Line data Source code
1 : /* +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
2 : Copyright (c) 2017 of Glen Hocky and Andrew White
3 :
4 : The eds module is free software: you can redistribute it and/or modify
5 : it under the terms of the GNU Lesser General Public License as published by
6 : the Free Software Foundation, either version 3 of the License, or
7 : (at your option) any later version.
8 :
9 : The eds module is distributed in the hope that it will be useful,
10 : but WITHOUT ANY WARRANTY; without even the implied warranty of
11 : MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12 : GNU Lesser General Public License for more details.
13 :
14 : You should have received a copy of the GNU Lesser General Public License
15 : along with plumed. If not, see <http://www.gnu.org/licenses/>.
16 : +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ */
17 : #include "bias/Bias.h"
18 : #include "bias/ReweightBase.h"
19 : #include "core/ActionAtomistic.h"
20 : #include "core/ActionRegister.h"
21 : #include "core/Atoms.h"
22 : #include "core/PlumedMain.h"
23 : #include "tools/File.h"
24 : #include "tools/Matrix.h"
25 : #include "tools/Random.h"
26 :
27 : #include <iostream>
28 :
29 : using namespace PLMD;
30 : using namespace bias;
31 :
32 : // namespace is lowercase to match
33 : // module names being all lowercase
34 :
35 : namespace PLMD {
36 : namespace eds {
37 :
38 : //+PLUMEDOC EDSMOD_BIAS EDS
39 : /*
40 : Add a linear bias on a set of observables.
41 :
42 : This force is the same as the linear part of the bias in \ref
43 : RESTRAINT, but this bias has the ability to compute the prefactors
44 : adaptively using the scheme of White and Voth \cite white2014efficient
45 : in order to match target observable values for a set of CVs.
46 : Further updates to the algorithm are described in \cite hocky2017cgds
47 : and you can read a review on the method and its applications here: \cite Amirkulova2019Recent.
48 :
49 : You can
50 : see a tutorial on EDS specifically for biasing coordination number at
51 : <a
52 : href="http://thewhitelab.org/blog/tutorial/2017/05/10/lammps-coordination-number-tutorial/">
53 : Andrew White's webpage</a>.
54 :
55 : The addition to the potential is of the form
56 : \f[
57 : \sum_i \frac{\alpha_i}{s_i} x_i
58 : \f]
59 :
60 : where for CV \f$x_i\f$, a coupling constant \f${\alpha}_i\f$ is determined
61 : adaptively or set by the user to match a target value for
62 : \f$x_i\f$. \f$s_i\f$ is a scale parameter, which by default is set to
63 : the target value. It may also be set separately.
64 :
65 : \warning
66 : It is not possible to set the target value of the observable
67 : to zero with the default value of \f$s_i\f$ as this will cause a
68 : divide-by-zero error. Instead, set \f$s_i=1\f$ or modify the CV so the
69 : desired target value is no longer zero.
70 :
71 : Notice that a similar method is available as \ref MAXENT, although with different features and using a different optimization algorithm.
72 :
73 : \par Virial
74 :
75 : The bias forces modify the virial and this can change your simulation density if the bias is used in an NPT simulation.
76 : One way to avoid changing the virial contribution from the bias is to add the keyword VIRIAL=1.0, which changes how the bias
77 : is computed to minimize its contribution to the virial. This can also lead to smaller magnitude biases that are more robust if
78 : transferred to other systems. VIRIAL=1.0 can be a reasonable starting point and increasing the value changes the balance between matching
79 : the set-points and minimizing the virial. See \cite Amirkulova2019Recent for details on the equations. Since the coupling constants
80 : are unique with a single CV, VIRIAL is not applicable with a single CV. When used with multiple CVs, the CVs should be correlated
81 : which is almost always the case.
82 :
83 : \par Weighting
84 :
85 : EDS computes means and variances as part of its algorithm. If you are
86 : also using a biasing method like metadynamics, you may wish to remove
87 : the effect of this bias in your EDS computations so that EDS works on
88 : the canonical values (reweighted to be unbiased). For example, you may be using
89 : metadynamics to bias a dihedral angle to enhance sampling and be using
90 : EDS to set the average distance between two particular atoms. Specifically:
91 :
92 : \plumedfile
93 : # set-up metadynamics
94 : t: TORSION ATOMS=1,2,3,4
95 : md: METAD ARG=d SIGMA=0.2 HEIGHT=0.3 PACE=500 TEMP=300
96 : # compute bias weights
97 : bias: REWEIGHT_METAD TEMP=300
98 : # now do EDS on distance while removing effect of metadynamics
99 : d: DISTANCE ATOMS=4,7
100 : eds: EDS ARG=d CENTER=3.0 PERIOD=100 TEMP=300 LOGWEIGHTS=bias
101 : \endplumedfile
102 :
103 : This is an approximation though because EDS uses a finite samples while running to get means/variances.
104 : At the end of a run,
105 : you should ensure this approach worked and indeed your reweighted CV matches the target value.
106 :
107 : \par Examples
108 :
109 : The following input for a harmonic oscillator of two beads will
110 : adaptively find a linear bias to change the mean and variance to the
111 : target values. The PRINT line shows how to access the value of the
112 : coupling constants.
113 :
114 : \plumedfile
115 : dist: DISTANCE ATOMS=1,2
116 : # this is the squared of the distance
117 : dist2: COMBINE ARG=dist POWERS=2 PERIODIC=NO
118 :
119 : # bias mean and variance
120 : eds: EDS ARG=dist,dist2 CENTER=2.0,1.0 PERIOD=100 TEMP=1.0
121 : PRINT ARG=dist,dist2,eds.dist_coupling,eds.dist2_coupling,eds.bias,eds.force2 FILE=colvars.dat STRIDE=100
122 : \endplumedfile
123 :
124 : <hr>
125 :
126 : Rather than trying to find the coupling constants adaptively, one can ramp up to a constant value.
127 : \plumedfile
128 : dist: DISTANCE ATOMS=1,2
129 : dist2: COMBINE ARG=dist POWERS=2 PERIODIC=NO
130 :
131 : # ramp couplings from 0,0 to -1,1 over 50000 steps
132 : eds: EDS ARG=dist,dist2 CENTER=2.0,1.0 FIXED=-1,1 RAMP PERIOD=50000 TEMP=1.0
133 :
134 : # same as above, except starting at -0.5,0.5 rather than default of 0,0
135 : eds2: EDS ARG=dist,dist2 CENTER=2.0,1.0 FIXED=-1,1 INIT=-0.5,0.5 RAMP PERIOD=50000 TEMP=1.0
136 : \endplumedfile
137 :
138 : <hr>
139 : A restart file can be added to dump information needed to restart/continue simulation using these parameters every PERIOD.
