src/gromacs/mdlib/expanded.cpp

   1 /*
   2  * This file is part of the GROMACS molecular simulation package.
   3  *
   4  * Copyright (c) 2012,2013,2014,2015,2016,2017,2018, by the GROMACS development team, led by
   5  * Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl,
   6  * and including many others, as listed in the AUTHORS file in the
   7  * top-level source directory and at http://www.gromacs.org.
   8  *
   9  * GROMACS is free software; you can redistribute it and/or
  10  * modify it under the terms of the GNU Lesser General Public License
  11  * as published by the Free Software Foundation; either version 2.1
  12  * of the License, or (at your option) any later version.
  13  *
  14  * GROMACS is distributed in the hope that it will be useful,
  15  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  16  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  17  * Lesser General Public License for more details.
  18  *
  19  * You should have received a copy of the GNU Lesser General Public
  20  * License along with GROMACS; if not, see
  21  * http://www.gnu.org/licenses, or write to the Free Software Foundation,
  22  * Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA.
  23  *
  24  * If you want to redistribute modifications to GROMACS, please
  25  * consider that scientific software is very special. Version
  26  * control is crucial - bugs must be traceable. We will be happy to
  27  * consider code for inclusion in the official distribution, but
  28  * derived work must not be called official GROMACS. Details are found
  29  * in the README & COPYING files - if they are missing, get the
  30  * official version at http://www.gromacs.org.
  31  *
  32  * To help us fund GROMACS development, we humbly ask that you cite
  33  * the research papers on the package. Check out http://www.gromacs.org.
  34  */
  35 #include "gmxpre.h"
  36
  37 #include "expanded.h"
  38
  39 #include <stdio.h>
  40
  41 #include <cmath>
  42
  43 #include <algorithm>
  44
  45 #include "gromacs/domdec/domdec.h"
  46 #include "gromacs/fileio/confio.h"
  47 #include "gromacs/fileio/gmxfio.h"
  48 #include "gromacs/fileio/xtcio.h"
  49 #include "gromacs/gmxlib/chargegroup.h"
  50 #include "gromacs/gmxlib/network.h"
  51 #include "gromacs/gmxlib/nrnb.h"
  52 #include "gromacs/listed-forces/disre.h"
  53 #include "gromacs/listed-forces/orires.h"
  54 #include "gromacs/math/functions.h"
  55 #include "gromacs/math/units.h"
  56 #include "gromacs/math/vec.h"
  57 #include "gromacs/mdlib/calcmu.h"
  58 #include "gromacs/mdlib/constr.h"
  59 #include "gromacs/mdlib/force.h"
  60 #include "gromacs/mdlib/update.h"
  61 #include "gromacs/mdtypes/enerdata.h"
  62 #include "gromacs/mdtypes/forcerec.h"
  63 #include "gromacs/mdtypes/inputrec.h"
  64 #include "gromacs/mdtypes/md_enums.h"
  65 #include "gromacs/mdtypes/mdatom.h"
  66 #include "gromacs/mdtypes/state.h"
  67 #include "gromacs/random/threefry.h"
  68 #include "gromacs/random/uniformrealdistribution.h"
  69 #include "gromacs/timing/wallcycle.h"
  70 #include "gromacs/utility/fatalerror.h"
  71 #include "gromacs/utility/gmxmpi.h"
  72 #include "gromacs/utility/smalloc.h"
  73
  74 static void init_df_history_weights(df_history_t *dfhist, t_expanded *expand, int nlim)
  75 {
  76     int i;
  77     dfhist->wl_delta = expand->init_wl_delta;
  78     for (i = 0; i < nlim; i++)
  79     {
  80         dfhist->sum_weights[i] = expand->init_lambda_weights[i];
  81         dfhist->sum_dg[i]      = expand->init_lambda_weights[i];
  82     }
  83 }
  84
  85 /* Eventually should contain all the functions needed to initialize expanded ensemble
  86    before the md loop starts */
  87 void init_expanded_ensemble(gmx_bool bStateFromCP, t_inputrec *ir, df_history_t *dfhist)
  88 {
  89     if (!bStateFromCP)
  90     {
  91         init_df_history_weights(dfhist, ir->expandedvals, ir->fepvals->n_lambda);
  92     }
  93 }
  94
  95 static void GenerateGibbsProbabilities(real *ene, double *p_k, double *pks, int minfep, int maxfep)
  96 {
  97
  98     int  i;
  99     real maxene;
 100
 101     *pks   = 0.0;
 102     maxene = ene[minfep];
 103     /* find the maximum value */
 104     for (i = minfep; i <= maxfep; i++)
 105     {
 106         if (ene[i] > maxene)
 107         {
 108             maxene = ene[i];
 109         }
 110     }
 111     /* find the denominator */
 112     for (i = minfep; i <= maxfep; i++)
 113     {
 114         *pks += std::exp(ene[i]-maxene);
 115     }
 116     /*numerators*/
 117     for (i = minfep; i <= maxfep; i++)
 118     {
 119         p_k[i] = std::exp(ene[i]-maxene) / *pks;
 120     }
 121 }
 122
 123 static void GenerateWeightedGibbsProbabilities(const real *ene, double *p_k, double *pks, int nlim, real *nvals, real delta)
 124 {
 125
 126     int   i;
 127     real  maxene;
 128     real *nene;
 129     *pks = 0.0;
 130
 131     snew(nene, nlim);
 132     for (i = 0; i < nlim; i++)
 133     {
 134         if (nvals[i] == 0)
 135         {
 136             /* add the delta, since we need to make sure it's greater than zero, and
 137                we need a non-arbitrary number? */
 138             nene[i] = ene[i] + std::log(nvals[i]+delta);
 139         }
 140         else
 141         {
 142             nene[i] = ene[i] + std::log(nvals[i]);
 143         }
 144     }
 145
 146     /* find the maximum value */
 147     maxene = nene[0];
 148     for (i = 0; i < nlim; i++)
 149     {
 150         if (nene[i] > maxene)
 151         {
 152             maxene = nene[i];
 153         }
 154     }
 155
 156     /* subtract off the maximum, avoiding overflow */
 157     for (i = 0; i < nlim; i++)
 158     {
 159         nene[i] -= maxene;
 160     }
 161
 162     /* find the denominator */
 163     for (i = 0; i < nlim; i++)
 164     {
 165         *pks += std::exp(nene[i]);
 166     }
 167
 168     /*numerators*/
 169     for (i = 0; i < nlim; i++)
 170     {
 171         p_k[i] = std::exp(nene[i]) / *pks;
 172     }
 173     sfree(nene);
 174 }
 175
 176 static int FindMinimum(const real *min_metric, int N)
 177 {
 178
 179     real min_val;
 180     int  min_nval, nval;
 181
 182     min_nval = 0;
 183     min_val  = min_metric[0];
 184
 185     for (nval = 0; nval < N; nval++)
 186     {
 187         if (min_metric[nval] < min_val)
 188         {
 189             min_val  = min_metric[nval];
 190             min_nval = nval;
 191         }
 192     }
 193     return min_nval;
 194 }
 195
 196 static gmx_bool CheckHistogramRatios(int nhisto, const real *histo, real ratio)
 197 {
 198
 199     int      i;
 200     real     nmean;
 201     gmx_bool bIfFlat;
 202
 203     nmean = 0;
