/Kcy, 2);
1362	>	break;
1363	>	case rnemdPx :
1364	>	diff = fastpow(1.0 - r1, 2) + fastpow(1.0 - r2, 2)
1365	>	+ fastpow(r1 * r1 / r2 / r2 - Kcz/Kcy, 2);
1366	>	break;
1367	>	case rnemdPy :
1368	>	diff = fastpow(1.0 - r1, 2) + fastpow(1.0 - r2, 2)
1369	>	+ fastpow(r1 * r1 / r2 / r2 - Kcz/Kcx, 2);
1370	>	break;
1371	>	case rnemdPz :
1372	>	diff = fastpow(1.0 - r1, 2) + fastpow(1.0 - r2, 2)
1373	>	+ fastpow(r1 * r1 / r2 / r2 - Kcy/Kcx, 2);
1374	>	default :
1375	>	break;
1376	>	}
1377	>	if (diff < smallestDiff) {
1378	>	smallestDiff = diff;
1379	>	bestPair = *rpi;
1380	>	}
1381	>	}
1382	>	#ifdef IS_MPI
1383	>	if (worldRank == 0) {
1384	>	#endif
1385	>	// sprintf(painCave.errMsg,
1386	>	//         "RNEMD: roots r1= %lf\tr2 = %lf\n",
1387	>	//         bestPair.first, bestPair.second);
1388	>	// painCave.isFatal = 0;
1389	>	// painCave.severity = OPENMD_INFO;
1390	>	// simError();
1391	>	#ifdef IS_MPI
1392	>	}
1393	>	#endif
1394	>
1395	>	switch(rnemdFluxType_) {
1396	>	case rnemdKE :
1397	>	x = bestPair.first;
1398	>	y = bestPair.first;
1399	>	z = bestPair.second;
1400	>	break;
1401	>	case rnemdPx :
1402	>	x = c;
1403	>	y = bestPair.first;
1404	>	z = bestPair.second;
1405	>	break;
1406	>	case rnemdPy :
1407	>	x = bestPair.first;
1408	>	y = c;
1409	>	z = bestPair.second;
1410	>	break;
1411	>	case rnemdPz :
1412	>	x = bestPair.first;
1413	>	y = bestPair.second;
1414	>	z = c;
1415	>	break;
1416	>	default :
1417	>	break;
1418	>	}
1419	>	vector<StuntDouble*>::iterator sdi;
1420	>	Vector3d vel;
1421	>	for (sdi = coldBin.begin(); sdi != coldBin.end(); ++sdi) {
1422	>	vel = (*sdi)->getVel();
1423	>	vel.x() *= x;
1424	>	vel.y() *= y;
1425	>	vel.z() *= z;
1426	>	(*sdi)->setVel(vel);
1427	>	}
1428	>	//convert to hotBin coefficient
1429	>	x = 1.0 + px * (1.0 - x);
1430	>	y = 1.0 + py * (1.0 - y);
1431	>	z = 1.0 + pz * (1.0 - z);
1432	>	for (sdi = hotBin.begin(); sdi != hotBin.end(); ++sdi) {
1433	>	vel = (*sdi)->getVel();
1434	>	vel.x() *= x;
1435	>	vel.y() *= y;
1436	>	vel.z() *= z;
1437	>	(*sdi)->setVel(vel);
1438	>	}
1439	>	successfulScale = true;
1440	>	switch(rnemdFluxType_) {
1441	>	case rnemdKE :
1442	>	kineticExchange_ += kineticTarget_;
1443	>	break;
1444	>	case rnemdPx :
1445	>	case rnemdPy :
1446	>	case rnemdPz :
1447	>	momentumExchange_ += momentumTarget_;
1448	>	break;
1449	>	default :
1450	>	break;
1451	>	}
1452	>	}
1453	>	}
1454	>	if (successfulScale != true) {
1455	>	sprintf(painCave.errMsg,
1456	>	"RNEMD::doNIVS exchange NOT performed - roots that solve\n"
1457	>	"\tthe constraint equations may not exist or there may be\n"
1458	>	"\tno selected objects in one or both slabs.\n");
1459	>	painCave.isFatal = 0;
1460	>	painCave.severity = OPENMD_INFO;
1461	>	simError();
1462	>	failTrialCount_++;
1463	>	}
1464	>	}
1465	>
1466	>	void RNEMD::doVSS(SelectionManager& smanA, SelectionManager& smanB) {
1467	>	if (!doRNEMD_) return;
1468	>	int selei;
1469	>	int selej;
1470	>
1471	>	Snapshot* currentSnap_ = info_->getSnapshotManager()->getCurrentSnapshot();
1472	>	RealType time = currentSnap_->getTime();
1473	>	Mat3x3d hmat = currentSnap_->getHmat();
1474	>
1475	>	StuntDouble* sd;
1476	>
1477	>	vector<StuntDouble*> hotBin, coldBin;
1478	>
1479	>	Vector3d Ph(V3Zero);
1480	>	Vector3d Lh(V3Zero);
1481	>	RealType Mh = 0.0;
1482	>	Mat3x3d Ih(0.0);
1483	>	RealType Kh = 0.0;
1484	>	Vector3d Pc(V3Zero);
1485	>	Vector3d Lc(V3Zero);
1486	>	RealType Mc = 0.0;
1487	>	Mat3x3d Ic(0.0);
1488	>	RealType Kc = 0.0;
1489	>
1490	>	// Constraints can be on only the linear or angular momentum, but
1491	>	// not both.  Usually, the user will specify which they want, but
1492	>	// in case they don't, the use of periodic boundaries should make
1493	>	// the choice for us.
1494	>	bool doLinearPart = false;
1495	>	bool doAngularPart = false;
1496	>
1497	>	switch (rnemdFluxType_) {
1498	>	case rnemdPx:
1499	>	case rnemdPy:
1500	>	case rnemdPz:
1501	>	case rnemdPvector:
1502	>	case rnemdKePx:
1503	>	case rnemdKePy:
1504	>	case rnemdKePvector:
1505	>	doLinearPart = true;
1506		break;
1507	<	case rnemdPzScale ://we don't really do this, do we?
1508	<	c = 1 - targetFlux_ / Pcz;
1509	<	a000 = Kcx;
1510	<	a110 = Kcy;
1511	<	c0 = Kcz * c * c - Kcx - Kcy - Kcz;
1512	<	a001 = px * px * Khx;
1513	<	a111 = py * py * Khy;
1514	<	b01 = -2.0 * Khx * px * (1.0 + px);
1515	<	b11 = -2.0 * Khy * py * (1.0 + py);
690	<	c1 = Khx * px * (2.0 + px) + Khy * py * (2.0 + py)
691	<	+ Khz * (fastpow(c * pz - pz - 1.0, 2) - 1.0);
692	<	break;
693	<	default :
1507	>	case rnemdLx:
1508	>	case rnemdLy:
1509	>	case rnemdLz:
1510	>	case rnemdLvector:
1511	>	case rnemdKeLx:
1512	>	case rnemdKeLy:
1513	>	case rnemdKeLz:
1514	>	case rnemdKeLvector:
1515	>	doAngularPart = true;
1516		break;
1517	+	case rnemdKE:
1518	+	case rnemdRotKE:
1519	+	case rnemdFullKE:
1520	+	default:
1521	+	if (usePeriodicBoundaryConditions_)
1522	+	doLinearPart = true;
1523	+	else
1524	+	doAngularPart = true;
1525	+	break;
1526		}
1527	+
1528	+	for (sd = smanA.beginSelected(selei); sd != NULL;
1529	+	sd = smanA.nextSelected(selei)) {
1530
1531	<	RealType v1 = a000 * a111 - a001 * a110;
698	<	RealType v2 = a000 * b01;
699	<	RealType v3 = a000 * b11;
700	<	RealType v4 = a000 * c1 - a001 * c0;
701	<	RealType v8 = a110 * b01;
702	<	RealType v10 = - b01 * c0;
1531	>	Vector3d pos = sd->getPos();
1532
1533	<	RealType u0 = v2 * v10 - v4 * v4;
1534	<	RealType u1 = -2.0 * v3 * v4;
1535	<	RealType u2 = -v2 * v8 - v3 * v3 - 2.0 * v1 * v4;
1536	<	RealType u3 = -2.0 * v1 * v3;
1537	<	RealType u4 = - v1 * v1;
1538	<	//rescale coefficients
1539	<	RealType maxAbs = fabs(u0);
1540	<	if (maxAbs < fabs(u1)) maxAbs = fabs(u1);
1541	<	if (maxAbs < fabs(u2)) maxAbs = fabs(u2);
1542	<	if (maxAbs < fabs(u3)) maxAbs = fabs(u3);
1543	<	if (maxAbs < fabs(u4)) maxAbs = fabs(u4);
1544	<	u0 /= maxAbs;
1545	<	u1 /= maxAbs;
1546	<	u2 /= maxAbs;
1547	<	u3 /= maxAbs;
1548	<	u4 /= maxAbs;
1549	<	//max_element(start, end) is also available.