140 : \plumedfile
141 : dist: DISTANCE ATOMS=1,2
142 : dist2: COMBINE ARG=dist POWERS=2 PERIODIC=NO
143 :
144 : # add the option to write to a restart file
145 : eds: EDS ARG=dist,dist2 CENTER=2.0,1.0 PERIOD=100 TEMP=1.0 OUT_RESTART=checkpoint.eds
146 : \endplumedfile
147 :
148 : The first few lines of the restart file that is output if we run a calculation with one CV will look something like this:
149 :
150 : \auxfile{restart.eds}
151 : #! FIELDS time d1_center d1_set d1_target d1_coupling d1_maxrange d1_maxgrad d1_accum d1_mean d1_std
152 : #! SET adaptive 1
153 : #! SET update_period 1
154 : #! SET seed 0
155 : #! SET kbt 2.4943
156 : 0.0000 1.0000 0.0000 0.0000 0.0000 7.4830 0.1497 0.0000 0.0000 0.0000
157 : 1.0000 1.0000 0.0000 0.0000 0.0000 7.4830 0.1497 0.0000 0.0000 0.0000
158 : 2.0000 1.0000 -7.4830 0.0000 0.0000 7.4830 0.1497 0.0224 0.0000 0.0000
159 : 3.0000 1.0000 -7.4830 0.0000 -7.4830 7.4830 0.1497 0.0224 0.0000 0.0000
160 : 4.0000 1.0000 -7.4830 0.0000 -7.4830 7.4830 0.1497 0.0224 0.0000 0.0000
161 : \endauxfile
162 :
163 : <hr>
164 :
165 : Read in a previous restart file. Adding RESTART flag makes output append
166 : \plumedfile
167 : d1: DISTANCE ATOMS=1,2
168 :
169 : eds: EDS ARG=d1 CENTER=2.0 PERIOD=100 TEMP=1.0 IN_RESTART=restart.eds RESTART=YES
170 : \endplumedfile
171 :
172 : <hr>
173 :
174 : Read in a previous restart file and freeze the bias at the final level from the previous simulation
175 : \plumedfile
176 : d1: DISTANCE ATOMS=1,2
177 :
178 : eds: EDS ARG=d1 CENTER=2.0 TEMP=1.0 IN_RESTART=restart.eds FREEZE
179 : \endplumedfile
180 :
181 : <hr>
182 :
183 : Read in a previous restart file and freeze the bias at the mean from the previous simulation
184 : \plumedfile
185 : d1: DISTANCE ATOMS=1,2
186 :
187 : eds: EDS ARG=d1 CENTER=2.0 TEMP=1.0 IN_RESTART=restart.eds FREEZE MEAN
188 : \endplumedfile
189 :
190 :
191 : */
192 : //+ENDPLUMEDOC
193 :
194 : class EDS : public Bias {
195 :
196 : private:
197 : /*We will get this and store it once, since on-the-fly changing number of CVs will be fatal*/
198 : const unsigned int ncvs_;
199 : std::vector<double> center_;
200 : std::vector<Value *> center_values_;
201 : ReweightBase *logweights_; // weights to use if reweighting averages
202 : std::vector<double> scale_;
203 : std::vector<double> current_coupling_; // actually current coupling
204 : std::vector<double> set_coupling_; // what our coupling is ramping up to. Equal to current_coupling when gathering stats
205 : std::vector<double> target_coupling_; // used when loaded to reach a value
206 : std::vector<double> max_coupling_range_; // used for adaptive range
207 : std::vector<double> max_coupling_grad_; // maximum allowed gradient
208 : std::vector<double> coupling_rate_;
209 : std::vector<double> coupling_accum_;
210 : std::vector<double> means_;
211 : std::vector<double> differences_;
212 : std::vector<double> alpha_vector_;
213 : std::vector<double> alpha_vector_2_;
214 : std::vector<double> ssds_;
215 : std::vector<double> step_size_;
216 : std::vector<double> pseudo_virial_;
217 : std::vector<Value *> out_coupling_;
218 : Matrix<double> covar_;
219 : Matrix<double> covar2_;
220 : Matrix<double> lm_inv_;
221 : std::string in_restart_name_;
222 : std::string out_restart_name_;
223 : std::string fmt_;
224 : OFile out_restart_;
225 : IFile in_restart_;
226 : bool b_c_values_;
227 : bool b_adaptive_;
228 : bool b_freeze_;
229 : bool b_equil_;
230 : bool b_ramp_;
231 : bool b_covar_;
232 : bool b_restart_;
233 : bool b_write_restart_;
234 : bool b_lm_;
235 : bool b_virial_;
236 : bool b_update_virial_;
237 : bool b_weights_;
238 : int seed_;
239 : int update_period_;
240 : int avg_coupling_count_;
241 : int update_calls_;
242 : double kbt_;
243 : double multi_prop_;
244 : double lm_mixing_par_;
245 : double virial_scaling_;
246 : double pseudo_virial_sum_; // net virial for all cvs in current period
247 : double max_logweight_; // maximum observed max logweight for period
248 : double wsum_; // sum of weights thus far
249 : Random rand_;
250 : Value *value_force2_;
251 : Value *value_pressure_;
252 :
253 : /*read input restart. b_mean sets if we use mean or final value for freeze*/
254 : void readInRestart(const bool b_mean);
255 : /*setup output restart*/
256 : void setupOutRestart();
257 : /*write output restart*/
258 : void writeOutRestart();
259 : void update_statistics();
260 : void update_pseudo_virial();
261 : void calc_lm_step_size();
262 : void calc_covar_step_size();
263 : void calc_ssd_step_size();
264 : void reset_statistics();
265 : void update_bias();
266 : void apply_bias();
267 :
268 : public:
269 : explicit EDS(const ActionOptions &);
270 : void calculate();
271 : void update();
272 : void turnOnDerivatives();
273 : static void registerKeywords(Keywords &keys);
274 : ~EDS();
275 : };
276 :
277 13801 : PLUMED_REGISTER_ACTION(EDS, "EDS")
278 :
279 12 : void EDS::registerKeywords(Keywords &keys) {
280 12 : Bias::registerKeywords(keys);
281 12 : keys.use("ARG");
282 24 : keys.add("optional", "CENTER", "The desired centers (equilibrium values) which will be sought during the adaptive linear biasing. This is for fixed centers");