 204     for (i = 0; i < nhisto; i++)
 205     {
 206         nmean += histo[i];
 207     }
 208
 209     if (nmean == 0)
 210     {
 211         /* no samples! is bad!*/
 212         bIfFlat = FALSE;
 213         return bIfFlat;
 214     }
 215     nmean /= (real)nhisto;
 216
 217     bIfFlat = TRUE;
 218     for (i = 0; i < nhisto; i++)
 219     {
 220         /* make sure that all points are in the ratio < x <  1/ratio range  */
 221         if (!((histo[i]/nmean < 1.0/ratio) && (histo[i]/nmean > ratio)))
 222         {
 223             bIfFlat = FALSE;
 224             break;
 225         }
 226     }
 227     return bIfFlat;
 228 }
 229
 230 static gmx_bool CheckIfDoneEquilibrating(int nlim, t_expanded *expand, df_history_t *dfhist, gmx_int64_t step)
 231 {
 232
 233     int      i, totalsamples;
 234     gmx_bool bDoneEquilibrating = TRUE;
 235     gmx_bool bIfFlat;
 236
 237     /* If we are doing slow growth to get initial values, we haven't finished equilibrating */
 238     if (expand->lmc_forced_nstart > 0)
 239     {
 240         for (i = 0; i < nlim; i++)
 241         {
 242             if (dfhist->n_at_lam[i] < expand->lmc_forced_nstart) /* we are still doing the initial sweep, so we're definitely not
 243                                                                     done equilibrating*/
 244             {
 245                 bDoneEquilibrating = FALSE;
 246                 break;
 247             }
 248         }
 249     }
 250     else
 251     {
 252         /* assume we have equilibrated the weights, then check to see if any of the conditions are not met */
 253         bDoneEquilibrating = TRUE;
 254
 255         /* calculate the total number of samples */
 256         switch (expand->elmceq)
 257         {
 258             case elmceqNO:
 259                 /* We have not equilibrated, and won't, ever. */
 260                 bDoneEquilibrating = FALSE;
 261                 break;
 262             case elmceqYES:
 263                 /* we have equilibrated -- we're done */
 264                 bDoneEquilibrating = TRUE;
 265                 break;
 266             case elmceqSTEPS:
 267                 /* first, check if we are equilibrating by steps, if we're still under */
 268                 if (step < expand->equil_steps)
 269                 {
 270                     bDoneEquilibrating = FALSE;
 271                 }
 272                 break;
 273             case elmceqSAMPLES:
 274                 totalsamples = 0;
 275                 for (i = 0; i < nlim; i++)
 276                 {
 277                     totalsamples += dfhist->n_at_lam[i];
 278                 }
 279                 if (totalsamples < expand->equil_samples)
 280                 {
 281                     bDoneEquilibrating = FALSE;
 282                 }
 283                 break;
 284             case elmceqNUMATLAM:
 285                 for (i = 0; i < nlim; i++)
 286                 {
 287                     if (dfhist->n_at_lam[i] < expand->equil_n_at_lam) /* we are still doing the initial sweep, so we're definitely not
 288                                                                          done equilibrating*/
 289                     {
 290                         bDoneEquilibrating  = FALSE;
 291                         break;
 292                     }
 293                 }
 294                 break;
 295             case elmceqWLDELTA:
 296                 if (EWL(expand->elamstats)) /* This check is in readir as well, but
 297                                                just to be sure */
 298                 {
 299                     if (dfhist->wl_delta > expand->equil_wl_delta)
 300                     {
 301                         bDoneEquilibrating = FALSE;
 302                     }
 303                 }
 304                 break;
 305             case elmceqRATIO:
 306                 /* we can use the flatness as a judge of good weights, as long as
 307                    we're not doing minvar, or Wang-Landau.
 308                    But turn off for now until we figure out exactly how we do this.
 309                  */
 310
 311                 if (!(EWL(expand->elamstats) || expand->elamstats == elamstatsMINVAR))
 312                 {
 313                     /* we want to use flatness -avoiding- the forced-through samples.  Plus, we need to convert to
 314                        floats for this histogram function. */
 315
 316                     real *modhisto;
 317                     snew(modhisto, nlim);
 318                     for (i = 0; i < nlim; i++)
 319                     {
 320                         modhisto[i] = 1.0*(dfhist->n_at_lam[i]-expand->lmc_forced_nstart);
 321                     }
 322                     bIfFlat = CheckHistogramRatios(nlim, modhisto, expand->equil_ratio);
 323                     sfree(modhisto);
 324                     if (!bIfFlat)
 325                     {
 326                         bDoneEquilibrating = FALSE;
 327                     }
 328                 }
 329                 break;
 330             default:
 331                 bDoneEquilibrating = TRUE;
 332                 break;
 333         }
 334     }
 335     return bDoneEquilibrating;
 336 }
 337
 338 static gmx_bool UpdateWeights(int nlim, t_expanded *expand, df_history_t *dfhist,
 339                               int fep_state, real *scaled_lamee, real *weighted_lamee, gmx_int64_t step)
 340 {
 341     gmx_bool  bSufficientSamples;
 342     int       i;
 343     int       n0, np1, nm1, nval, min_nvalm, min_nvalp, maxc;
 344     real      omega_m1_0, omega_p1_0, clam_osum;
 345     real      de, de_function;
 346     real      cnval, zero_sum_weights;
 347     real     *omegam_array, *weightsm_array, *omegap_array, *weightsp_array, *varm_array, *varp_array, *dwp_array, *dwm_array;
 348     real      clam_varm, clam_varp, clam_weightsm, clam_weightsp, clam_minvar;
 349     real     *lam_variance, *lam_dg;
 350     double   *p_k;
 351     double    pks = 0;
 352     real      chi_m1_0, chi_p1_0, chi_m2_0, chi_p2_0, chi_p1_m1, chi_p2_m1, chi_m1_p1, chi_m2_p1;
 353
 354     /* if we have equilibrated the weights, exit now */
 355     if (dfhist->bEquil)
 356     {
 357         return FALSE;
 358     }
 359
 360     if (CheckIfDoneEquilibrating(nlim, expand, dfhist, step))
 361     {
 362         dfhist->bEquil = TRUE;
 363         /* zero out the visited states so we know how many equilibrated states we have
 364            from here on out.*/
 365         for (i = 0; i < nlim; i++)
 366         {
 367             dfhist->n_at_lam[i] = 0;
 368         }
 369         return TRUE;
 370     }