1550	<	Polynomial<RealType> poly; //same as DoublePolynomial poly;
1551	<	poly.setCoefficient(4, u4);
1552	<	poly.setCoefficient(3, u3);
1553	<	poly.setCoefficient(2, u2);
1554	<	poly.setCoefficient(1, u1);
1555	<	poly.setCoefficient(0, u0);
1556	<	std::vector<RealType> realRoots = poly.FindRealRoots();
1557	<
1558	<	std::vector<RealType>::iterator ri;
1559	<	RealType r1, r2, alpha0;
1560	<	std::vector<std::pair<RealType,RealType> > rps;
1561	<	for (ri = realRoots.begin(); ri !=realRoots.end(); ri++) {
1562	<	r2 = *ri;
1563	<	//check if FindRealRoots() give the right answer
1564	<	if ( fabs(u0 + r2 * (u1 + r2 * (u2 + r2 * (u3 + r2 * u4)))) > 1e-6 ) {
1565	<	sprintf(painCave.errMsg,
1566	<	"RNEMD Warning: polynomial solve seems to have an error!");
1567	<	painCave.isFatal = 0;
1568	<	simError();
740	<	failRootCount_++;
741	<	}
742	<	//might not be useful w/o rescaling coefficients
743	<	alpha0 = -c0 - a110 * r2 * r2;
744	<	if (alpha0 >= 0.0) {
745	<	r1 = sqrt(alpha0 / a000);
746	<	if (fabs(c1 + r1 * (b01 + r1 * a001) + r2 * (b11 + r2 * a111)) < 1e-6)
747	<	{ rps.push_back(std::make_pair(r1, r2)); }
748	<	if (r1 > 1e-6) { //r1 non-negative
749	<	r1 = -r1;
750	<	if (fabs(c1 + r1 * (b01 + r1 * a001) + r2 * (b11 + r2 * a111)) <1e-6)
751	<	{ rps.push_back(std::make_pair(r1, r2)); }
1533	>	// wrap the stuntdouble's position back into the box:
1534	>
1535	>	if (usePeriodicBoundaryConditions_)
1536	>	currentSnap_->wrapVector(pos);
1537	>
1538	>	RealType mass = sd->getMass();
1539	>	Vector3d vel = sd->getVel();
1540	>	Vector3d rPos = sd->getPos() - coordinateOrigin_;
1541	>	RealType r2;
1542	>
1543	>	hotBin.push_back(sd);
1544	>	Ph += mass * vel;
1545	>	Mh += mass;
1546	>	Kh += mass * vel.lengthSquare();
1547	>	Lh += mass * cross(rPos, vel);
1548	>	Ih -= outProduct(rPos, rPos) * mass;
1549	>	r2 = rPos.lengthSquare();
1550	>	Ih(0, 0) += mass * r2;
1551	>	Ih(1, 1) += mass * r2;
1552	>	Ih(2, 2) += mass * r2;
1553	>
1554	>	if (rnemdFluxType_ == rnemdFullKE) {
1555	>	if (sd->isDirectional()) {
1556	>	Vector3d angMom = sd->getJ();
1557	>	Mat3x3d I = sd->getI();
1558	>	if (sd->isLinear()) {
1559	>	int i = sd->linearAxis();
1560	>	int j = (i + 1) % 3;
1561	>	int k = (i + 2) % 3;
1562	>	Kh += angMom[j] * angMom[j] / I(j, j) +
1563	>	angMom[k] * angMom[k] / I(k, k);
1564	>	} else {
1565	>	Kh += angMom[0] * angMom[0] / I(0, 0) +
1566	>	angMom[1] * angMom[1] / I(1, 1) +
1567	>	angMom[2] * angMom[2] / I(2, 2);
1568	>	}
1569		}
1570		}
1571		}
1572	<	// Consider combininig together the solving pair part w/ the searching
1573	<	// best solution part so that we don't need the pairs vector
1574	<	if (!rps.empty()) {
1575	<	RealType smallestDiff = HONKING_LARGE_VALUE;
1576	<	RealType diff;
1577	<	std::pair<RealType,RealType> bestPair = std::make_pair(1.0, 1.0);
1578	<	std::vector<std::pair<RealType,RealType> >::iterator rpi;
1579	<	for (rpi = rps.begin(); rpi != rps.end(); rpi++) {
1580	<	r1 = (*rpi).first;
1581	<	r2 = (*rpi).second;
1582	<	switch(rnemdType_) {
1583	<	case rnemdKineticScale :
1584	<	diff = fastpow(1.0 - r1, 2) + fastpow(1.0 - r2, 2)
1585	<	+ fastpow(r1 * r1 / r2 / r2 - Kcz/Kcx, 2)
1586	<	+ fastpow(r1 * r1 / r2 / r2 - Kcz/Kcy, 2);
1587	<	break;
1588	<	case rnemdPxScale :
1589	<	diff = fastpow(1.0 - r1, 2) + fastpow(1.0 - r2, 2)
1590	<	+ fastpow(r1 * r1 / r2 / r2 - Kcz/Kcy, 2);
1591	<	break;
1592	<	case rnemdPyScale :
1593	<	diff = fastpow(1.0 - r1, 2) + fastpow(1.0 - r2, 2)
1594	<	+ fastpow(r1 * r1 / r2 / r2 - Kcz/Kcx, 2);
1595	<	break;
1596	<	case rnemdPzScale :
1597	<	default :
1598	<	break;
1572	>	for (sd = smanB.beginSelected(selej); sd != NULL;
1573	>	sd = smanB.nextSelected(selej)) {
1574	>
1575	>	Vector3d pos = sd->getPos();
1576	>
1577	>	// wrap the stuntdouble's position back into the box:
1578	>
1579	>	if (usePeriodicBoundaryConditions_)
1580	>	currentSnap_->wrapVector(pos);
1581	>
1582	>	RealType mass = sd->getMass();
1583	>	Vector3d vel = sd->getVel();
1584	>	Vector3d rPos = sd->getPos() - coordinateOrigin_;
1585	>	RealType r2;
1586	>
1587	>	coldBin.push_back(sd);
1588	>	Pc += mass * vel;
1589	>	Mc += mass;
1590	>	Kc += mass * vel.lengthSquare();
1591	>	Lc += mass * cross(rPos, vel);
1592	>	Ic -= outProduct(rPos, rPos) * mass;
1593	>	r2 = rPos.lengthSquare();
1594	>	Ic(0, 0) += mass * r2;
1595	>	Ic(1, 1) += mass * r2;
1596	>	Ic(2, 2) += mass * r2;
1597	>
1598	>	if (rnemdFluxType_ == rnemdFullKE) {
1599	>	if (sd->isDirectional()) {
1600	>	Vector3d angMom = sd->getJ();
1601	>	Mat3x3d I = sd->getI();
1602	>	if (sd->isLinear()) {
1603	>	int i = sd->linearAxis();
1604	>	int j = (i + 1) % 3;
1605	>	int k = (i + 2) % 3;
1606	>	Kc += angMom[j] * angMom[j] / I(j, j) +
1607	>	angMom[k] * angMom[k] / I(k, k);
1608	>	} else {
1609	>	Kc += angMom[0] * angMom[0] / I(0, 0) +
1610	>	angMom[1] * angMom[1] / I(1, 1) +
1611	>	angMom[2] * angMom[2] / I(2, 2);
1612	>	}
1613		}
783	–	if (diff < smallestDiff) {
784	–	smallestDiff = diff;
785	–	bestPair = *rpi;
786	–	}
1614		}
1615	+	}
1616	+
1617	+	Kh *= 0.5;
1618	+	Kc *= 0.5;
1619	+
1620		#ifdef IS_MPI
1621	<	if (worldRank == 0) {
1621	>	MPI_Allreduce(MPI_IN_PLACE, &Ph[0], 3, MPI_REALTYPE, MPI_SUM,
1622	>	MPI_COMM_WORLD);
1623	>	MPI_Allreduce(MPI_IN_PLACE, &Pc[0], 3, MPI_REALTYPE, MPI_SUM,
1624	>	MPI_COMM_WORLD);
1625	>	MPI_Allreduce(MPI_IN_PLACE, &Lh[0], 3, MPI_REALTYPE, MPI_SUM,
1626	>	MPI_COMM_WORLD);
1627	>	MPI_Allreduce(MPI_IN_PLACE, &Lc[0], 3, MPI_REALTYPE, MPI_SUM,
1628	>	MPI_COMM_WORLD);
1629	>	MPI_Allreduce(MPI_IN_PLACE, &Mh, 1, MPI_REALTYPE, MPI_SUM, MPI_COMM_WORLD);