283 24 : keys.add("optional", "CENTER_ARG", "The desired centers (equilibrium values) which will be sought during the adaptive linear biasing. "
284 : "CENTER_ARG is for calculated centers, e.g. from a CV or analysis. ");
285 :
286 24 : keys.add("optional", "PERIOD", "Steps over which to adjust bias for adaptive or ramping");
287 24 : keys.add("compulsory", "RANGE", "25.0", "The (starting) maximum increase in coupling constant per PERIOD (in \\f$k_B T\\f$/[BIAS_SCALE unit]) for each CV biased");
288 24 : keys.add("compulsory", "SEED", "0", "Seed for random order of changing bias");
289 24 : keys.add("compulsory", "INIT", "0", "Starting value for coupling constant");
290 24 : keys.add("compulsory", "FIXED", "0", "Fixed target values for coupling constant. Non-adaptive.");
291 24 : keys.add("optional", "BIAS_SCALE", "A divisor to set the units of the bias. "
292 : "If not set, this will be the CENTER value by default (as is done in White and Voth 2014).");
293 24 : keys.add("optional", "TEMP", "The system temperature. If not provided will be taken from MD code (if available)");
294 24 : keys.add("optional", "MULTI_PROP", "What proportion of dimensions to update at each step. "
295 : "Must be in interval [1,0), where 1 indicates all and any other indicates a stochastic update. "
296 : "If not set, default is 1 / N, where N is the number of CVs. ");
297 24 : keys.add("optional", "VIRIAL", "Add an update penalty for having non-zero virial contributions. Only makes sense with multiple correlated CVs.");
298 24 : keys.add("optional", "LOGWEIGHTS", "Add weights to use for computing statistics. For example, if biasing with metadynamics.");
299 24 : keys.addFlag("LM", false, "Use Levenberg-Marquadt algorithm along with simultaneous keyword. Otherwise use gradient descent.");
300 24 : keys.add("compulsory", "LM_MIXING", "1", "Initial mixing parameter when using Levenberg-Marquadt minimization.");
301 24 : keys.add("optional", "RESTART_FMT", "the format that should be used to output real numbers in EDS restarts");
302 24 : keys.add("optional", "OUT_RESTART", "Output file for all information needed to continue EDS simulation. "
303 : "If you have the RESTART directive set (global or for EDS), this file will be appended to. "
304 : "Note that the header will be printed again if appending.");
305 24 : keys.add("optional", "IN_RESTART", "Read this file to continue an EDS simulation. "
306 : "If same as OUT_RESTART and you have not set the RESTART directive, the file will be backed-up and overwritten with new output. "
307 : "If you do have the RESTART flag set and it is the same name as OUT_RESTART, this file will be appended.");
308 :
309 24 : keys.addFlag("RAMP", false, "Slowly increase bias constant to a fixed value");
310 24 : keys.addFlag("COVAR", false, "Utilize the covariance matrix when updating the bias. Default Off, but may be enabled due to other options");
311 24 : keys.addFlag("FREEZE", false, "Fix bias at current level (only used for restarting).");
312 24 : keys.addFlag("MEAN", false, "Instead of using final bias level from restart, use average. Can only be used in conjunction with FREEZE");
313 :
314 12 : keys.use("RESTART");
315 :
316 24 : keys.addOutputComponent("force2", "default", "squared value of force from the bias");
317 24 : keys.addOutputComponent("pressure", "default", "If using virial keyword, this is the current sum of virial terms. It is in units of pressure (energy / vol^3)");
318 24 : keys.addOutputComponent("_coupling", "default", "For each named CV biased, there will be a corresponding output CV_coupling storing the current linear bias prefactor.");
319 12 : }
320 :
321 8 : EDS::EDS(const ActionOptions &ao) : PLUMED_BIAS_INIT(ao),
322 8 : ncvs_(getNumberOfArguments()),
323 8 : scale_(ncvs_, 0.0),
324 8 : current_coupling_(ncvs_, 0.0),
325 8 : set_coupling_(ncvs_, 0.0),
326 8 : target_coupling_(ncvs_, 0.0),
327 8 : max_coupling_range_(ncvs_, 25.0),
328 8 : max_coupling_grad_(ncvs_, 0.0),
329 8 : coupling_rate_(ncvs_, 1.0),
330 8 : coupling_accum_(ncvs_, 0.0),
331 8 : means_(ncvs_, 0.0),
332 8 : step_size_(ncvs_, 0.0),
333 8 : pseudo_virial_(ncvs_),
334 8 : out_coupling_(ncvs_, NULL),
335 8 : in_restart_name_(""),
336 8 : out_restart_name_(""),
337 8 : fmt_("%f"),
338 8 : b_adaptive_(true),
339 8 : b_freeze_(false),
340 8 : b_equil_(true),
341 8 : b_ramp_(false),
342 8 : b_covar_(false),
343 8 : b_restart_(false),
344 8 : b_write_restart_(false),
345 8 : b_lm_(false),
346 8 : b_virial_(false),
347 8 : b_weights_(false),
348 8 : seed_(0),
349 8 : update_period_(0),
350 8 : avg_coupling_count_(1),
351 8 : update_calls_(0),
352 8 : kbt_(0.0),
353 8 : multi_prop_(-1.0),
354 8 : lm_mixing_par_(0.1),
355 8 : virial_scaling_(0.),
356 8 : pseudo_virial_sum_(0.0),
357 8 : max_logweight_(0.0),
358 8 : wsum_(0.0),
359 32 : value_force2_(NULL) {
360 8 : double temp = -1.0;
361 8 : bool b_mean = false;
362 : std::vector<Value *> wvalues;
363 :
364 8 : addComponent("force2");
365 8 : componentIsNotPeriodic("force2");
366 8 : value_force2_ = getPntrToComponent("force2");
367 :
368 20 : for (unsigned int i = 0; i < ncvs_; ++i) {
369 12 : std::string comp = getPntrToArgument(i)->getName() + "_coupling";
370 12 : addComponent(comp);
371 12 : componentIsNotPeriodic(comp);