 371
 372     /* If we reached this far, we have not equilibrated yet, keep on
 373        going resetting the weights */
 374
 375     if (EWL(expand->elamstats))
 376     {
 377         if (expand->elamstats == elamstatsWL)  /* Standard Wang-Landau */
 378         {
 379             dfhist->sum_weights[fep_state] -= dfhist->wl_delta;
 380             dfhist->wl_histo[fep_state]    += 1.0;
 381         }
 382         else if (expand->elamstats == elamstatsWWL) /* Weighted Wang-Landau */
 383         {
 384             snew(p_k, nlim);
 385
 386             /* first increment count */
 387             GenerateGibbsProbabilities(weighted_lamee, p_k, &pks, 0, nlim-1);
 388             for (i = 0; i < nlim; i++)
 389             {
 390                 dfhist->wl_histo[i] += (real)p_k[i];
 391             }
 392
 393             /* then increment weights (uses count) */
 394             pks = 0.0;
 395             GenerateWeightedGibbsProbabilities(weighted_lamee, p_k, &pks, nlim, dfhist->wl_histo, dfhist->wl_delta);
 396
 397             for (i = 0; i < nlim; i++)
 398             {
 399                 dfhist->sum_weights[i] -= dfhist->wl_delta*(real)p_k[i];
 400             }
 401             /* Alternate definition, using logarithms. Shouldn't make very much difference! */
 402             /*
 403                real di;
 404                for (i=0;i<nlim;i++)
 405                {
 406                 di = (real)1.0 + dfhist->wl_delta*(real)p_k[i];
 407                 dfhist->sum_weights[i] -= log(di);
 408                }
 409              */
 410             sfree(p_k);
 411         }
 412
 413         zero_sum_weights =  dfhist->sum_weights[0];
 414         for (i = 0; i < nlim; i++)
 415         {
 416             dfhist->sum_weights[i] -= zero_sum_weights;
 417         }
 418     }
 419
 420     if (expand->elamstats == elamstatsBARKER || expand->elamstats == elamstatsMETROPOLIS || expand->elamstats == elamstatsMINVAR)
 421     {
 422
 423         de_function = 0;  /* to get rid of warnings, but this value will not be used because of the logic */
 424         maxc        = 2*expand->c_range+1;
 425
 426         snew(lam_dg, nlim);
 427         snew(lam_variance, nlim);
 428
 429         snew(omegap_array, maxc);
 430         snew(weightsp_array, maxc);
 431         snew(varp_array, maxc);
 432         snew(dwp_array, maxc);
 433
 434         snew(omegam_array, maxc);
 435         snew(weightsm_array, maxc);
 436         snew(varm_array, maxc);
 437         snew(dwm_array, maxc);
 438
 439         /* unpack the current lambdas -- we will only update 2 of these */
 440
 441         for (i = 0; i < nlim-1; i++)
 442         {   /* only through the second to last */
 443             lam_dg[i]       = dfhist->sum_dg[i+1] - dfhist->sum_dg[i];
 444             lam_variance[i] = gmx::square(dfhist->sum_variance[i+1]) - gmx::square(dfhist->sum_variance[i]);
 445         }
 446
 447         /* accumulate running averages */
 448         for (nval = 0; nval < maxc; nval++)
 449         {
 450             /* constants for later use */
 451             cnval = (real)(nval-expand->c_range);
 452             /* actually, should be able to rewrite it w/o exponential, for better numerical stability */
 453             if (fep_state > 0)
 454             {
 455                 de = std::exp(cnval - (scaled_lamee[fep_state]-scaled_lamee[fep_state-1]));
 456                 if (expand->elamstats == elamstatsBARKER || expand->elamstats == elamstatsMINVAR)
 457                 {
 458                     de_function = 1.0/(1.0+de);
 459                 }
 460                 else if (expand->elamstats == elamstatsMETROPOLIS)
 461                 {
 462                     if (de < 1.0)
 463                     {
 464                         de_function = 1.0;
 465                     }
 466                     else
 467                     {
 468                         de_function = 1.0/de;
 469                     }
 470                 }
 471                 dfhist->accum_m[fep_state][nval]  += de_function;
 472                 dfhist->accum_m2[fep_state][nval] += de_function*de_function;
 473             }
 474
 475             if (fep_state < nlim-1)
 476             {
 477                 de = std::exp(-cnval + (scaled_lamee[fep_state+1]-scaled_lamee[fep_state]));
 478                 if (expand->elamstats == elamstatsBARKER || expand->elamstats == elamstatsMINVAR)
 479                 {
 480                     de_function = 1.0/(1.0+de);
 481                 }
 482                 else if (expand->elamstats == elamstatsMETROPOLIS)
 483                 {
 484                     if (de < 1.0)
 485                     {
 486                         de_function = 1.0;
 487                     }
 488                     else
 489                     {
 490                         de_function = 1.0/de;
 491                     }
 492                 }
 493                 dfhist->accum_p[fep_state][nval]  += de_function;
 494                 dfhist->accum_p2[fep_state][nval] += de_function*de_function;
 495             }
 496
 497             /* Metropolis transition and Barker transition (unoptimized Bennett) acceptance weight determination */
 498
 499             n0  = dfhist->n_at_lam[fep_state];
 500             if (fep_state > 0)
 501             {
 502                 nm1 = dfhist->n_at_lam[fep_state-1];
 503             }
 504             else
 505             {
 506                 nm1 = 0;
 507             }
 508             if (fep_state < nlim-1)
 509             {
 510                 np1 = dfhist->n_at_lam[fep_state+1];
 511             }
 512             else
 513             {
 514                 np1 = 0;
 515             }
 516
 517             /* logic SHOULD keep these all set correctly whatever the logic, but apparently it can't figure it out. */
 518             chi_m1_0 = chi_p1_0 = chi_m2_0 = chi_p2_0 = chi_p1_m1 = chi_p2_m1 = chi_m1_p1 = chi_m2_p1 = 0;
 519
 520             if (n0 > 0)
 521             {
 522                 chi_m1_0 = dfhist->accum_m[fep_state][nval]/n0;
 523                 chi_p1_0 = dfhist->accum_p[fep_state][nval]/n0;
 524                 chi_m2_0 = dfhist->accum_m2[fep_state][nval]/n0;
 525                 chi_p2_0 = dfhist->accum_p2[fep_state][nval]/n0;
 526             }
 527
 528             if ((fep_state > 0 ) && (nm1 > 0))
 529             {
 530                 chi_p1_m1 = dfhist->accum_p[fep_state-1][nval]/nm1;
 531                 chi_p2_m1 = dfhist->accum_p2[fep_state-1][nval]/nm1;
 532             }
 533
 534             if ((fep_state < nlim-1) && (np1 > 0))
 535             {
