src/gromacs/applied_forces/awh/bias.cpp

   1 /*
   2  * This file is part of the GROMACS molecular simulation package.
   3  *
   4  * Copyright (c) 2015,2016,2017,2018,2019,2020,2021, 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
  36 /*! \internal \file
  37  * \brief
  38  * Implements the Bias class.
  39  *
  40  * \author Viveca Lindahl
  41  * \author Berk Hess <hess@kth.se>
  42  * \ingroup module_awh
  43  */
  44
  45 #include "gmxpre.h"
  46
  47 #include "bias.h"
  48
  49 #include <cassert>
  50 #include <cinttypes>
  51 #include <cmath>
  52 #include <cstdio>
  53 #include <cstdlib>
  54 #include <cstring>
  55
  56 #include <algorithm>
  57 #include <memory>
  58
  59 #include "gromacs/fileio/gmxfio.h"
  60 #include "gromacs/gmxlib/network.h"
  61 #include "gromacs/math/functions.h"
  62 #include "gromacs/math/utilities.h"
  63 #include "gromacs/mdtypes/awh_history.h"
  64 #include "gromacs/mdtypes/awh_params.h"
  65 #include "gromacs/mdtypes/commrec.h"
  66 #include "gromacs/utility/exceptions.h"
  67 #include "gromacs/utility/gmxassert.h"
  68 #include "gromacs/utility/stringutil.h"
  69
  70 #include "correlationgrid.h"
  71 #include "correlationhistory.h"
  72 #include "pointstate.h"
  73
  74 namespace gmx
  75 {
  76
  77 void Bias::warnForHistogramAnomalies(double t, int64_t step, FILE* fplog)
  78 {
  79     const int maxNumWarningsInCheck = 1;  /* The maximum number of warnings to print per check */
  80     const int maxNumWarningsInRun   = 10; /* The maximum number of warnings to print in a run */
  81
  82     if (fplog == nullptr || numWarningsIssued_ >= maxNumWarningsInRun || state_.inInitialStage()
  83         || !params_.isCheckHistogramForAnomaliesStep(step))
  84     {
  85         return;
  86     }
  87
  88     numWarningsIssued_ +=
  89             state_.warnForHistogramAnomalies(grid_, biasIndex(), t, fplog, maxNumWarningsInCheck);
  90
  91     if (numWarningsIssued_ >= maxNumWarningsInRun)
  92     {
  93         fprintf(fplog, "\nawh%d: suppressing future AWH warnings.\n", biasIndex() + 1);
  94     }
  95 }
  96
  97 void Bias::doSkippedUpdatesForAllPoints()
  98 {
  99     if (params_.skipUpdates())
 100     {
 101         state_.doSkippedUpdatesForAllPoints(params_);
 102     }
 103 }
 104
 105 gmx::ArrayRef<const double> Bias::calcForceAndUpdateBias(const awh_dvec         coordValue,
 106                                                          ArrayRef<const double> neighborLambdaEnergies,
 107                                                          ArrayRef<const double> neighborLambdaDhdl,
 108                                                          double*                awhPotential,
 109                                                          double*                potentialJump,
 110                                                          const t_commrec*       commRecord,
 111                                                          const gmx_multisim_t*  ms,
 112                                                          double                 t,
 113                                                          int64_t                step,
 114                                                          int64_t                seed,
 115                                                          FILE*                  fplog)
 116 {
 117     if (step < 0)
 118     {
 119         GMX_THROW(InvalidInputError(
 120                 "The step number is negative which is not supported by the AWH code."));
 121     }
 122
 123     GMX_RELEASE_ASSERT(!(params_.convolveForce && grid_.hasLambdaAxis()),
 124                        "When using AWH to sample an FEP lambda dimension the AWH potential cannot "
 125                        "be convolved.");
 126
 127     state_.setCoordValue(grid_, coordValue);
 128
 129     std::vector<double, AlignedAllocator<double>>& probWeightNeighbor = alignedTempWorkSpace_;
 130
 131     /* If the convolved force is needed or this is a sampling step,
 132      * the bias in the current neighborhood needs to be up-to-date
 133      * and the probablity weights need to be calculated.