372 12 : out_coupling_[i] = getPntrToComponent(comp);
373 : }
374 :
375 8 : parseVector("CENTER", center_);
376 8 : parseArgumentList("CENTER_ARG", center_values_);
377 8 : parseArgumentList("LOGWEIGHTS", wvalues);
378 8 : parseVector("BIAS_SCALE", scale_);
379 8 : parseVector("RANGE", max_coupling_range_);
380 8 : parseVector("FIXED", target_coupling_);
381 8 : parseVector("INIT", set_coupling_);
382 8 : parse("PERIOD", update_period_);
383 8 : parse("TEMP", temp);
384 8 : parse("SEED", seed_);
385 8 : parse("MULTI_PROP", multi_prop_);
386 8 : parse("LM_MIXING", lm_mixing_par_);
387 8 : parse("RESTART_FMT", fmt_);
388 8 : parse("VIRIAL", virial_scaling_);
389 8 : fmt_ = " " + fmt_; // add space since parse strips them
390 8 : parse("OUT_RESTART", out_restart_name_);
391 8 : parseFlag("LM", b_lm_);
392 8 : parseFlag("RAMP", b_ramp_);
393 8 : parseFlag("FREEZE", b_freeze_);
394 8 : parseFlag("MEAN", b_mean);
395 8 : parseFlag("COVAR", b_covar_);
396 8 : parse("IN_RESTART", in_restart_name_);
397 8 : checkRead();
398 :
399 : /*
400 : * Things that are different when using changing centers:
401 : * 1. Scale
402 : * 2. The log file
403 : * 3. Reading Restarts
404 : */
405 :
406 8 : if (center_.size() == 0) {
407 1 : if (center_values_.size() == 0) {
408 0 : error("Must set either CENTER or CENTER_ARG");
409 1 : } else if (center_values_.size() != ncvs_) {
410 0 : error("CENTER_ARG must contain the same number of variables as ARG");
411 : }
412 1 : b_c_values_ = true;
413 1 : center_.resize(ncvs_);
414 1 : log.printf(" EDS will use possibly varying centers\n");
415 : } else {
416 7 : if (center_.size() != ncvs_) {
417 0 : error("Must have same number of CENTER arguments as ARG arguments");
418 7 : } else if (center_values_.size() != 0) {
419 0 : error("You can only set CENTER or CENTER_ARG. Not both");
420 : }
421 7 : b_c_values_ = false;
422 7 : log.printf(" EDS will use fixed centers\n");
423 : }
424 :
425 : // check for weights
426 8 : if (wvalues.size() > 1) {
427 0 : error("LOGWEIGHTS can only support one weight set. Please only pass one action");
428 8 : } else if (wvalues.size() == 1) {
429 1 : logweights_ = dynamic_cast<ReweightBase *>(wvalues[0]->getPntrToAction());
430 1 : b_weights_ = true;
431 : }
432 :
433 8 : log.printf(" setting scaling:");
434 8 : if (scale_.size() > 0 && scale_.size() < ncvs_) {
435 0 : error("the number of BIAS_SCALE values be the same as number of CVs");
436 8 : } else if (scale_.size() == 0 && b_c_values_) {
437 0 : log.printf(" Setting SCALE to be 1 for all CVs\n");
438 0 : scale_.resize(ncvs_);
439 0 : for (unsigned int i = 0; i < ncvs_; ++i) {
440 0 : scale_[i] = 1;
441 : }
442 8 : } else if (scale_.size() == 0 && !b_c_values_) {
443 2 : log.printf(" (default) ");
444 :
445 2 : scale_.resize(ncvs_);
446 6 : for (unsigned int i = 0; i < scale_.size(); ++i) {
447 4 : if (center_[i] == 0) {
448 0 : error("BIAS_SCALE parameter has been set to CENTER value of 0 (as is default). This will divide by 0, so giving up. See doc for EDS bias");
449 : }
450 4 : scale_[i] = center_[i];
451 : }
452 : } else {
453 14 : for (unsigned int i = 0; i < scale_.size(); ++i) {
454 8 : log.printf(" %f", scale_[i]);
455 : }
456 : }
457 8 : log.printf("\n");
458 :
459 8 : if (b_lm_) {
460 1 : log.printf(" EDS will perform Levenberg-Marquardt minimization with mixing parameter = %f\n", lm_mixing_par_);
461 1 : differences_.resize(ncvs_);
462 1 : alpha_vector_.resize(ncvs_);
463 1 : alpha_vector_2_.resize(ncvs_);
464 1 : covar_.resize(ncvs_, ncvs_);
465 1 : covar2_.resize(ncvs_, ncvs_);
466 1 : lm_inv_.resize(ncvs_, ncvs_);
467 1 : covar2_ *= 0;
468 1 : lm_inv_ *= 0;
469 1 : if (multi_prop_ != 1) {
470 0 : log.printf(" WARNING - doing LM minimization but MULTI_PROP!=1\n");
471 : }
472 7 : } else if (b_covar_) {
473 1 : log.printf(" EDS will utilize covariance matrix for update steps\n");
474 1 : covar_.resize(ncvs_, ncvs_);
475 : } else {
476 6 : log.printf(" EDS will utilize variance for update steps\n");
477 6 : ssds_.resize(ncvs_);
478 : }
479 :
480 8 : b_virial_ = virial_scaling_;
481 :
482 8 : if (b_virial_) {
483 1 : if (ncvs_ == 1) {
484 0 : error("Minimizing the virial is only valid with multiply correlated collective variables.");
485 : }
486 : // check that the CVs can be used to compute pseudo-virial
487 1 : log.printf(" EDS will compute virials of CVs and penalize with scale of %f. Checking CVs are valid...", virial_scaling_);
488 4 : for (unsigned int i = 0; i < ncvs_; ++i) {
489 3 : auto a = dynamic_cast<ActionAtomistic *>(getPntrToArgument(i)->getPntrToAction());
490 3 : if (!a) {
491 0 : error("If using VIRIAL keyword, you must have normal CVs as arguments to EDS. Offending action: " + getPntrToArgument(i)->getPntrToAction()->getName());
492 : }
493 : // cppcheck-suppress nullPointerRedundantCheck
494 3 : if (!(a->getPbc().isOrthorombic())) {
495 3 : log.printf(" WARNING: EDS Virial should have a orthorombic cell\n");
496 : }
497 : }
498 1 : log.printf("done.\n");
499 1 : addComponent("pressure");
500 1 : componentIsNotPeriodic("pressure");
501 1 : value_pressure_ = getPntrToComponent("pressure");
502 : }
503 :
504 8 : if (b_mean && !b_freeze_) {
505 0 : error("EDS keyword MEAN can only be used along with keyword FREEZE");