 536                 chi_m1_p1 = dfhist->accum_m[fep_state+1][nval]/np1;
 537                 chi_m2_p1 = dfhist->accum_m2[fep_state+1][nval]/np1;
 538             }
 539
 540             omega_m1_0    = 0;
 541             omega_p1_0    = 0;
 542             clam_weightsm = 0;
 543             clam_weightsp = 0;
 544             clam_varm     = 0;
 545             clam_varp     = 0;
 546
 547             if (fep_state > 0)
 548             {
 549                 if (n0 > 0)
 550                 {
 551                     omega_m1_0 = chi_m2_0/(chi_m1_0*chi_m1_0) - 1.0;
 552                     if (nm1 > 0)
 553                     {
 554                         real omega_p1_m1 = chi_p2_m1/(chi_p1_m1*chi_p1_m1) - 1.0;
 555                         clam_weightsm = (std::log(chi_m1_0) - std::log(chi_p1_m1)) + cnval;
 556                         clam_varm     = (1.0/n0)*(omega_m1_0) + (1.0/nm1)*(omega_p1_m1);
 557                     }
 558                 }
 559             }
 560
 561             if (fep_state < nlim-1)
 562             {
 563                 if (n0 > 0)
 564                 {
 565                     omega_p1_0 = chi_p2_0/(chi_p1_0*chi_p1_0) - 1.0;
 566                     if (np1 > 0)
 567                     {
 568                         real omega_m1_p1 = chi_m2_p1/(chi_m1_p1*chi_m1_p1) - 1.0;
 569                         clam_weightsp = (std::log(chi_m1_p1) - std::log(chi_p1_0)) + cnval;
 570                         clam_varp     = (1.0/np1)*(omega_m1_p1) + (1.0/n0)*(omega_p1_0);
 571                     }
 572                 }
 573             }
 574
 575             if (n0 > 0)
 576             {
 577                 omegam_array[nval]             = omega_m1_0;
 578             }
 579             else
 580             {
 581                 omegam_array[nval]             = 0;
 582             }
 583             weightsm_array[nval]           = clam_weightsm;
 584             varm_array[nval]               = clam_varm;
 585             if (nm1 > 0)
 586             {
 587                 dwm_array[nval]  = fabs( (cnval + std::log((1.0*n0)/nm1)) - lam_dg[fep_state-1] );
 588             }
 589             else
 590             {
 591                 dwm_array[nval]  = std::fabs( cnval - lam_dg[fep_state-1] );
 592             }
 593
 594             if (n0 > 0)
 595             {
 596                 omegap_array[nval]             = omega_p1_0;
 597             }
 598             else
 599             {
 600                 omegap_array[nval]             = 0;
 601             }
 602             weightsp_array[nval]           = clam_weightsp;
 603             varp_array[nval]               = clam_varp;
 604             if ((np1 > 0) && (n0 > 0))
 605             {
 606                 dwp_array[nval]  = fabs( (cnval + std::log((1.0*np1)/n0)) - lam_dg[fep_state] );
 607             }
 608             else
 609             {
 610                 dwp_array[nval]  = std::fabs( cnval - lam_dg[fep_state] );
 611             }
 612
 613         }
 614
 615         /* find the C's closest to the old weights value */
 616
 617         min_nvalm     = FindMinimum(dwm_array, maxc);
 618         omega_m1_0    = omegam_array[min_nvalm];
 619         clam_weightsm = weightsm_array[min_nvalm];
 620         clam_varm     = varm_array[min_nvalm];
 621
 622         min_nvalp     = FindMinimum(dwp_array, maxc);
 623         omega_p1_0    = omegap_array[min_nvalp];
 624         clam_weightsp = weightsp_array[min_nvalp];
 625         clam_varp     = varp_array[min_nvalp];
 626
 627         clam_osum   = omega_m1_0 + omega_p1_0;
 628         clam_minvar = 0;
 629         if (clam_osum > 0)
 630         {
 631             clam_minvar = 0.5*std::log(clam_osum);
 632         }
 633
 634         if (fep_state > 0)
 635         {
 636             lam_dg[fep_state-1]       = clam_weightsm;
 637             lam_variance[fep_state-1] = clam_varm;
 638         }
 639
 640         if (fep_state < nlim-1)
 641         {
 642             lam_dg[fep_state]       = clam_weightsp;
 643             lam_variance[fep_state] = clam_varp;
 644         }
 645
 646         if (expand->elamstats == elamstatsMINVAR)
 647         {
 648             bSufficientSamples = TRUE;
 649             /* make sure they are all past a threshold */
 650             for (i = 0; i < nlim; i++)
 651             {
 652                 if (dfhist->n_at_lam[i] < expand->minvarmin)
 653                 {
 654                     bSufficientSamples = FALSE;
 655                 }
 656             }
 657             if (bSufficientSamples)
 658             {
 659                 dfhist->sum_minvar[fep_state] = clam_minvar;
 660                 if (fep_state == 0)
 661                 {
 662                     for (i = 0; i < nlim; i++)
 663                     {
 664                         dfhist->sum_minvar[i] += (expand->minvar_const-clam_minvar);
 665                     }
 666                     expand->minvar_const          = clam_minvar;
 667                     dfhist->sum_minvar[fep_state] = 0.0;
 668                 }
 669                 else
 670                 {
 671                     dfhist->sum_minvar[fep_state] -= expand->minvar_const;
 672                 }
 673             }
 674         }
 675
 676         /* we need to rezero minvar now, since it could change at fep_state = 0 */
 677         dfhist->sum_dg[0]       = 0.0;
 678         dfhist->sum_variance[0] = 0.0;
 679         dfhist->sum_weights[0]  = dfhist->sum_dg[0] + dfhist->sum_minvar[0]; /* should be zero */
 680
 681         for (i = 1; i < nlim; i++)
 682         {
 683             dfhist->sum_dg[i]       = lam_dg[i-1] + dfhist->sum_dg[i-1];
 684             dfhist->sum_variance[i] = std::sqrt(lam_variance[i-1] + gmx::square(dfhist->sum_variance[i-1]));
 685             dfhist->sum_weights[i]  = dfhist->sum_dg[i] + dfhist->sum_minvar[i];
 686         }
 687
 688         sfree(lam_dg);
 689         sfree(lam_variance);
 690
 691         sfree(omegam_array);
 692         sfree(weightsm_array);
 693         sfree(varm_array);
 694         sfree(dwm_array);
 695
 696         sfree(omegap_array);
 697         sfree(weightsp_array);
 698         sfree(varp_array);
 699         sfree(dwp_array);
 700     }
 701     return FALSE;
 702 }
 703
 704 static int ChooseNewLambda(int nlim, t_expanded *expand, df_history_t *dfhist, int fep_state, real *weighted_lamee, double *p_k,
 705                            gmx_int64_t seed, gmx_int64_t step)
 706 {
 707     /* Choose new lambda value, and update transition matrix */
 708
 709     int                                  i, ifep, minfep, maxfep, lamnew, lamtrial, starting_fep_state;