1630	>	MPI_Allreduce(MPI_IN_PLACE, &Kh, 1, MPI_REALTYPE, MPI_SUM, MPI_COMM_WORLD);
1631	>	MPI_Allreduce(MPI_IN_PLACE, &Mc, 1, MPI_REALTYPE, MPI_SUM, MPI_COMM_WORLD);
1632	>	MPI_Allreduce(MPI_IN_PLACE, &Kc, 1, MPI_REALTYPE, MPI_SUM, MPI_COMM_WORLD);
1633	>	MPI_Allreduce(MPI_IN_PLACE, Ih.getArrayPointer(), 9,
1634	>	MPI_REALTYPE, MPI_SUM, MPI_COMM_WORLD);
1635	>	MPI_Allreduce(MPI_IN_PLACE, Ic.getArrayPointer(), 9,
1636	>	MPI_REALTYPE, MPI_SUM, MPI_COMM_WORLD);
1637		#endif
1638	<	std::cerr << "we choose r1 = " << bestPair.first
792	<	<< " and r2 = " << bestPair.second << "\n";
793	<	#ifdef IS_MPI
794	<	}
795	<	#endif
1638	>
1639
1640	<	RealType x, y, z;
1641	<	switch(rnemdType_) {
1642	<	case rnemdKineticScale :
1643	<	x = bestPair.first;
1644	<	y = bestPair.first;
1645	<	z = bestPair.second;
1646	<	break;
1647	<	case rnemdPxScale :
1648	<	x = c;
1649	<	y = bestPair.first;
1650	<	z = bestPair.second;
1651	<	break;
1652	<	case rnemdPyScale :
1653	<	x = bestPair.first;
1654	<	y = c;
1655	<	z = bestPair.second;
1656	<	break;
1657	<	case rnemdPzScale :
1658	<	x = bestPair.first;
1659	<	y = bestPair.second;
1660	<	z = c;
1661	<	break;
1662	<	default :
1663	<	break;
1664	<	}
1665	<	std::vector<StuntDouble*>::iterator sdi;
1666	<	Vector3d vel;
1667	<	for (sdi = coldBin.begin(); sdi != coldBin.end(); sdi++) {
1668	<	vel = (*sdi)->getVel();
1669	<	vel.x() *= x;
1670	<	vel.y() *= y;
1671	<	vel.z() *= z;
1672	<	(*sdi)->setVel(vel);
1673	<	}
1674	<	//convert to hotBin coefficient
1675	<	x = 1.0 + px * (1.0 - x);
1676	<	y = 1.0 + py * (1.0 - y);
1677	<	z = 1.0 + pz * (1.0 - z);
1678	<	for (sdi = hotBin.begin(); sdi != hotBin.end(); sdi++) {
1679	<	vel = (*sdi)->getVel();
1680	<	vel.x() *= x;
1681	<	vel.y() *= y;
1682	<	vel.z() *= z;
1683	<	(*sdi)->setVel(vel);
1640	>	Vector3d ac, acrec, bc, bcrec;
1641	>	Vector3d ah, ahrec, bh, bhrec;
1642	>
1643	>	bool successfulExchange = false;
1644	>	if ((Mh > 0.0) && (Mc > 0.0)) {//both slabs are not empty
1645	>	Vector3d vc = Pc / Mc;
1646	>	ac = -momentumTarget_ / Mc + vc;
1647	>	acrec = -momentumTarget_ / Mc;
1648	>
1649	>	// We now need the inverse of the inertia tensor to calculate the
1650	>	// angular velocity of the cold slab;
1651	>	Mat3x3d Ici = Ic.inverse();
1652	>	Vector3d omegac = Ici * Lc;
1653	>	bc  = -(Ici * angularMomentumTarget_) + omegac;
1654	>	bcrec = bc - omegac;
1655	>
1656	>	RealType cNumerator = Kc - kineticTarget_;
1657	>	if (doLinearPart)
1658	>	cNumerator -= 0.5 * Mc * ac.lengthSquare();
1659	>
1660	>	if (doAngularPart)
1661	>	cNumerator -= 0.5 * ( dot(bc, Ic * bc));
1662	>
1663	>	if (cNumerator > 0.0) {
1664	>
1665	>	RealType cDenominator = Kc;
1666	>
1667	>	if (doLinearPart)
1668	>	cDenominator -= 0.5 * Mc * vc.lengthSquare();
1669	>
1670	>	if (doAngularPart)
1671	>	cDenominator -= 0.5*(dot(omegac, Ic * omegac));
1672	>
1673	>	if (cDenominator > 0.0) {
1674	>	RealType c = sqrt(cNumerator / cDenominator);
1675	>	if ((c > 0.9) && (c < 1.1)) {//restrict scaling coefficients
1676	>
1677	>	Vector3d vh = Ph / Mh;
1678	>	ah = momentumTarget_ / Mh + vh;
1679	>	ahrec = momentumTarget_ / Mh;
1680	>
1681	>	// We now need the inverse of the inertia tensor to
1682	>	// calculate the angular velocity of the hot slab;
1683	>	Mat3x3d Ihi = Ih.inverse();
1684	>	Vector3d omegah = Ihi * Lh;
1685	>	bh  = (Ihi * angularMomentumTarget_) + omegah;
1686	>	bhrec = bh - omegah;
1687	>
1688	>	RealType hNumerator = Kh + kineticTarget_;
1689	>	if (doLinearPart)
1690	>	hNumerator -= 0.5 * Mh * ah.lengthSquare();
1691	>
1692	>	if (doAngularPart)
1693	>	hNumerator -= 0.5 * ( dot(bh, Ih * bh));
1694	>
1695	>	if (hNumerator > 0.0) {
1696	>
1697	>	RealType hDenominator = Kh;
1698	>	if (doLinearPart)
1699	>	hDenominator -= 0.5 * Mh * vh.lengthSquare();
1700	>	if (doAngularPart)
1701	>	hDenominator -= 0.5*(dot(omegah, Ih * omegah));
1702	>
1703	>	if (hDenominator > 0.0) {
1704	>	RealType h = sqrt(hNumerator / hDenominator);
1705	>	if ((h > 0.9) && (h < 1.1)) {
1706	>
1707	>	vector<StuntDouble*>::iterator sdi;
1708	>	Vector3d vel;
1709	>	Vector3d rPos;
1710	>
1711	>	for (sdi = coldBin.begin(); sdi != coldBin.end(); ++sdi) {
1712	>	//vel = (*sdi)->getVel();
1713	>	rPos = (*sdi)->getPos() - coordinateOrigin_;
1714	>	if (doLinearPart)
1715	>	vel = ((*sdi)->getVel() - vc) * c + ac;
1716	>	if (doAngularPart)
1717	>	vel = ((*sdi)->getVel() - cross(omegac, rPos)) * c + cross(bc, rPos);
1718	>
1719	>	(*sdi)->setVel(vel);
1720	>	if (rnemdFluxType_ == rnemdFullKE) {
1721	>	if ((*sdi)->isDirectional()) {
1722	>	Vector3d angMom = (*sdi)->getJ() * c;
1723	>	(*sdi)->setJ(angMom);
1724	>	}
1725	>	}
1726	>	}
1727	>	for (sdi = hotBin.begin(); sdi != hotBin.end(); ++sdi) {
1728	>	//vel = (*sdi)->getVel();
1729	>	rPos = (*sdi)->getPos() - coordinateOrigin_;
1730	>	if (doLinearPart)
1731	>	vel = ((*sdi)->getVel() - vh) * h + ah;
1732	>	if (doAngularPart)
1733	>	vel = ((*sdi)->getVel() - cross(omegah, rPos)) * h + cross(bh, rPos);
1734	>
1735	>	(*sdi)->setVel(vel);
1736	>	if (rnemdFluxType_ == rnemdFullKE) {
1737	>	if ((*sdi)->isDirectional()) {
1738	>	Vector3d angMom = (*sdi)->getJ() * h;
1739	>	(*sdi)->setJ(angMom);
1740	>	}
1741	>	}
1742	>	}
1743	>	successfulExchange = true;
1744	>	kineticExchange_ += kineticTarget_;
1745	>	momentumExchange_ += momentumTarget_;
1746	>	angularMomentumExchange_ += angularMomentumTarget_;
1747	>	}
1748	>	}
1749	>	}
1750	>	}
1751	>	}
1752		}
1753	<	exchangeSum_ += targetFlux_;
1754	<	//we may want to check whether the exchange has been successful