506 : }
507 :
508 8 : if (in_restart_name_ != "") {
509 2 : b_restart_ = true;
510 2 : log.printf(" reading simulation information from file: %s\n", in_restart_name_.c_str());
511 2 : readInRestart(b_mean);
512 : } else {
513 :
514 6 : if (temp >= 0.0) {
515 6 : kbt_ = plumed.getAtoms().getKBoltzmann() * temp;
516 : } else {
517 0 : kbt_ = plumed.getAtoms().getKbT();
518 : }
519 :
520 : // in driver, this results in kbt of 0
521 6 : if (kbt_ == 0) {
522 0 : error(" Unable to determine valid kBT. "
523 : "Could be because you are runnning from driver or MD didn't give temperature.\n"
524 : "Consider setting temperature manually with the TEMP keyword.");
525 : kbt_ = 1;
526 : }
527 :
528 6 : log.printf(" kBT = %f\n", kbt_);
529 6 : log.printf(" Updating every %i steps\n", update_period_);
530 :
531 6 : if (!b_c_values_) {
532 5 : log.printf(" with centers:");
533 14 : for (unsigned int i = 0; i < ncvs_; ++i) {
534 9 : log.printf(" %f ", center_[i]);
535 : }
536 : } else {
537 1 : log.printf(" with actions centers:");
538 2 : for (unsigned int i = 0; i < ncvs_; ++i) {
539 1 : log.printf(" %s ", center_values_[i]->getName().c_str());
540 : // add dependency on these actions
541 1 : addDependency(center_values_[i]->getPntrToAction());
542 : }
543 : }
544 :
545 6 : log.printf("\n with initial ranges / rates:\n");
546 16 : for (unsigned int i = 0; i < max_coupling_range_.size(); ++i) {
547 : // this is just an empirical guess. Bigger range, bigger grads. Less frequent updates, bigger changes
548 : //
549 : // using the current maxing out scheme, max_coupling_range is the biggest step that can be taken in any given interval
550 10 : max_coupling_range_[i] *= kbt_;
551 10 : max_coupling_grad_[i] = max_coupling_range_[i];
552 10 : log.printf(" %f / %f\n", max_coupling_range_[i], max_coupling_grad_[i]);
553 : }
554 :
555 6 : if (seed_ > 0) {
556 2 : log.printf(" setting random seed = %i", seed_);
557 2 : rand_.setSeed(seed_);
558 : }
559 :
560 16 : for (unsigned int i = 0; i < ncvs_; ++i)
561 10 : if (target_coupling_[i] != 0.0) {
562 1 : b_adaptive_ = false;
563 : }
564 :
565 6 : if (!b_adaptive_) {
566 1 : if (b_ramp_) {
567 1 : log.printf(" ramping up coupling constants over %i steps\n", update_period_);
568 : }
569 :
570 1 : log.printf(" with starting coupling constants");
571 2 : for (unsigned int i = 0; i < set_coupling_.size(); ++i) {
572 1 : log.printf(" %f", set_coupling_[i]);
573 : }
574 1 : log.printf("\n");
575 1 : log.printf(" and final coupling constants");
576 2 : for (unsigned int i = 0; i < target_coupling_.size(); ++i) {
577 1 : log.printf(" %f", target_coupling_[i]);
578 : }
579 1 : log.printf("\n");
580 : }
581 :
582 : // now do setup
583 6 : if (b_ramp_) {
584 1 : update_period_ *= -1;
585 : }
586 :
587 16 : for (unsigned int i = 0; i < set_coupling_.size(); ++i) {
588 10 : current_coupling_[i] = set_coupling_[i];
589 : }
590 :
591 : // if b_adaptive_, then first half will be used for equilibrating and second half for statistics
592 6 : if (update_period_ > 0) {
593 5 : update_period_ /= 2;
594 : }
595 : }
596 :
597 8 : if (b_freeze_) {
598 1 : b_adaptive_ = false;
599 1 : update_period_ = 0;
600 1 : if (b_mean) {
601 1 : log.printf(" freezing bias at the average level from the restart file\n");
602 : } else {
603 0 : log.printf(" freezing bias at current level\n");
604 : }
605 : }
606 :
607 8 : if (multi_prop_ == -1.0) {
608 5 : log.printf(" Will update each dimension stochastically with probability 1 / number of CVs\n");
609 5 : multi_prop_ = 1.0 / ncvs_;
610 3 : } else if (multi_prop_ > 0 && multi_prop_ <= 1.0) {
611 3 : log.printf(" Will update each dimension stochastically with probability %f\n", multi_prop_);
612 : } else {
613 0 : error(" MULTI_PROP must be between 0 and 1\n");
614 : }
615 :
616 8 : if (out_restart_name_.length() > 0) {
617 8 : log.printf(" writing restart information every %i steps to file %s with format %s\n", abs(update_period_), out_restart_name_.c_str(), fmt_.c_str());
618 8 : b_write_restart_ = true;
619 8 : setupOutRestart();
620 : }
621 :
622 16 : log << " Bibliography " << plumed.cite("White and Voth, J. Chem. Theory Comput. 10 (8), 3023-3030 (2014)") << "\n";
623 16 : log << " Bibliography " << plumed.cite("G. M. Hocky, T. Dannenhoffer-Lafage, G. A. Voth, J. Chem. Theory Comput. 13 (9), 4593-4603 (2017)") << "\n";
624 8 : }
625 :
626 2 : void EDS::readInRestart(const bool b_mean) {
627 2 : int adaptive_i = 0;
628 :
629 2 : in_restart_.open(in_restart_name_);
630 :
631 4 : if (in_restart_.FieldExist("kbt")) {
632 2 : in_restart_.scanField("kbt", kbt_);
633 : } else {
634 0 : error("No field 'kbt' in restart file");
635 : }
636 2 : log.printf(" with kBT = %f\n", kbt_);
637 :
638 4 : if (in_restart_.FieldExist("update_period")) {
639 2 : in_restart_.scanField("update_period", update_period_);
640 : } else {
641 0 : error("No field 'update_period' in restart file");
642 : }
643 2 : log.printf(" Updating every %i steps\n", update_period_);
644 :
645 4 : if (in_restart_.FieldExist("adaptive")) {
646 : // note, no version of scanField for boolean
647 2 : in_restart_.scanField("adaptive", adaptive_i);
648 : } else {
649 0 : error("No field 'adaptive' in restart file");