 710     real                                 r1, r2, de, trialprob, tprob = 0;
 711     double                              *propose, *accept, *remainder;
 712     double                               pks;
 713     real                                 pnorm;
 714     gmx::ThreeFry2x64<0>                 rng(seed, gmx::RandomDomain::ExpandedEnsemble); // We only draw once, so zero bits internal counter is fine
 715     gmx::UniformRealDistribution<real>   dist;
 716
 717     starting_fep_state = fep_state;
 718     lamnew             = fep_state; /* so that there is a default setting -- stays the same */
 719
 720     if (!EWL(expand->elamstats))    /* ignore equilibrating the weights if using WL */
 721     {
 722         if ((expand->lmc_forced_nstart > 0) && (dfhist->n_at_lam[nlim-1] <= expand->lmc_forced_nstart))
 723         {
 724             /* Use a marching method to run through the lambdas and get preliminary free energy data,
 725                before starting 'free' sampling.  We start free sampling when we have enough at each lambda */
 726
 727             /* if we have enough at this lambda, move on to the next one */
 728
 729             if (dfhist->n_at_lam[fep_state] == expand->lmc_forced_nstart)
 730             {
 731                 lamnew = fep_state+1;
 732                 if (lamnew == nlim)  /* whoops, stepped too far! */
 733                 {
 734                     lamnew -= 1;
 735                 }
 736             }
 737             else
 738             {
 739                 lamnew = fep_state;
 740             }
 741             return lamnew;
 742         }
 743     }
 744
 745     snew(propose, nlim);
 746     snew(accept, nlim);
 747     snew(remainder, nlim);
 748
 749     for (i = 0; i < expand->lmc_repeats; i++)
 750     {
 751         rng.restart(step, i);
 752         dist.reset();
 753
 754         for (ifep = 0; ifep < nlim; ifep++)
 755         {
 756             propose[ifep] = 0;
 757             accept[ifep]  = 0;
 758         }
 759
 760         if ((expand->elmcmove == elmcmoveGIBBS) || (expand->elmcmove == elmcmoveMETGIBBS))
 761         {
 762             /* use the Gibbs sampler, with restricted range */
 763             if (expand->gibbsdeltalam < 0)
 764             {
 765                 minfep      = 0;
 766                 maxfep      = nlim-1;
 767             }
 768             else
 769             {
 770                 minfep = fep_state - expand->gibbsdeltalam;
 771                 maxfep = fep_state + expand->gibbsdeltalam;
 772                 if (minfep < 0)
 773                 {
 774                     minfep = 0;
 775                 }
 776                 if (maxfep > nlim-1)
 777                 {
 778                     maxfep = nlim-1;
 779                 }
 780             }
 781
 782             GenerateGibbsProbabilities(weighted_lamee, p_k, &pks, minfep, maxfep);
 783
 784             if (expand->elmcmove == elmcmoveGIBBS)
 785             {
 786                 for (ifep = minfep; ifep <= maxfep; ifep++)
 787                 {
 788                     propose[ifep] = p_k[ifep];
 789                     accept[ifep]  = 1.0;
 790                 }
 791                 /* Gibbs sampling */
 792                 r1 = dist(rng);
 793                 for (lamnew = minfep; lamnew <= maxfep; lamnew++)
 794                 {
 795                     if (r1 <= p_k[lamnew])
 796                     {
 797                         break;
 798                     }
 799                     r1 -= p_k[lamnew];
 800                 }
 801             }
 802             else if (expand->elmcmove == elmcmoveMETGIBBS)
 803             {
 804
 805                 /* Metropolized Gibbs sampling */
 806                 for (ifep = minfep; ifep <= maxfep; ifep++)
 807                 {
 808                     remainder[ifep] = 1 - p_k[ifep];
 809                 }
 810
 811                 /* find the proposal probabilities */
 812
 813                 if (remainder[fep_state] == 0)
 814                 {
 815                     /* only the current state has any probability */
 816                     /* we have to stay at the current state */
 817                     lamnew = fep_state;
 818                 }
 819                 else
 820                 {
 821                     for (ifep = minfep; ifep <= maxfep; ifep++)
 822                     {
 823                         if (ifep != fep_state)
 824                         {
 825                             propose[ifep] = p_k[ifep]/remainder[fep_state];
 826                         }
 827                         else
 828                         {
 829                             propose[ifep] = 0;
 830                         }
 831                     }
 832
 833                     r1 = dist(rng);
 834                     for (lamtrial = minfep; lamtrial <= maxfep; lamtrial++)
 835                     {
 836                         pnorm = p_k[lamtrial]/remainder[fep_state];
 837                         if (lamtrial != fep_state)
 838                         {
 839                             if (r1 <= pnorm)
 840                             {
 841                                 break;
 842                             }
 843                             r1 -= pnorm;
 844                         }
 845                     }
 846
 847                     /* we have now selected lamtrial according to p(lamtrial)/1-p(fep_state) */
 848                     tprob = 1.0;
 849                     /* trial probability is min{1,\frac{1 - p(old)}{1-p(new)} MRS 1/8/2008 */
 850                     trialprob = (remainder[fep_state])/(remainder[lamtrial]);
 851                     if (trialprob < tprob)
 852                     {
 853                         tprob = trialprob;
 854                     }
 855                     r2 = dist(rng);
 856                     if (r2 < tprob)
 857                     {
 858                         lamnew = lamtrial;
 859                     }
 860                     else
 861                     {
 862                         lamnew = fep_state;
 863                     }
 864                 }
 865
 866                 /* now figure out the acceptance probability for each */
 867                 for (ifep = minfep; ifep <= maxfep; ifep++)
 868                 {
 869                     tprob = 1.0;
 870                     if (remainder[ifep] != 0)
 871                     {
 872                         trialprob = (remainder[fep_state])/(remainder[ifep]);
 873                     }
 874                     else
 875                     {
 876                         trialprob = 1.0; /* this state is the only choice! */
 877                     }