1755	<	} else {
1756	<	std::cerr << "exchange NOT performed!\n";//MPI incompatible
1753	>	}
1754	>	if (successfulExchange != true) {
1755	>	sprintf(painCave.errMsg,
1756	>	"RNEMD::doVSS exchange NOT performed - roots that solve\n"
1757	>	"\tthe constraint equations may not exist or there may be\n"
1758	>	"\tno selected objects in one or both slabs.\n");
1759	>	painCave.isFatal = 0;
1760	>	painCave.severity = OPENMD_INFO;
1761	>	simError();
1762		failTrialCount_++;
1763		}
848	–
1764		}
1765
1766	+	RealType RNEMD::getDividingArea() {
1767	+
1768	+	if (hasDividingArea_) return dividingArea_;
1769	+
1770	+	RealType areaA, areaB;
1771	+	Snapshot* snap = info_->getSnapshotManager()->getCurrentSnapshot();
1772	+
1773	+	if (hasSelectionA_) {
1774	+
1775	+	if (evaluatorA_.hasSurfaceArea())
1776	+	areaA = evaluatorA_.getSurfaceArea();
1777	+	else {
1778	+
1779	+	cerr << "selection A did not have surface area, recomputing\n";
1780	+	int isd;
1781	+	StuntDouble* sd;
1782	+	vector<StuntDouble*> aSites;
1783	+	seleManA_.setSelectionSet(evaluatorA_.evaluate());
1784	+	for (sd = seleManA_.beginSelected(isd); sd != NULL;
1785	+	sd = seleManA_.nextSelected(isd)) {
1786	+	aSites.push_back(sd);
1787	+	}
1788	+	#if defined(HAVE_QHULL)
1789	+	ConvexHull* surfaceMeshA = new ConvexHull();
1790	+	surfaceMeshA->computeHull(aSites);
1791	+	cerr << "flag1\n";
1792	+	areaA = surfaceMeshA->getArea();
1793	+	cerr << "Flag2 " << areaA << "\n";
1794	+	delete surfaceMeshA;
1795	+	#else
1796	+	sprintf( painCave.errMsg,
1797	+	"RNEMD::getDividingArea : Hull calculation is not possible\n"
1798	+	"\twithout libqhull. Please rebuild OpenMD with qhull enabled.");
1799	+	painCave.severity = OPENMD_ERROR;
1800	+	painCave.isFatal = 1;
1801	+	simError();
1802	+	#endif
1803	+	}
1804	+
1805	+	} else {
1806	+	if (usePeriodicBoundaryConditions_) {
1807	+	// in periodic boundaries, the surface area is twice the x-y
1808	+	// area of the current box:
1809	+	areaA = 2.0 * snap->getXYarea();
1810	+	} else {
1811	+	// in non-periodic simulations, without explicitly setting
1812	+	// selections, the sphere radius sets the surface area of the
1813	+	// dividing surface:
1814	+	areaA = 4.0 * M_PI * pow(sphereARadius_, 2);
1815	+	}
1816	+	}
1817	+
1818	+	if (hasSelectionB_) {
1819	+	if (evaluatorB_.hasSurfaceArea())
1820	+	areaB = evaluatorB_.getSurfaceArea();
1821	+	else {
1822	+	cerr << "selection B did not have surface area, recomputing\n";
1823	+
1824	+	int isd;
1825	+	StuntDouble* sd;
1826	+	vector<StuntDouble*> bSites;
1827	+	seleManB_.setSelectionSet(evaluatorB_.evaluate());
1828	+	for (sd = seleManB_.beginSelected(isd); sd != NULL;
1829	+	sd = seleManB_.nextSelected(isd)) {
1830	+	bSites.push_back(sd);
1831	+	}
1832	+
1833	+	#if defined(HAVE_QHULL)
1834	+	ConvexHull* surfaceMeshB = new ConvexHull();
1835	+	surfaceMeshB->computeHull(bSites);
1836	+	areaB = surfaceMeshB->getArea();
1837	+	delete surfaceMeshB;
1838	+	#else
1839	+	sprintf( painCave.errMsg,
1840	+	"RNEMD::getDividingArea : Hull calculation is not possible\n"
1841	+	"\twithout libqhull. Please rebuild OpenMD with qhull enabled.");
1842	+	painCave.severity = OPENMD_ERROR;
1843	+	painCave.isFatal = 1;
1844	+	simError();
1845	+	#endif
1846	+	}
1847	+
1848	+	} else {
1849	+	if (usePeriodicBoundaryConditions_) {
1850	+	// in periodic boundaries, the surface area is twice the x-y
1851	+	// area of the current box:
1852	+	areaB = 2.0 * snap->getXYarea();
1853	+	} else {
1854	+	// in non-periodic simulations, without explicitly setting
1855	+	// selections, but if a sphereBradius has been set, just use that:
1856	+	areaB = 4.0 * M_PI * pow(sphereBRadius_, 2);
1857	+	}
1858	+	}
1859	+
1860	+	dividingArea_ = min(areaA, areaB);
1861	+	hasDividingArea_ = true;
1862	+	return dividingArea_;
1863	+	}
1864	+
1865		void RNEMD::doRNEMD() {
1866	+	if (!doRNEMD_) return;
1867	+	trialCount_++;
1868
1869	<	switch(rnemdType_) {
1870	<	case rnemdKineticScale :
1871	<	case rnemdPxScale :
1872	<	case rnemdPyScale :
1873	<	case rnemdPzScale :
1874	<	doScale();
1869	>	// object evaluator:
1870	>	evaluator_.loadScriptString(rnemdObjectSelection_);
1871	>	seleMan_.setSelectionSet(evaluator_.evaluate());
1872	>
1873	>	evaluatorA_.loadScriptString(selectionA_);
1874	>	evaluatorB_.loadScriptString(selectionB_);
1875	>
1876	>	seleManA_.setSelectionSet(evaluatorA_.evaluate());
1877	>	seleManB_.setSelectionSet(evaluatorB_.evaluate());
1878	>
1879	>	commonA_ = seleManA_ & seleMan_;
1880	>	commonB_ = seleManB_ & seleMan_;
1881	>
1882	>	// Target exchange quantities (in each exchange) = dividingArea * dt * flux
1883	>	// dt = exchange time interval
1884	>	// flux = target flux
1885	>	// dividingArea = smallest dividing surface between the two regions
1886	>
1887	>	hasDividingArea_ = false;
1888	>	RealType area = getDividingArea();
1889	>
1890	>	kineticTarget_ = kineticFlux_ * exchangeTime_ * area;
1891	>	momentumTarget_ = momentumFluxVector_ * exchangeTime_ * area;
1892	>	angularMomentumTarget_ = angularMomentumFluxVector_ * exchangeTime_ * area;
1893	>
1894	>	switch(rnemdMethod_) {
1895	>	case rnemdSwap:
1896	>	doSwap(commonA_, commonB_);
1897		break;
1898	<	case rnemdKineticSwap :
1899	<	case rnemdPx :
862	<	case rnemdPy :
863	<	case rnemdPz :
864	<	doSwap();
1898	>	case rnemdNIVS:
1899	>	doNIVS(commonA_, commonB_);
1900		break;
1901	<	case rnemdUnknown :
1901	>	case rnemdVSS:
1902	>	doVSS(commonA_, commonB_);
1903	>	break;
1904	>	case rnemdUnkownMethod:
1905		default :
1906		break;
1907		}
1908		}
1909