650 : }
651 2 : b_adaptive_ = bool(adaptive_i);
652 :
653 4 : if (in_restart_.FieldExist("seed")) {
654 2 : in_restart_.scanField("seed", seed_);
655 : } else {
656 0 : error("No field 'seed' in restart file");
657 : }
658 2 : if (seed_ > 0) {
659 0 : log.printf(" setting random seed = %i", seed_);
660 0 : rand_.setSeed(seed_);
661 : }
662 :
663 : double time, tmp;
664 2 : std::vector<double> avg_bias = std::vector<double>(center_.size());
665 : unsigned int N = 0;
666 : std::string cv_name;
667 :
668 24 : while (in_restart_.scanField("time", time)) {
669 :
670 20 : for (unsigned int i = 0; i < ncvs_; ++i) {
671 : cv_name = getPntrToArgument(i)->getName();
672 10 : in_restart_.scanField(cv_name + "_center", set_coupling_[i]);
673 20 : in_restart_.scanField(cv_name + "_set", set_coupling_[i]);
674 20 : in_restart_.scanField(cv_name + "_target", target_coupling_[i]);
675 20 : in_restart_.scanField(cv_name + "_coupling", current_coupling_[i]);
676 20 : in_restart_.scanField(cv_name + "_maxrange", max_coupling_range_[i]);
677 20 : in_restart_.scanField(cv_name + "_maxgrad", max_coupling_grad_[i]);
678 20 : in_restart_.scanField(cv_name + "_accum", coupling_accum_[i]);
679 10 : in_restart_.scanField(cv_name + "_mean", means_[i]);
680 20 : if (in_restart_.FieldExist(cv_name + "_pseudovirial")) {
681 0 : if (b_virial_) {
682 0 : in_restart_.scanField(cv_name + "_pseudovirial", pseudo_virial_[i]);
683 : } else { // discard the field
684 0 : in_restart_.scanField(cv_name + "_pseudovirial", tmp);
685 : }
686 : }
687 : // unused due to difference between covar/nocovar
688 20 : in_restart_.scanField(cv_name + "_std", tmp);
689 :
690 10 : avg_bias[i] += current_coupling_[i];
691 : }
692 10 : N++;
693 :
694 10 : in_restart_.scanField();
695 : }
696 :
697 2 : log.printf(" with centers:");
698 4 : for (unsigned int i = 0; i < center_.size(); ++i) {
699 2 : log.printf(" %f", center_[i]);
700 : }
701 2 : log.printf("\n and scaling:");
702 4 : for (unsigned int i = 0; i < scale_.size(); ++i) {
703 2 : log.printf(" %f", scale_[i]);
704 : }
705 :
706 2 : log.printf("\n with initial ranges / rates:\n");
707 4 : for (unsigned int i = 0; i < max_coupling_range_.size(); ++i) {
708 2 : log.printf(" %f / %f\n", max_coupling_range_[i], max_coupling_grad_[i]);
709 : }
710 :
711 2 : if (!b_adaptive_ && update_period_ < 0) {
712 0 : log.printf(" ramping up coupling constants over %i steps\n", -update_period_);
713 : }
714 :
715 2 : if (b_mean) {
716 1 : log.printf("Loaded in averages for coupling constants...\n");
717 2 : for (unsigned int i = 0; i < current_coupling_.size(); ++i) {
718 1 : current_coupling_[i] = avg_bias[i] / N;
719 : }
720 2 : for (unsigned int i = 0; i < current_coupling_.size(); ++i) {
721 1 : set_coupling_[i] = avg_bias[i] / N;
722 : }
723 : }
724 :
725 2 : log.printf(" with current coupling constants:\n ");
726 4 : for (unsigned int i = 0; i < current_coupling_.size(); ++i) {
727 2 : log.printf(" %f", current_coupling_[i]);
728 : }
729 2 : log.printf("\n");
730 2 : log.printf(" with initial coupling constants:\n ");
731 4 : for (unsigned int i = 0; i < set_coupling_.size(); ++i) {
732 2 : log.printf(" %f", set_coupling_[i]);
733 : }
734 2 : log.printf("\n");
735 2 : log.printf(" and final coupling constants:\n ");
736 4 : for (unsigned int i = 0; i < target_coupling_.size(); ++i) {
737 2 : log.printf(" %f", target_coupling_[i]);
738 : }
739 2 : log.printf("\n");
740 :
741 2 : in_restart_.close();
742 2 : }
743 :
744 8 : void EDS::setupOutRestart() {
745 8 : out_restart_.link(*this);
746 8 : out_restart_.fmtField(fmt_);
747 8 : out_restart_.open(out_restart_name_);
748 : out_restart_.setHeavyFlush();
749 :
750 16 : out_restart_.addConstantField("adaptive").printField("adaptive", b_adaptive_);
751 16 : out_restart_.addConstantField("update_period").printField("update_period", update_period_);
752 16 : out_restart_.addConstantField("seed").printField("seed", seed_);
753 16 : out_restart_.addConstantField("kbt").printField("kbt", kbt_);
754 8 : }
755 :
756 27 : void EDS::writeOutRestart() {
757 : std::string cv_name;
758 27 : out_restart_.printField("time", getTimeStep() * getStep());
759 :
760 66 : for (unsigned int i = 0; i < ncvs_; ++i) {
761 : cv_name = getPntrToArgument(i)->getName();
762 78 : out_restart_.printField(cv_name + "_center", center_[i]);
763 78 : out_restart_.printField(cv_name + "_set", set_coupling_[i]);
764 78 : out_restart_.printField(cv_name + "_target", target_coupling_[i]);
765 78 : out_restart_.printField(cv_name + "_coupling", current_coupling_[i]);
766 78 : out_restart_.printField(cv_name + "_maxrange", max_coupling_range_[i]);
767 78 : out_restart_.printField(cv_name + "_maxgrad", max_coupling_grad_[i]);
768 78 : out_restart_.printField(cv_name + "_accum", coupling_accum_[i]);
769 39 : out_restart_.printField(cv_name + "_mean", means_[i]);
770 39 : if (b_virial_) {
771 18 : out_restart_.printField(cv_name + "_pseudovirial", pseudo_virial_[i]);
772 : }
773 39 : if (!b_covar_ && !b_lm_) {
774 42 : out_restart_.printField(cv_name + "_std", ssds_[i] / (fmax(1, update_calls_ - 1)));
775 : } else {
776 36 : out_restart_.printField(cv_name + "_std", covar_(i, i) / (fmax(1, update_calls_ - 1)));