 878                     if (trialprob < tprob)
 879                     {
 880                         tprob = trialprob;
 881                     }
 882                     /* probability for fep_state=0, but that's fine, it's never proposed! */
 883                     accept[ifep] = tprob;
 884                 }
 885             }
 886
 887             if (lamnew > maxfep)
 888             {
 889                 /* it's possible some rounding is failing */
 890                 if (gmx_within_tol(remainder[fep_state], 0, 50*GMX_DOUBLE_EPS))
 891                 {
 892                     /* numerical rounding error -- no state other than the original has weight */
 893                     lamnew = fep_state;
 894                 }
 895                 else
 896                 {
 897                     /* probably not a numerical issue */
 898                     int   loc    = 0;
 899                     int   nerror = 200+(maxfep-minfep+1)*60;
 900                     char *errorstr;
 901                     snew(errorstr, nerror);
 902                     /* if its greater than maxfep, then something went wrong -- probably underflow in the calculation
 903                        of sum weights. Generated detailed info for failure */
 904                     loc += sprintf(errorstr, "Something wrong in choosing new lambda state with a Gibbs move -- probably underflow in weight determination.\nDenominator is: %3d%17.10e\n  i                dE        numerator          weights\n", 0, pks);
 905                     for (ifep = minfep; ifep <= maxfep; ifep++)
 906                     {
 907                         loc += sprintf(&errorstr[loc], "%3d %17.10e%17.10e%17.10e\n", ifep, weighted_lamee[ifep], p_k[ifep], dfhist->sum_weights[ifep]);
 908                     }
 909                     gmx_fatal(FARGS, errorstr);
 910                 }
 911             }
 912         }
 913         else if ((expand->elmcmove == elmcmoveMETROPOLIS) || (expand->elmcmove == elmcmoveBARKER))
 914         {
 915             /* use the metropolis sampler with trial +/- 1 */
 916             r1 = dist(rng);
 917             if (r1 < 0.5)
 918             {
 919                 if (fep_state == 0)
 920                 {
 921                     lamtrial = fep_state;
 922                 }
 923                 else
 924                 {
 925                     lamtrial = fep_state-1;
 926                 }
 927             }
 928             else
 929             {
 930                 if (fep_state == nlim-1)
 931                 {
 932                     lamtrial = fep_state;
 933                 }
 934                 else
 935                 {
 936                     lamtrial = fep_state+1;
 937                 }
 938             }
 939
 940             de = weighted_lamee[lamtrial] - weighted_lamee[fep_state];
 941             if (expand->elmcmove == elmcmoveMETROPOLIS)
 942             {
 943                 tprob     = 1.0;
 944                 trialprob = std::exp(de);
 945                 if (trialprob < tprob)
 946                 {
 947                     tprob = trialprob;
 948                 }
 949                 propose[fep_state] = 0;
 950                 propose[lamtrial]  = 1.0; /* note that this overwrites the above line if fep_state = ntrial, which only occurs at the ends */
 951                 accept[fep_state]  = 1.0; /* doesn't actually matter, never proposed unless fep_state = ntrial, in which case it's 1.0 anyway */
 952                 accept[lamtrial]   = tprob;
 953
 954             }
 955             else if (expand->elmcmove == elmcmoveBARKER)
 956             {
 957                 tprob = 1.0/(1.0+std::exp(-de));
 958
 959                 propose[fep_state] = (1-tprob);
 960                 propose[lamtrial] += tprob; /* we add, to account for the fact that at the end, they might be the same point */
 961                 accept[fep_state]  = 1.0;
 962                 accept[lamtrial]   = 1.0;
 963             }
 964
 965             r2 = dist(rng);
 966             if (r2 < tprob)
 967             {
 968                 lamnew = lamtrial;
 969             }
 970             else
 971             {
 972                 lamnew = fep_state;
 973             }
 974         }
 975
 976         for (ifep = 0; ifep < nlim; ifep++)
 977         {
 978             dfhist->Tij[fep_state][ifep]      += propose[ifep]*accept[ifep];
 979             dfhist->Tij[fep_state][fep_state] += propose[ifep]*(1.0-accept[ifep]);
 980         }
 981         fep_state = lamnew;
 982     }
 983
 984     dfhist->Tij_empirical[starting_fep_state][lamnew] += 1.0;
 985
 986     sfree(propose);
 987     sfree(accept);
 988     sfree(remainder);
 989
 990     return lamnew;
 991 }
 992
 993 /* print out the weights to the log, along with current state */
 994 void PrintFreeEnergyInfoToFile(FILE *outfile, t_lambda *fep, t_expanded *expand, t_simtemp *simtemp, df_history_t *dfhist,
 995                                int fep_state, int frequency, gmx_int64_t step)
 996 {
 997     int         nlim, i, ifep, jfep;
 998     real        dw, dg, dv, Tprint;
 999     const char *print_names[efptNR] = {" FEPL", "MassL", "CoulL", " VdwL", "BondL", "RestT", "Temp.(K)"};
1000     gmx_bool    bSimTemp            = FALSE;
1001
1002     nlim = fep->n_lambda;
1003     if (simtemp != nullptr)
1004     {
1005         bSimTemp = TRUE;
1006     }
1007
1008     if (step % frequency == 0)
1009     {
1010         fprintf(outfile, "             MC-lambda information\n");
1011         if (EWL(expand->elamstats) && (!(dfhist->bEquil)))
1012         {
1013             fprintf(outfile, "  Wang-Landau incrementor is: %11.5g\n", dfhist->wl_delta);
1014         }
1015         fprintf(outfile, "  N");
1016         for (i = 0; i < efptNR; i++)
1017         {
1018             if (fep->separate_dvdl[i])
1019             {
1020                 fprintf(outfile, "%7s", print_names[i]);
1021             }
1022             else if ((i == efptTEMPERATURE) && bSimTemp)
1023             {
1024                 fprintf(outfile, "%10s", print_names[i]); /* more space for temperature formats */
1025             }
1026         }
1027         fprintf(outfile, "    Count   ");
1028         if (expand->elamstats == elamstatsMINVAR)
1029         {
1030             fprintf(outfile, "W(in kT)   G(in kT)  dG(in kT)  dV(in kT)\n");
1031         }
1032         else
1033         {
1034             fprintf(outfile, "G(in kT)  dG(in kT)\n");
1035         }
1036         for (ifep = 0; ifep < nlim; ifep++)
1037         {
1038             if (ifep == nlim-1)
1039             {
1040                 dw = 0.0;
1041                 dg = 0.0;