1910		void RNEMD::collectData() {
1911	<
1911	>	if (!doRNEMD_) return;
1912		Snapshot* currentSnap_ = info_->getSnapshotManager()->getCurrentSnapshot();
1913	+
1914	+	// collectData can be called more frequently than the doRNEMD, so use the
1915	+	// computed area from the last exchange time:
1916	+	RealType area = getDividingArea();
1917	+	areaAccumulator_->add(area);
1918		Mat3x3d hmat = currentSnap_->getHmat();
1919	+	Vector3d u = angularMomentumFluxVector_;
1920	+	u.normalize();
1921
1922		seleMan_.setSelectionSet(evaluator_.evaluate());
1923
1924	<	int selei;
1924	>	int selei(0);
1925		StuntDouble* sd;
1926	<	int idx;
1926	>	int binNo;
1927	>	RealType mass;
1928	>	Vector3d vel;
1929	>	Vector3d rPos;
1930	>	RealType KE;
1931	>	Vector3d L;
1932	>	Mat3x3d I;
1933	>	RealType r2;
1934
1935	+	vector<RealType> binMass(nBins_, 0.0);
1936	+	vector<Vector3d> binP(nBins_, V3Zero);
1937	+	vector<RealType> binOmega(nBins_, 0.0);
1938	+	vector<Vector3d> binL(nBins_, V3Zero);
1939	+	vector<Mat3x3d>  binI(nBins_);
1940	+	vector<RealType> binKE(nBins_, 0.0);
1941	+	vector<int> binDOF(nBins_, 0);
1942	+	vector<int> binCount(nBins_, 0);
1943	+
1944	+	// alternative approach, track all molecules instead of only those
1945	+	// selected for scaling/swapping:
1946	+	/*
1947	+	SimInfo::MoleculeIterator miter;
1948	+	vector<StuntDouble*>::iterator iiter;
1949	+	Molecule* mol;
1950	+	StuntDouble* sd;
1951	+	for (mol = info_->beginMolecule(miter); mol != NULL;
1952	+	mol = info_->nextMolecule(miter))
1953	+	sd is essentially sd
1954	+	for (sd = mol->beginIntegrableObject(iiter);
1955	+	sd != NULL;
1956	+	sd = mol->nextIntegrableObject(iiter))
1957	+	*/
1958	+
1959		for (sd = seleMan_.beginSelected(selei); sd != NULL;
1960	<	sd = seleMan_.nextSelected(selei)) {
1961	<
886	<	idx = sd->getLocalIndex();
887	<
1960	>	sd = seleMan_.nextSelected(selei)) {
1961	>
1962		Vector3d pos = sd->getPos();
1963
1964		// wrap the stuntdouble's position back into the box:
1965
1966	<	if (usePeriodicBoundaryConditions_)
1966	>	if (usePeriodicBoundaryConditions_) {
1967		currentSnap_->wrapVector(pos);
1968	<
1969	<	// which bin is this stuntdouble in?
1970	<	// wrapped positions are in the range [-0.5*hmat(2,2), +0.5*hmat(2,2)]
1971	<
1972	<	int binNo = int(nBins_ * (pos.z() / hmat(2,2) + 0.5)) % nBins_;
1968	>	// which bin is this stuntdouble in?
1969	>	// wrapped positions are in the range [-0.5*hmat(2,2), +0.5*hmat(2,2)]
1970	>	// Shift molecules by half a box to have bins start at 0
1971	>	// The modulo operator is used to wrap the case when we are
1972	>	// beyond the end of the bins back to the beginning.
1973	>	binNo = int(nBins_ * (pos.z() / hmat(2,2) + 0.5)) % nBins_;
1974	>	} else {
1975	>	Vector3d rPos = pos - coordinateOrigin_;
1976	>	binNo = int(rPos.length() / binWidth_);
1977	>	}
1978
1979	<	if (rnemdLogWidth_ == midBin_ + 1)
1980	<	if (binNo > midBin_)
1981	<	binNo = nBins_ - binNo;
1979	>	mass = sd->getMass();
1980	>	vel = sd->getVel();
1981	>	rPos = sd->getPos() - coordinateOrigin_;
1982	>	KE = 0.5 * mass * vel.lengthSquare();
1983	>	L = mass * cross(rPos, vel);
1984	>	I = outProduct(rPos, rPos) * mass;
1985	>	r2 = rPos.lengthSquare();
1986	>	I(0, 0) += mass * r2;
1987	>	I(1, 1) += mass * r2;
1988	>	I(2, 2) += mass * r2;
1989
1990	<	RealType mass = sd->getMass();
1991	<	Vector3d vel = sd->getVel();
1992	<	RealType value;
1993	<	RealType xVal, yVal, zVal;
1990	>	// Project the relative position onto a plane perpendicular to
1991	>	// the angularMomentumFluxVector:
1992	>	// Vector3d rProj = rPos - dot(rPos, u) * u;
1993	>	// Project the velocity onto a plane perpendicular to the
1994	>	// angularMomentumFluxVector:
1995	>	// Vector3d vProj = vel  - dot(vel, u) * u;
1996	>	// Compute angular velocity vector (should be nearly parallel to
1997	>	// angularMomentumFluxVector
1998	>	// Vector3d aVel = cross(rProj, vProj);
1999
2000	<	switch(rnemdType_) {
2001	<	case rnemdKineticSwap :
2002	<	case rnemdKineticScale :
2003	<
2004	<	value = mass * (vel[0]*vel[0] + vel[1]*vel[1] +
2005	<	vel[2]*vel[2]);
2006	<
2007	<	valueCount_[binNo] += 3;
2008	<	if (sd->isDirectional()) {
2009	<	Vector3d angMom = sd->getJ();
2010	<	Mat3x3d I = sd->getI();
2011	<
2012	<	if (sd->isLinear()) {
2013	<	int i = sd->linearAxis();
2014	<	int j = (i + 1) % 3;
2015	<	int k = (i + 2) % 3;
2016	<	value += angMom[j] * angMom[j] / I(j, j) +
2017	<	angMom[k] * angMom[k] / I(k, k);
2018	<
2019	<	valueCount_[binNo] +=2;
2020	<
2021	<	} else {
2022	<	value += angMom[0]*angMom[0]/I(0, 0)
2023	<	+ angMom[1]*angMom[1]/I(1, 1)
2024	<	+ angMom[2]*angMom[2]/I(2, 2);
2025	<	valueCount_[binNo] +=3;
935	<	}
936	<	}
937	<	value = value / PhysicalConstants::energyConvert / PhysicalConstants::kb;
938	<
939	<	break;
940	<	case rnemdPx :
941	<	case rnemdPxScale :
942	<	value = mass * vel[0];
943	<	valueCount_[binNo]++;
944	<	xVal = mass * vel.x() * vel.x() / PhysicalConstants::energyConvert
945	<	/ PhysicalConstants::kb;
946	<	yVal = mass * vel.y() * vel.y() / PhysicalConstants::energyConvert
947	<	/ PhysicalConstants::kb;
948	<	zVal = mass * vel.z() * vel.z() / PhysicalConstants::energyConvert
949	<	/ PhysicalConstants::kb;
950	<	xTempHist_[binNo] += xVal;
951	<	yTempHist_[binNo] += yVal;
952	<	zTempHist_[binNo] += zVal;
953	<	break;
954	<	case rnemdPy :
955	<	case rnemdPyScale :
956	<	value = mass * vel[1];
957	<	valueCount_[binNo]++;
958	<	break;
959	<	case rnemdPz :
960	<	case rnemdPzScale :
961	<	value = mass * vel[2];
962	<	valueCount_[binNo]++;
963	<	break;
964	<	case rnemdUnknown :