777 : }
778 : }
779 27 : out_restart_.printField();
780 27 : }
781 :
782 40 : void EDS::calculate() {
783 :
784 : // get center values from action if necessary
785 40 : if (b_c_values_)
786 10 : for (unsigned int i = 0; i < ncvs_; ++i) {
787 5 : center_[i] = center_values_[i]->get();
788 : }
789 :
790 40 : apply_bias();
791 40 : }
792 :
793 40 : void EDS::apply_bias() {
794 : // Compute linear force as in "restraint"
795 : double ene = 0, totf2 = 0, cv, m, f;
796 :
797 100 : for (unsigned int i = 0; i < ncvs_; ++i) {
798 60 : cv = difference(i, center_[i], getArgument(i));
799 60 : m = current_coupling_[i];
800 60 : f = -m;
801 60 : ene += m * cv;
802 : setOutputForce(i, f);
803 60 : totf2 += f * f;
804 : }
805 :
806 : setBias(ene);
807 40 : value_force2_->set(totf2);
808 40 : }
809 :
810 12 : void EDS::update_statistics() {
811 : double s, N, w = 1.0;
812 12 : std::vector<double> deltas(ncvs_);
813 :
814 : // update weight max, if necessary
815 12 : if (b_weights_) {
816 2 : w = logweights_->getLogWeight();
817 2 : if (max_logweight_ < w) {
818 : // we have new max. Need to shift existing values
819 0 : wsum_ *= exp(max_logweight_ - w);
820 0 : max_logweight_ = w;
821 : }
822 : // convert to weight
823 2 : w = exp(w - max_logweight_);
824 2 : wsum_ += w;
825 : N = wsum_;
826 : } else {
827 10 : N = fmax(1, update_calls_);
828 : }
829 :
830 : // Welford, West, and Hanso online variance method
831 : // with weights (default = 1.0)
832 32 : for (unsigned int i = 0; i < ncvs_; ++i) {
833 20 : deltas[i] = difference(i, means_[i], getArgument(i)) * w;
834 20 : means_[i] += deltas[i] / N;
835 20 : if (!b_covar_ && !b_lm_) {
836 8 : ssds_[i] += deltas[i] * difference(i, means_[i], getArgument(i));
837 : }
838 : }
839 12 : if (b_covar_ || b_lm_) {
840 16 : for (unsigned int i = 0; i < ncvs_; ++i) {
841 36 : for (unsigned int j = i; j < ncvs_; ++j) {
842 24 : s = (N - 1) * deltas[i] * deltas[j] / N / N - covar_(i, j) / N;
843 24 : covar_(i, j) += s;
844 : // do this so we don't double count
845 24 : covar_(j, i) = covar_(i, j);
846 : }
847 : }
848 : }
849 12 : if (b_virial_) {
850 2 : update_pseudo_virial();
851 : }
852 12 : }
853 :
854 8 : void EDS::reset_statistics() {
855 20 : for (unsigned int i = 0; i < ncvs_; ++i) {
856 12 : means_[i] = 0;
857 12 : if (!b_covar_ && !b_lm_) {
858 6 : ssds_[i] = 0;
859 : }
860 : }
861 8 : if (b_covar_ || b_lm_)
862 8 : for (unsigned int i = 0; i < ncvs_; ++i)
863 24 : for (unsigned int j = 0; j < ncvs_; ++j) {
864 18 : covar_(i, j) = 0;
865 : }
866 8 : if (b_virial_) {
867 4 : for (unsigned int i = 0; i < ncvs_; ++i) {
868 3 : pseudo_virial_[i] = 0;
869 : }
870 1 : pseudo_virial_sum_ = 0;
871 : }
872 8 : if (b_weights_) {
873 2 : wsum_ = 0;
874 2 : max_logweight_ = 0;
875 : }
876 8 : }
877 :
878 1 : void EDS::calc_lm_step_size() {
879 : // calulcate step size
880 : // uses scale here, which by default is center
881 :
882 1 : mult(covar_, covar_, covar2_);
883 4 : for (unsigned int i = 0; i < ncvs_; ++i) {
884 3 : differences_[i] = difference(i, center_[i], means_[i]);
885 3 : covar2_[i][i] += lm_mixing_par_ * covar2_[i][i];
886 : }
887 :
888 : // "step_size_vec" = 2*inv(covar*covar+ lambda diag(covar*covar))*covar*(mean-center)
889 1 : mult(covar_, differences_, alpha_vector_);
890 1 : Invert(covar2_, lm_inv_);
891 1 : mult(lm_inv_, alpha_vector_, alpha_vector_2_);
892 :
893 4 : for (unsigned int i = 0; i < ncvs_; ++i) {
894 3 : step_size_[i] = 2 * alpha_vector_2_[i] / kbt_ / scale_[i];
895 : }
896 1 : }
897 :
898 1 : void EDS::calc_covar_step_size() {
899 : // calulcate step size
900 : // uses scale here, which by default is center
901 : double tmp;
902 4 : for (unsigned int i = 0; i < ncvs_; ++i) {
903 : tmp = 0;
904 12 : for (unsigned int j = 0; j < ncvs_; ++j) {
905 9 : tmp += difference(i, center_[i], means_[i]) * covar_(i, j);
906 : }
907 3 : step_size_[i] = 2 * tmp / kbt_ / scale_[i] * update_calls_ / fmax(1, update_calls_ - 1);
908 : }
909 1 : }
910 :
911 6 : void EDS::calc_ssd_step_size() {
912 : double tmp;
913 12 : for (unsigned int i = 0; i < ncvs_; ++i) {
914 6 : tmp = 2. * difference(i, center_[i], means_[i]) * ssds_[i] / fmax(1, update_calls_ - 1);
915 6 : step_size_[i] = tmp / kbt_ / scale_[i];
916 : }
917 6 : }
918 :
919 2 : void EDS::update_pseudo_virial() {
920 : // We want to compute the bias force on each atom times the position
921 : // of the atoms.
922 : double p, netp = 0, netpv = 0;
923 : double volume = 0;
924 8 : for (unsigned int i = 0; i < ncvs_; ++i) {
925 : // checked in setup to ensure this cast is valid.
926 6 : ActionAtomistic *cv = dynamic_cast<ActionAtomistic *>(getPntrToArgument(i)->getPntrToAction());
927 : Tensor &v(cv->modifyVirial());
928 6 : Tensor box(cv->getBox());
929 : const unsigned int natoms = cv->getNumberOfAtoms();
930 6 : if (!volume) {
931 2 : volume = box.determinant();
932 : }
933 :
934 : // pressure contribution is -dBias / dV
935 : // dBias / dV = alpha / w * dCV / dV
936 : // to get partial of CV wrt to volume
937 : // dCV/dV = sum dCV/dvij * vij / V
938 : // where vij is box element
939 : // add diagonal of virial tensor to get net pressure
940 : // TODO: replace this with adjugate (Jacobi's Formula) for non-orthorombic case(?)