1042                 dv = 0.0;
1043             }
1044             else
1045             {
1046                 dw = dfhist->sum_weights[ifep+1] - dfhist->sum_weights[ifep];
1047                 dg = dfhist->sum_dg[ifep+1] - dfhist->sum_dg[ifep];
1048                 dv = std::sqrt(gmx::square(dfhist->sum_variance[ifep+1]) - gmx::square(dfhist->sum_variance[ifep]));
1049             }
1050             fprintf(outfile, "%3d", (ifep+1));
1051             for (i = 0; i < efptNR; i++)
1052             {
1053                 if (fep->separate_dvdl[i])
1054                 {
1055                     fprintf(outfile, "%7.3f", fep->all_lambda[i][ifep]);
1056                 }
1057                 else if (i == efptTEMPERATURE && bSimTemp)
1058                 {
1059                     fprintf(outfile, "%9.3f", simtemp->temperatures[ifep]);
1060                 }
1061             }
1062             if (EWL(expand->elamstats) && (!(dfhist->bEquil)))  /* if performing WL and still haven't equilibrated */
1063             {
1064                 if (expand->elamstats == elamstatsWL)
1065                 {
1066                     fprintf(outfile, " %8d", (int)dfhist->wl_histo[ifep]);
1067                 }
1068                 else
1069                 {
1070                     fprintf(outfile, " %8.3f", dfhist->wl_histo[ifep]);
1071                 }
1072             }
1073             else   /* we have equilibrated weights */
1074             {
1075                 fprintf(outfile, " %8d", dfhist->n_at_lam[ifep]);
1076             }
1077             if (expand->elamstats == elamstatsMINVAR)
1078             {
1079                 fprintf(outfile, " %10.5f %10.5f %10.5f %10.5f", dfhist->sum_weights[ifep], dfhist->sum_dg[ifep], dg, dv);
1080             }
1081             else
1082             {
1083                 fprintf(outfile, " %10.5f %10.5f", dfhist->sum_weights[ifep], dw);
1084             }
1085             if (ifep == fep_state)
1086             {
1087                 fprintf(outfile, " <<\n");
1088             }
1089             else
1090             {
1091                 fprintf(outfile, "   \n");
1092             }
1093         }
1094         fprintf(outfile, "\n");
1095
1096         if ((step % expand->nstTij == 0) && (expand->nstTij > 0) && (step > 0))
1097         {
1098             fprintf(outfile, "                     Transition Matrix\n");
1099             for (ifep = 0; ifep < nlim; ifep++)
1100             {
1101                 fprintf(outfile, "%12d", (ifep+1));
1102             }
1103             fprintf(outfile, "\n");
1104             for (ifep = 0; ifep < nlim; ifep++)
1105             {
1106                 for (jfep = 0; jfep < nlim; jfep++)
1107                 {
1108                     if (dfhist->n_at_lam[ifep] > 0)
1109                     {
1110                         if (expand->bSymmetrizedTMatrix)
1111                         {
1112                             Tprint = (dfhist->Tij[ifep][jfep]+dfhist->Tij[jfep][ifep])/(dfhist->n_at_lam[ifep]+dfhist->n_at_lam[jfep]);
1113                         }
1114                         else
1115                         {
1116                             Tprint = (dfhist->Tij[ifep][jfep])/(dfhist->n_at_lam[ifep]);
1117                         }
1118                     }
1119                     else
1120                     {
1121                         Tprint = 0.0;
1122                     }
1123                     fprintf(outfile, "%12.8f", Tprint);
1124                 }
1125                 fprintf(outfile, "%3d\n", (ifep+1));
1126             }
1127
1128             fprintf(outfile, "                  Empirical Transition Matrix\n");
1129             for (ifep = 0; ifep < nlim; ifep++)
1130             {
1131                 fprintf(outfile, "%12d", (ifep+1));
1132             }
1133             fprintf(outfile, "\n");
1134             for (ifep = 0; ifep < nlim; ifep++)
1135             {
1136                 for (jfep = 0; jfep < nlim; jfep++)
1137                 {
1138                     if (dfhist->n_at_lam[ifep] > 0)
1139                     {
1140                         if (expand->bSymmetrizedTMatrix)
1141                         {
1142                             Tprint = (dfhist->Tij_empirical[ifep][jfep]+dfhist->Tij_empirical[jfep][ifep])/(dfhist->n_at_lam[ifep]+dfhist->n_at_lam[jfep]);
1143                         }
1144                         else
1145                         {
1146                             Tprint = dfhist->Tij_empirical[ifep][jfep]/(dfhist->n_at_lam[ifep]);
1147                         }
1148                     }
1149                     else
1150                     {
1151                         Tprint = 0.0;
1152                     }
1153                     fprintf(outfile, "%12.8f", Tprint);
1154                 }
1155                 fprintf(outfile, "%3d\n", (ifep+1));
1156             }
1157         }
1158     }
1159 }
1160
1161 int ExpandedEnsembleDynamics(FILE *log, t_inputrec *ir, gmx_enerdata_t *enerd,
1162                              t_state *state, t_extmass *MassQ, int fep_state, df_history_t *dfhist,
1163                              gmx_int64_t step,
1164                              rvec *v, t_mdatoms *mdatoms)
1165 /* Note that the state variable is only needed for simulated tempering, not
1166    Hamiltonian expanded ensemble.  May be able to remove it after integrator refactoring. */
1167 {
1168     real       *pfep_lamee, *scaled_lamee, *weighted_lamee;
1169     double     *p_k;
1170     int         i, nlim, lamnew, totalsamples;
1171     real        oneovert, maxscaled = 0, maxweighted = 0;
1172     t_expanded *expand;
1173     t_simtemp  *simtemp;
1174     gmx_bool    bIfReset, bSwitchtoOneOverT, bDoneEquilibrating = FALSE;
1175
1176     expand  = ir->expandedvals;
1177     simtemp = ir->simtempvals;
1178     nlim    = ir->fepvals->n_lambda;
1179
1180     snew(scaled_lamee, nlim);
1181     snew(weighted_lamee, nlim);
1182     snew(pfep_lamee, nlim);
1183     snew(p_k, nlim);
1184
1185     /* update the count at the current lambda*/
1186     dfhist->n_at_lam[fep_state]++;
1187
1188     /* need to calculate the PV term somewhere, but not needed here? Not until there's a lambda state that's
1189        pressure controlled.*/
1190     /*
1191        pVTerm = 0;
1192        where does this PV term go?
1193        for (i=0;i<nlim;i++)
1194        {
1195        fep_lamee[i] += pVTerm;
1196        }
1197      */
1198
1199     /* determine the minimum value to avoid overflow.  Probably a better way to do this */
1200     /* we don't need to include the pressure term, since the volume is the same between the two.