965	<	default :
966	<	break;
2000	>	if (binNo >= 0 && binNo < nBins_)  {
2001	>	binCount[binNo]++;
2002	>	binMass[binNo] += mass;
2003	>	binP[binNo] += mass*vel;
2004	>	binKE[binNo] += KE;
2005	>	binI[binNo] += I;
2006	>	binL[binNo] += L;
2007	>	binDOF[binNo] += 3;
2008	>
2009	>	if (sd->isDirectional()) {
2010	>	Vector3d angMom = sd->getJ();
2011	>	Mat3x3d Ia = sd->getI();
2012	>	if (sd->isLinear()) {
2013	>	int i = sd->linearAxis();
2014	>	int j = (i + 1) % 3;
2015	>	int k = (i + 2) % 3;
2016	>	binKE[binNo] += 0.5 * (angMom[j] * angMom[j] / Ia(j, j) +
2017	>	angMom[k] * angMom[k] / Ia(k, k));
2018	>	binDOF[binNo] += 2;
2019	>	} else {
2020	>	binKE[binNo] += 0.5 * (angMom[0] * angMom[0] / Ia(0, 0) +
2021	>	angMom[1] * angMom[1] / Ia(1, 1) +
2022	>	angMom[2] * angMom[2] / Ia(2, 2));
2023	>	binDOF[binNo] += 3;
2024	>	}
2025	>	}
2026		}
2027	<	valueHist_[binNo] += value;
2027	>	}
2028	>
2029	>	#ifdef IS_MPI
2030	>
2031	>	for (int i = 0; i < nBins_; i++) {
2032	>
2033	>	MPI_Allreduce(MPI_IN_PLACE, &binCount[i],
2034	>	1, MPI_INT, MPI_SUM, MPI_COMM_WORLD);
2035	>	MPI_Allreduce(MPI_IN_PLACE, &binMass[i],
2036	>	1, MPI_REALTYPE, MPI_SUM, MPI_COMM_WORLD);
2037	>	MPI_Allreduce(MPI_IN_PLACE, binP[i].getArrayPointer(),
2038	>	3, MPI_REALTYPE, MPI_SUM, MPI_COMM_WORLD);
2039	>	MPI_Allreduce(MPI_IN_PLACE, binL[i].getArrayPointer(),
2040	>	3, MPI_REALTYPE, MPI_SUM, MPI_COMM_WORLD);
2041	>	MPI_Allreduce(MPI_IN_PLACE, binI[i].getArrayPointer(),
2042	>	9, MPI_REALTYPE, MPI_SUM, MPI_COMM_WORLD);
2043	>	MPI_Allreduce(MPI_IN_PLACE, &binKE[i],
2044	>	1, MPI_REALTYPE, MPI_SUM, MPI_COMM_WORLD);
2045	>	MPI_Allreduce(MPI_IN_PLACE, &binDOF[i],
2046	>	1, MPI_INT, MPI_SUM, MPI_COMM_WORLD);
2047	>	//MPI_Allreduce(MPI_IN_PLACE, &binOmega[i],
2048	>	//                          1, MPI_REALTYPE, MPI_SUM, MPI_COMM_WORLD);
2049		}
2050	+
2051	+	#endif
2052
2053	+	Vector3d omega;
2054	+	RealType den;
2055	+	RealType temp;
2056	+	RealType z;
2057	+	RealType r;
2058	+	for (int i = 0; i < nBins_; i++) {
2059	+	if (usePeriodicBoundaryConditions_) {
2060	+	z = (((RealType)i + 0.5) / (RealType)nBins_) * hmat(2,2);
2061	+	den = binMass[i] * nBins_ * PhysicalConstants::densityConvert
2062	+	/ currentSnap_->getVolume() ;
2063	+	} else {
2064	+	r = (((RealType)i + 0.5) * binWidth_);
2065	+	RealType rinner = (RealType)i * binWidth_;
2066	+	RealType router = (RealType)(i+1) * binWidth_;
2067	+	den = binMass[i] * 3.0 * PhysicalConstants::densityConvert
2068	+	/ (4.0 * M_PI * (pow(router,3) - pow(rinner,3)));
2069	+	}
2070	+	vel = binP[i] / binMass[i];
2071	+
2072	+	omega = binI[i].inverse() * binL[i];
2073	+
2074	+	// omega = binOmega[i] / binCount[i];
2075	+
2076	+	if (binCount[i] > 0) {
2077	+	// only add values if there are things to add
2078	+	temp = 2.0 * binKE[i] / (binDOF[i] * PhysicalConstants::kb *
2079	+	PhysicalConstants::energyConvert);
2080	+
2081	+	for (unsigned int j = 0; j < outputMask_.size(); ++j) {
2082	+	if(outputMask_[j]) {
2083	+	switch(j) {
2084	+	case Z:
2085	+	dynamic_cast<Accumulator *>(data_[j].accumulator[i])->add(z);
2086	+	break;
2087	+	case R:
2088	+	dynamic_cast<Accumulator *>(data_[j].accumulator[i])->add(r);
2089	+	break;
2090	+	case TEMPERATURE:
2091	+	dynamic_cast<Accumulator *>(data_[j].accumulator[i])->add(temp);
2092	+	break;
2093	+	case VELOCITY:
2094	+	dynamic_cast<VectorAccumulator *>(data_[j].accumulator[i])->add(vel);
2095	+	break;
2096	+	case ANGULARVELOCITY:
2097	+	dynamic_cast<VectorAccumulator *>(data_[j].accumulator[i])->add(omega);
2098	+	break;
2099	+	case DENSITY:
2100	+	dynamic_cast<Accumulator *>(data_[j].accumulator[i])->add(den);
2101	+	break;
2102	+	}
2103	+	}
2104	+	}
2105	+	}
2106	+	}
2107	+	hasData_ = true;
2108		}
2109
2110		void RNEMD::getStarted() {
2111	<	Snapshot* currentSnap_ = info_->getSnapshotManager()->getCurrentSnapshot();
2112	<	Stats& stat = currentSnap_->statData;
2113	<	stat[Stats::RNEMD_EXCHANGE_TOTAL] = exchangeSum_;
2111	>	if (!doRNEMD_) return;
2112	>	hasDividingArea_ = false;
2113	>	collectData();
2114	>	writeOutputFile();
2115		}
2116
2117	<	void RNEMD::getStatus() {
2117	>	void RNEMD::parseOutputFileFormat(const std::string& format) {
2118	>	if (!doRNEMD_) return;
2119	>	StringTokenizer tokenizer(format, " ,;|\t\n\r");
2120	>
2121	>	while(tokenizer.hasMoreTokens()) {
2122	>	std::string token(tokenizer.nextToken());
2123	>	toUpper(token);
2124	>	OutputMapType::iterator i = outputMap_.find(token);
2125	>	if (i != outputMap_.end()) {
2126	>	outputMask_.set(i->second);
2127	>	} else {
2128	>	sprintf( painCave.errMsg,
2129	>	"RNEMD::parseOutputFileFormat: %s is not a recognized\n"
2130	>	"\toutputFileFormat keyword.\n", token.c_str() );
2131	>	painCave.isFatal = 0;
2132	>	painCave.severity = OPENMD_ERROR;
2133	>	simError();
2134	>	}
2135	>	}
2136	>	}
2137	>
2138	>	void RNEMD::writeOutputFile() {
2139	>	if (!doRNEMD_) return;
2140	>	if (!hasData_) return;
2141	>
2142	>	#ifdef IS_MPI
2143	>	// If we're the root node, should we print out the results
2144	>	int worldRank;
2145	>	MPI_Comm_rank( MPI_COMM_WORLD, &worldRank);
2146
2147	<	Snapshot* currentSnap_ = info_->getSnapshotManager()->getCurrentSnapshot();
2148	<	Stats& stat = currentSnap_->statData;
2149	<	RealType time = currentSnap_->getTime();
2147	>	if (worldRank == 0) {
2148	>	#endif
2149	>	rnemdFile_.open(rnemdFileName_.c_str(), std::ios::out | std::ios::trunc );
2150	>
2151	>	if( !rnemdFile_ ){
2152	>	sprintf( painCave.errMsg,
2153	>	"Could not open \"%s\" for RNEMD output.\n",
2154	>	rnemdFileName_.c_str());