941 6 : p = v(0, 0) * box(0, 0) + v(1, 1) * box(1, 1) + v(2, 2) * box(2, 2);
942 6 : p /= volume;
943 :
944 6 : netp += p;
945 :
946 : // now scale for correct units in EDS algorithm
947 6 : p *= (volume) / (kbt_ * natoms);
948 :
949 : // compute running mean of scaled
950 6 : if (set_coupling_[i] != 0) {
951 0 : pseudo_virial_[i] = (p - pseudo_virial_[i]) / (fmax(1, update_calls_));
952 : } else {
953 6 : pseudo_virial_[i] = 0;
954 : }
955 : // update net pressure
956 6 : netpv += pseudo_virial_[i];
957 : }
958 : // update pressure
959 2 : value_pressure_->set(netp);
960 2 : pseudo_virial_sum_ = netpv;
961 2 : }
962 :
963 8 : void EDS::update_bias() {
964 8 : log.flush();
965 8 : if (b_lm_) {
966 1 : calc_lm_step_size();
967 7 : } else if (b_covar_) {
968 1 : calc_covar_step_size();
969 : } else {
970 6 : calc_ssd_step_size();
971 : }
972 :
973 20 : for (unsigned int i = 0; i < ncvs_; ++i) {
974 :
975 : // multidimesional stochastic step
976 12 : if (ncvs_ == 1 || (rand_.RandU01() < (multi_prop_))) {
977 :
978 12 : if (b_virial_) {
979 : // apply virial regularization
980 : // P * dP/dcoupling
981 : // coupling is already included in virial term due to plumed propogating from bias to forces
982 : // thus we need to divide by it to get the derivative (since force is linear in coupling)
983 3 : if (fabs(set_coupling_[i]) > 0.000000001) // my heuristic for if EDS has started to prevent / 0
984 : // scale^2 here is to align units
985 : {
986 0 : step_size_[i] -= 2 * scale_[i] * scale_[i] * virial_scaling_ * pseudo_virial_sum_ * pseudo_virial_sum_ / set_coupling_[i];
987 : }
988 : }
989 12 : if (step_size_[i] == 0) {
990 4 : continue;
991 : }
992 :
993 : // clip gradient
994 8 : step_size_[i] = copysign(fmin(fabs(step_size_[i]), max_coupling_grad_[i]), step_size_[i]);
995 8 : coupling_accum_[i] += step_size_[i] * step_size_[i];
996 :
997 : // equation 5 in White and Voth, JCTC 2014
998 : // no negative sign because it's in step_size
999 8 : set_coupling_[i] += step_size_[i] * max_coupling_range_[i] / sqrt(coupling_accum_[i]);
1000 8 : coupling_rate_[i] = (set_coupling_[i] - current_coupling_[i]) / update_period_;
1001 : } else {
1002 : // do not change the bias
1003 0 : coupling_rate_[i] = 0;
1004 : }
1005 : }
1006 :
1007 : // reset means/vars
1008 8 : reset_statistics();
1009 8 : }
1010 :
1011 40 : void EDS::update() {
1012 : // adjust parameters according to EDS recipe
1013 40 : update_calls_++;
1014 :
1015 : // if we aren't wating for the bias to equilibrate, set flag to collect data
1016 : // want statistics before writing restart
1017 40 : if (!b_equil_ && update_period_ > 0) {
1018 12 : update_statistics();
1019 : }
1020 :
1021 : // write restart with correct statistics before bias update
1022 : // check if we're ramping or doing normal updates and then restart if needed. The ramping check
1023 : // is complicated because we could be frozen, finished ramping or not ramping.
1024 : // The + 2 is so we have an extra line showing that the bias isn't changing (for my sanity and yours)
1025 40 : if (b_write_restart_) {
1026 40 : if (getStep() == 0 ||
1027 32 : ((update_period_ < 0 && !b_freeze_ && update_calls_ <= fabs(update_period_) + 2) ||
1028 24 : (update_period_ > 0 && update_calls_ % update_period_ == 0))) {
1029 27 : writeOutRestart();
1030 : }
1031 : }
1032 :
1033 : int b_finished_equil_flag = 1;
1034 :
1035 : // assume forces already applied and saved
1036 : // are we ramping to a constant value and not done equilibrating?
1037 40 : if (update_period_ < 0) {
1038 5 : if (update_calls_ <= fabs(update_period_) && !b_freeze_) {
1039 4 : for (unsigned int i = 0; i < ncvs_; ++i) {
1040 2 : current_coupling_[i] += (target_coupling_[i] - set_coupling_[i]) / fabs(update_period_);
1041 : }
1042 : }
1043 : // make sure we don't reset update calls
1044 : b_finished_equil_flag = 0;
1045 35 : } else if (update_period_ == 0) {
1046 : // do we have a no-update case?
1047 : // not updating
1048 : // pass
1049 30 : } else if (b_equil_) {
1050 : // equilibrating
1051 : // check if we've reached the setpoint
1052 48 : for (unsigned int i = 0; i < ncvs_; ++i) {
1053 30 : if (coupling_rate_[i] == 0 || pow(current_coupling_[i] - set_coupling_[i], 2) < pow(coupling_rate_[i], 2)) {
1054 14 : b_finished_equil_flag &= 1;
1055 : } else {
1056 16 : current_coupling_[i] += coupling_rate_[i];
1057 : b_finished_equil_flag = 0;
1058 : }
1059 : }
1060 : }
1061 :
1062 : // reduce all the flags
1063 40 : if (b_equil_ && b_finished_equil_flag) {
1064 11 : b_equil_ = false;
1065 11 : update_calls_ = 0;
1066 : }
1067 :
1068 : // Now we update coupling constant, if necessary
1069 40 : if (!b_equil_ && update_period_ > 0 && update_calls_ == update_period_ && !b_freeze_) {
1070 8 : update_bias();
1071 8 : update_calls_ = 0;
1072 8 : avg_coupling_count_++;
1073 8 : b_equil_ = true; // back to equilibration now
1074 : } // close update if
1075 :
1076 : // pass couplings out so they are accessible
1077 100 : for (unsigned int i = 0; i < ncvs_; ++i) {
1078 60 : out_coupling_[i]->set(current_coupling_[i]);
1079 : }
1080 40 : }
1081 :
1082 16 : EDS::~EDS() {
1083 8 : out_restart_.close();
1084 24 : }
1085 :
1086 0 : void EDS::turnOnDerivatives() {
1087 : // do nothing
1088 : // this is to avoid errors triggered when a bias is used as a CV
1089 : // (This is done in ExtendedLagrangian.cpp)
1090 0 : }
1091 :
1092 : }
1093 : } // close the 2 namespaces
|