1201        is there some term we are neglecting, however? */
1202
1203     if (ir->efep != efepNO)
1204     {
1205         for (i = 0; i < nlim; i++)
1206         {
1207             if (ir->bSimTemp)
1208             {
1209                 /* Note -- this assumes no mass changes, since kinetic energy is not added  . . . */
1210                 scaled_lamee[i] = (enerd->enerpart_lambda[i+1]-enerd->enerpart_lambda[0])/(simtemp->temperatures[i]*BOLTZ)
1211                     + enerd->term[F_EPOT]*(1.0/(simtemp->temperatures[i])- 1.0/(simtemp->temperatures[fep_state]))/BOLTZ;
1212             }
1213             else
1214             {
1215                 scaled_lamee[i] = (enerd->enerpart_lambda[i+1]-enerd->enerpart_lambda[0])/(expand->mc_temp*BOLTZ);
1216                 /* mc_temp is currently set to the system reft unless otherwise defined */
1217             }
1218
1219             /* save these energies for printing, so they don't get overwritten by the next step */
1220             /* they aren't overwritten in the non-free energy case, but we always print with these
1221                for simplicity */
1222         }
1223     }
1224     else
1225     {
1226         if (ir->bSimTemp)
1227         {
1228             for (i = 0; i < nlim; i++)
1229             {
1230                 scaled_lamee[i] = enerd->term[F_EPOT]*(1.0/simtemp->temperatures[i] - 1.0/simtemp->temperatures[fep_state])/BOLTZ;
1231             }
1232         }
1233     }
1234
1235     for (i = 0; i < nlim; i++)
1236     {
1237         pfep_lamee[i] = scaled_lamee[i];
1238
1239         weighted_lamee[i] = dfhist->sum_weights[i] - scaled_lamee[i];
1240         if (i == 0)
1241         {
1242             maxscaled   = scaled_lamee[i];
1243             maxweighted = weighted_lamee[i];
1244         }
1245         else
1246         {
1247             if (scaled_lamee[i] > maxscaled)
1248             {
1249                 maxscaled = scaled_lamee[i];
1250             }
1251             if (weighted_lamee[i] > maxweighted)
1252             {
1253                 maxweighted = weighted_lamee[i];
1254             }
1255         }
1256     }
1257
1258     for (i = 0; i < nlim; i++)
1259     {
1260         scaled_lamee[i]   -= maxscaled;
1261         weighted_lamee[i] -= maxweighted;
1262     }
1263
1264     /* update weights - we decide whether or not to actually do this inside */
1265
1266     bDoneEquilibrating = UpdateWeights(nlim, expand, dfhist, fep_state, scaled_lamee, weighted_lamee, step);
1267     if (bDoneEquilibrating)
1268     {
1269         if (log)
1270         {
1271             fprintf(log, "\nStep %d: Weights have equilibrated, using criteria: %s\n", (int)step, elmceq_names[expand->elmceq]);
1272         }
1273     }
1274
1275     lamnew = ChooseNewLambda(nlim, expand, dfhist, fep_state, weighted_lamee, p_k,
1276                              ir->expandedvals->lmc_seed, step);
1277     /* if using simulated tempering, we need to adjust the temperatures */
1278     if (ir->bSimTemp && (lamnew != fep_state)) /* only need to change the temperatures if we change the state */
1279     {
1280         int   i, j, n, d;
1281         real *buf_ngtc;
1282         real  told;
1283         int   nstart, nend, gt;
1284
1285         snew(buf_ngtc, ir->opts.ngtc);
1286
1287         for (i = 0; i < ir->opts.ngtc; i++)
1288         {
1289             if (ir->opts.ref_t[i] > 0)
1290             {
1291                 told              = ir->opts.ref_t[i];
1292                 ir->opts.ref_t[i] =  simtemp->temperatures[lamnew];
1293                 buf_ngtc[i]       = std::sqrt(ir->opts.ref_t[i]/told); /* using the buffer as temperature scaling */
1294             }
1295         }
1296
1297         /* we don't need to manipulate the ekind information, as it isn't due to be reset until the next step anyway */
1298
1299         nstart = 0;
1300         nend   = mdatoms->homenr;
1301         for (n = nstart; n < nend; n++)
1302         {
1303             gt = 0;
1304             if (mdatoms->cTC)
1305             {
1306                 gt = mdatoms->cTC[n];
1307             }
1308             for (d = 0; d < DIM; d++)
1309             {
1310                 v[n][d] *= buf_ngtc[gt];
1311             }
1312         }
1313
1314         if (inputrecNptTrotter(ir) || inputrecNphTrotter(ir) || inputrecNvtTrotter(ir))
1315         {
1316             /* we need to recalculate the masses if the temperature has changed */
1317             init_npt_masses(ir, state, MassQ, FALSE);
1318             for (i = 0; i < state->nnhpres; i++)
1319             {
1320                 for (j = 0; j < ir->opts.nhchainlength; j++)
1321                 {
1322                     state->nhpres_vxi[i+j] *= buf_ngtc[i];
1323                 }
1324             }
1325             for (i = 0; i < ir->opts.ngtc; i++)
1326             {
1327                 for (j = 0; j < ir->opts.nhchainlength; j++)
1328                 {
1329                     state->nosehoover_vxi[i+j] *= buf_ngtc[i];
1330                 }
1331             }
1332         }
1333         sfree(buf_ngtc);
1334     }
1335
1336     /* now check on the Wang-Landau updating critera */
1337
1338     if (EWL(expand->elamstats))
1339     {
1340         bSwitchtoOneOverT = FALSE;
1341         if (expand->bWLoneovert)
1342         {
1343             totalsamples = 0;
1344             for (i = 0; i < nlim; i++)
1345             {
1346                 totalsamples += dfhist->n_at_lam[i];
1347             }
1348             oneovert = (1.0*nlim)/totalsamples;
1349             /* oneovert has decreasd by a bit since last time, so we actually make sure its within one of this number */
1350             /* switch to 1/t incrementing when wl_delta has decreased at least once, and wl_delta is now less than 1/t */
1351             if ((dfhist->wl_delta <= ((totalsamples)/(totalsamples-1.00001))*oneovert) &&
1352                 (dfhist->wl_delta < expand->init_wl_delta))
1353             {
1354                 bSwitchtoOneOverT = TRUE;
1355             }
1356         }
1357         if (bSwitchtoOneOverT)
1358         {
1359             dfhist->wl_delta = oneovert; /* now we reduce by this each time, instead of only at flatness */
1360         }
1361         else
1362         {
1363             bIfReset = CheckHistogramRatios(nlim, dfhist->wl_histo, expand->wl_ratio);
1364             if (bIfReset)
1365             {
1366                 for (i = 0; i < nlim; i++)
1367                 {
1368                     dfhist->wl_histo[i] = 0;
1369                 }
1370                 dfhist->wl_delta *= expand->wl_scale;
1371                 if (log)
1372                 {
1373                     fprintf(log, "\nStep %d: weights are now:", (int)step);
1374                     for (i = 0; i < nlim; i++)
1375                     {
1376                         fprintf(log, " %.5f", dfhist->sum_weights[i]);
1377                     }
1378                     fprintf(log, "\n");
1379                 }
1380             }
1381         }
1382     }
1383     sfree(pfep_lamee);
1384     sfree(scaled_lamee);
1385     sfree(weighted_lamee);
1386     sfree(p_k);
1387
1388     return lamnew;
1389 }