2155	>	painCave.isFatal = 1;
2156	>	simError();
2157	>	}
2158
2159	<	stat[Stats::RNEMD_EXCHANGE_TOTAL] = exchangeSum_;
986	<	//or to be more meaningful, define another item as exchangeSum_ / time
987	<	int j;
2159	>	Snapshot* currentSnap_ = info_->getSnapshotManager()->getCurrentSnapshot();
2160
2161	<	#ifdef IS_MPI
2161	>	RealType time = currentSnap_->getTime();
2162	>	RealType avgArea;
2163	>	areaAccumulator_->getAverage(avgArea);
2164
2165	<	// all processors have the same number of bins, and STL vectors pack their
2166	<	// arrays, so in theory, this should be safe:
2165	>	RealType Jz(0.0);
2166	>	Vector3d JzP(V3Zero);
2167	>	Vector3d JzL(V3Zero);
2168	>	if (time >= info_->getSimParams()->getDt()) {
2169	>	Jz = kineticExchange_ / (time * avgArea)
2170	>	/ PhysicalConstants::energyConvert;
2171	>	JzP = momentumExchange_ / (time * avgArea);
2172	>	JzL = angularMomentumExchange_ / (time * avgArea);
2173	>	}
2174
2175	<	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &valueHist_[0],
2176	<	rnemdLogWidth_, MPI::REALTYPE, MPI::SUM);
2177	<	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &valueCount_[0],
2178	<	rnemdLogWidth_, MPI::INT, MPI::SUM);
2179	<	if (rnemdType_ == rnemdPx || rnemdType_ == rnemdPxScale) {
2180	<	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &xTempHist_[0],
2181	<	rnemdLogWidth_, MPI::REALTYPE, MPI::SUM);
1001	<	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &yTempHist_[0],
1002	<	rnemdLogWidth_, MPI::REALTYPE, MPI::SUM);
1003	<	MPI::COMM_WORLD.Allreduce(MPI::IN_PLACE, &zTempHist_[0],
1004	<	rnemdLogWidth_, MPI::REALTYPE, MPI::SUM);
1005	<	}
1006	<	// If we're the root node, should we print out the results
1007	<	int worldRank = MPI::COMM_WORLD.Get_rank();
1008	<	if (worldRank == 0) {
1009	<	#endif
1010	<	rnemdLog_ << time;
1011	<	for (j = 0; j < rnemdLogWidth_; j++) {
1012	<	rnemdLog_ << "\t" << valueHist_[j] / (RealType)valueCount_[j];
2175	>	rnemdFile_ << "#######################################################\n";
2176	>	rnemdFile_ << "# RNEMD {\n";
2177	>
2178	>	map<string, RNEMDMethod>::iterator mi;
2179	>	for(mi = stringToMethod_.begin(); mi != stringToMethod_.end(); ++mi) {
2180	>	if ( (*mi).second == rnemdMethod_)
2181	>	rnemdFile_ << "#    exchangeMethod  = \"" << (*mi).first << "\";\n";
2182		}
2183	<	rnemdLog_ << "\n";
2184	<	if (rnemdType_ == rnemdPx || rnemdType_ == rnemdPxScale ) {
2185	<	xTempLog_ << time;
2186	<	for (j = 0; j < rnemdLogWidth_; j++) {
2187	<	xTempLog_ << "\t" << xTempHist_[j] / (RealType)valueCount_[j];
2183	>	map<string, RNEMDFluxType>::iterator fi;
2184	>	for(fi = stringToFluxType_.begin(); fi != stringToFluxType_.end(); ++fi) {
2185	>	if ( (*fi).second == rnemdFluxType_)
2186	>	rnemdFile_ << "#    fluxType  = \"" << (*fi).first << "\";\n";
2187	>	}
2188	>
2189	>	rnemdFile_ << "#    exchangeTime = " << exchangeTime_ << ";\n";
2190	>
2191	>	rnemdFile_ << "#    objectSelection = \""
2192	>	<< rnemdObjectSelection_ << "\";\n";
2193	>	rnemdFile_ << "#    selectionA = \"" << selectionA_ << "\";\n";
2194	>	rnemdFile_ << "#    selectionB = \"" << selectionB_ << "\";\n";
2195	>	rnemdFile_ << "# }\n";
2196	>	rnemdFile_ << "#######################################################\n";
2197	>	rnemdFile_ << "# RNEMD report:\n";
2198	>	rnemdFile_ << "#      running time = " << time << " fs\n";
2199	>	rnemdFile_ << "# Target flux:\n";
2200	>	rnemdFile_ << "#           kinetic = "
2201	>	<< kineticFlux_ / PhysicalConstants::energyConvert
2202	>	<< " (kcal/mol/A^2/fs)\n";
2203	>	rnemdFile_ << "#          momentum = " << momentumFluxVector_
2204	>	<< " (amu/A/fs^2)\n";
2205	>	rnemdFile_ << "#  angular momentum = " << angularMomentumFluxVector_
2206	>	<< " (amu/A^2/fs^2)\n";
2207	>	rnemdFile_ << "# Target one-time exchanges:\n";
2208	>	rnemdFile_ << "#          kinetic = "
2209	>	<< kineticTarget_ / PhysicalConstants::energyConvert
2210	>	<< " (kcal/mol)\n";
2211	>	rnemdFile_ << "#          momentum = " << momentumTarget_
2212	>	<< " (amu*A/fs)\n";
2213	>	rnemdFile_ << "#  angular momentum = " << angularMomentumTarget_
2214	>	<< " (amu*A^2/fs)\n";
2215	>	rnemdFile_ << "# Actual exchange totals:\n";
2216	>	rnemdFile_ << "#          kinetic = "
2217	>	<< kineticExchange_ / PhysicalConstants::energyConvert
2218	>	<< " (kcal/mol)\n";
2219	>	rnemdFile_ << "#          momentum = " << momentumExchange_
2220	>	<< " (amu*A/fs)\n";
2221	>	rnemdFile_ << "#  angular momentum = " << angularMomentumExchange_
2222	>	<< " (amu*A^2/fs)\n";
2223	>	rnemdFile_ << "# Actual flux:\n";
2224	>	rnemdFile_ << "#          kinetic = " << Jz
2225	>	<< " (kcal/mol/A^2/fs)\n";
2226	>	rnemdFile_ << "#          momentum = " << JzP
2227	>	<< " (amu/A/fs^2)\n";
2228	>	rnemdFile_ << "#  angular momentum = " << JzL
2229	>	<< " (amu/A^2/fs^2)\n";
2230	>	rnemdFile_ << "# Exchange statistics:\n";
2231	>	rnemdFile_ << "#               attempted = " << trialCount_ << "\n";
2232	>	rnemdFile_ << "#                  failed = " << failTrialCount_ << "\n";
2233	>	if (rnemdMethod_ == rnemdNIVS) {
2234	>	rnemdFile_ << "#  NIVS root-check errors = "
2235	>	<< failRootCount_ << "\n";
2236	>	}
2237	>	rnemdFile_ << "#######################################################\n";
2238	>
2239	>
2240	>
2241	>	//write title
2242	>	rnemdFile_ << "#";
2243	>	for (unsigned int i = 0; i < outputMask_.size(); ++i) {
2244	>	if (outputMask_[i]) {
2245	>	rnemdFile_ << "\t" << data_[i].title <<
2246	>	"(" << data_[i].units << ")";
2247	>	// add some extra tabs for column alignment
2248	>	if (data_[i].dataType == "Vector3d") rnemdFile_ << "\t\t";
2249		}
2250	<	xTempLog_ << "\n";
2251	<	yTempLog_ << time;
2252	<	for (j = 0; j < rnemdLogWidth_; j++) {
2253	<	yTempLog_ << "\t" << yTempHist_[j] / (RealType)valueCount_[j];
2250	>	}
2251	>	rnemdFile_ << std::endl;
2252	>
2253	>	rnemdFile_.precision(8);
2254	>
2255	>	for (int j = 0; j < nBins_; j++) {
2256	>
2257	>	for (unsigned int i = 0; i < outputMask_.size(); ++i) {
2258	>	if (outputMask_[i]) {
2259	>	if (data_[i].dataType == "RealType")
2260	>	writeReal(i,j);
2261	>	else if (data_[i].dataType == "Vector3d")
2262	>	writeVector(i,j);
2263	>	else {
2264	>	sprintf( painCave.errMsg,
2265	>	"RNEMD found an unknown data type for: %s ",
2266	>	data_[i].title.c_str());
2267	>	painCave.isFatal = 1;
2268	>	simError();
2269	>	}
2270	>	}
2271		}
2272	<	yTempLog_ << "\n";
2273	<	zTempLog_ << time;
2274	<	for (j = 0; j < rnemdLogWidth_; j++) {
2275	<	zTempLog_ << "\t" << zTempHist_[j] / (RealType)valueCount_[j];
2272	>	rnemdFile_ << std::endl;
2273	>
2274	>	}
2275	>
2276	>	rnemdFile_ << "#######################################################\n";
2277	>	rnemdFile_ << "# 95% confidence intervals in those quantities follow:\n";
2278	>	rnemdFile_ << "#######################################################\n";
2279	>
2280	>
2281	>	for (int j = 0; j < nBins_; j++) {
2282	>	rnemdFile_ << "#";
2283	>	for (unsigned int i = 0; i < outputMask_.size(); ++i) {
2284	>	if (outputMask_[i]) {
2285	>	if (data_[i].dataType == "RealType")
2286	>	writeRealErrorBars(i,j);
2287	>	else if (data_[i].dataType == "Vector3d")
2288	>	writeVectorErrorBars(i,j);
2289	>	else {
2290	>	sprintf( painCave.errMsg,
2291	>	"RNEMD found an unknown data type for: %s ",
2292	>	data_[i].title.c_str());
2293	>	painCave.isFatal = 1;
2294	>	simError();
2295	>	}
2296	>	}
2297		}
2298	<	zTempLog_ << "\n";
2299	<	}
2298	>	rnemdFile_ << std::endl;
2299	>
2300	>	}
2301	>
2302	>	rnemdFile_.flush();
2303	>	rnemdFile_.close();
2304	>
2305		#ifdef IS_MPI
2306		}
2307		#endif
2308	<	for (j = 0; j < rnemdLogWidth_; j++) {
2309	<	valueCount_[j] = 0;
2310	<	valueHist_[j] = 0.0;
2308	>
2309	>	}
2310	>
2311	>	void RNEMD::writeReal(int index, unsigned int bin) {
2312	>	if (!doRNEMD_) return;
2313	>	assert(index >=0 && index < ENDINDEX);
2314	>	assert(int(bin) < nBins_);
2315	>	RealType s;
2316	>	int count;
2317	>
2318	>	count = data_[index].accumulator[bin]->count();
2319	>	if (count == 0) return;
2320	>
2321	>	dynamic_cast<Accumulator *>(data_[index].accumulator[bin])->getAverage(s);
2322	>
2323	>	if (! isinf(s) && ! isnan(s)) {
2324	>	rnemdFile_ << "\t" << s;
2325	>	} else{
2326	>	sprintf( painCave.errMsg,
2327	>	"RNEMD detected a numerical error writing: %s for bin %u",
2328	>	data_[index].title.c_str(), bin);
2329	>	painCave.isFatal = 1;
2330	>	simError();
2331	>	}
2332	>	}
2333	>
2334	>	void RNEMD::writeVector(int index, unsigned int bin) {
2335	>	if (!doRNEMD_) return;
2336	>	assert(index >=0 && index < ENDINDEX);
2337	>	assert(int(bin) < nBins_);
2338	>	Vector3d s;
2339	>	int count;
2340	>
2341	>	count = data_[index].accumulator[bin]->count();
2342	>
2343	>	if (count == 0) return;
2344	>
2345	>	dynamic_cast<VectorAccumulator*>(data_[index].accumulator[bin])->getAverage(s);
2346	>	if (isinf(s[0]) || isnan(s[0]) ||
2347	>	isinf(s[1]) || isnan(s[1]) ||
2348	>	isinf(s[2]) || isnan(s[2]) ) {
2349	>	sprintf( painCave.errMsg,
2350	>	"RNEMD detected a numerical error writing: %s for bin %u",
2351	>	data_[index].title.c_str(), bin);
2352	>	painCave.isFatal = 1;
2353	>	simError();
2354	>	} else {
2355	>	rnemdFile_ << "\t" << s[0] << "\t" << s[1] << "\t" << s[2];
2356		}
2357	<	if (rnemdType_ == rnemdPx || rnemdType_ == rnemdPxScale)
2358	<	for (j = 0; j < rnemdLogWidth_; j++) {
2359	<	xTempHist_[j] = 0.0;
2360	<	yTempHist_[j] = 0.0;
2361	<	zTempHist_[j] = 0.0;
2362	<	}
2357	>	}
2358	>
2359	>	void RNEMD::writeRealErrorBars(int index, unsigned int bin) {
2360	>	if (!doRNEMD_) return;
2361	>	assert(index >=0 && index < ENDINDEX);
2362	>	assert(int(bin) < nBins_);
2363	>	RealType s;
2364	>	int count;
2365	>
2366	>	count = data_[index].accumulator[bin]->count();
2367	>	if (count == 0) return;
2368	>
2369	>	dynamic_cast<Accumulator *>(data_[index].accumulator[bin])->get95percentConfidenceInterval(s);
2370	>
2371	>	if (! isinf(s) && ! isnan(s)) {
2372	>	rnemdFile_ << "\t" << s;
2373	>	} else{
2374	>	sprintf( painCave.errMsg,
2375	>	"RNEMD detected a numerical error writing: %s std. dev. for bin %u",
2376	>	data_[index].title.c_str(), bin);
2377	>	painCave.isFatal = 1;
2378	>	simError();
2379	>	}
2380		}
2381	+
2382	+	void RNEMD::writeVectorErrorBars(int index, unsigned int bin) {
2383	+	if (!doRNEMD_) return;
2384	+	assert(index >=0 && index < ENDINDEX);
2385	+	assert(int(bin) < nBins_);
2386	+	Vector3d s;
2387	+	int count;
2388	+
2389	+	count = data_[index].accumulator[bin]->count();
2390	+	if (count == 0) return;
2391	+
2392	+	dynamic_cast<VectorAccumulator*>(data_[index].accumulator[bin])->get95percentConfidenceInterval(s);
2393	+	if (isinf(s[0]) || isnan(s[0]) ||
2394	+	isinf(s[1]) || isnan(s[1]) ||
2395	+	isinf(s[2]) || isnan(s[2]) ) {
2396	+	sprintf( painCave.errMsg,
2397	+	"RNEMD detected a numerical error writing: %s std. dev. for bin %u",
2398	+	data_[index].title.c_str(), bin);
2399	+	painCave.isFatal = 1;
2400	+	simError();
2401	+	} else {
2402	+	rnemdFile_ << "\t" << s[0] << "\t" << s[1] << "\t" << s[2];
2403	+	}
2404	+	}
2405		}
2406	+

Diff Legend

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