 134      */
 135     const bool        isSampleCoordStep = params_.isSampleCoordStep(step);
 136     const bool        moveUmbrella      = (isSampleCoordStep || step == 0);
 137     double            convolvedBias     = 0;
 138     const CoordState& coordState        = state_.coordState();
 139
 140     if (params_.convolveForce || moveUmbrella || isSampleCoordStep)
 141     {
 142         if (params_.skipUpdates())
 143         {
 144             state_.doSkippedUpdatesInNeighborhood(params_, grid_);
 145         }
 146         convolvedBias = state_.updateProbabilityWeightsAndConvolvedBias(
 147                 dimParams_, grid_, moveUmbrella ? neighborLambdaEnergies : ArrayRef<const double>{}, &probWeightNeighbor);
 148
 149         if (isSampleCoordStep)
 150         {
 151             updateForceCorrelationGrid(probWeightNeighbor, neighborLambdaDhdl, t);
 152
 153             state_.sampleCoordAndPmf(dimParams_, grid_, probWeightNeighbor, convolvedBias);
 154         }
 155     }
 156
 157     /* Set the bias force and get the potential contribution from this bias.
 158      * The potential jump occurs at different times depending on how
 159      * the force is applied (and how the potential is normalized).
 160      * For the convolved force it happens when the bias is updated,
 161      * for the umbrella when the umbrella is moved.
 162      */
 163     *potentialJump = 0;
 164     double potential;
 165     if (params_.convolveForce)
 166     {
 167         state_.calcConvolvedForce(dimParams_,
 168                                   grid_,
 169                                   probWeightNeighbor,
 170                                   moveUmbrella ? neighborLambdaDhdl : ArrayRef<const double>{},
 171                                   tempForce_,
 172                                   biasForce_);
 173
 174         potential = -convolvedBias * params_.invBeta;
 175     }
 176     else
 177     {
 178         /* Umbrella force */
 179         GMX_RELEASE_ASSERT(state_.points()[coordState.umbrellaGridpoint()].inTargetRegion(),
 180                            "AWH bias grid point for the umbrella reference value is outside of the "
 181                            "target region.");
 182         potential = state_.calcUmbrellaForceAndPotential(
 183                 dimParams_,
 184                 grid_,
 185                 coordState.umbrellaGridpoint(),
 186                 moveUmbrella ? neighborLambdaDhdl : ArrayRef<const double>{},
 187                 biasForce_);
 188
 189         /* Moving the umbrella results in a force correction and
 190          * a new potential. The umbrella center is sampled as often as
 191          * the coordinate so we know the probability weights needed
 192          * for moving the umbrella are up-to-date.
 193          */
 194         if (moveUmbrella)
 195         {
 196             const bool onlySampleUmbrellaGridpoint = false;
 197             double     newPotential                = state_.moveUmbrella(dimParams_,
 198                                                       grid_,
 199                                                       probWeightNeighbor,
 200                                                       neighborLambdaDhdl,
 201                                                       biasForce_,
 202                                                       step,
 203                                                       seed,
 204                                                       params_.biasIndex,
 205                                                       onlySampleUmbrellaGridpoint);
 206             *potentialJump                         = newPotential - potential;
 207         }
 208     }
 209
 210     /* Update the free energy estimates and bias and other history dependent method parameters */
 211     if (params_.isUpdateFreeEnergyStep(step))
 212     {
 213         state_.updateFreeEnergyAndAddSamplesToHistogram(
 214                 dimParams_, grid_, params_, commRecord, ms, t, step, fplog, &updateList_);
 215
 216         if (params_.convolveForce)
 217         {
 218             /* The update results in a potential jump, so we need the new convolved potential. */
 219             double newPotential = -calcConvolvedBias(coordState.coordValue()) * params_.invBeta;
 220             *potentialJump      = newPotential - potential;
 221         }
 222     }
 223     /* If there is a lambda axis it is still controlled using an umbrella even if the force
 224      * is convolved in the other dimensions. */
 225     if (moveUmbrella && params_.convolveForce && grid_.hasLambdaAxis())
 226     {
 227         const bool onlySampleUmbrellaGridpoint = true;
 228         state_.moveUmbrella(dimParams_,
 229                             grid_,
 230                             probWeightNeighbor,
 231                             neighborLambdaDhdl,
 232                             biasForce_,
 233                             step,
 234                             seed,
 235                             params_.biasIndex,
 236                             onlySampleUmbrellaGridpoint);
 237     }
 238
 239     /* Return the potential. */
 240     *awhPotential = potential;
 241
 242     /* Check the sampled histograms and potentially warn user if something is suspicious */
 243     warnForHistogramAnomalies(t, step, fplog);
 244
 245     return biasForce_;
 246 }
 247
 248 /*! \brief
 249  * Count the total number of samples / sample weight over all grid points.
 250  *
 251  * \param[in] pointState  The state of the points in a bias.
 252  * \returns the total sample count.
 253  */
 254 static int64_t countSamples(ArrayRef<const PointState> pointState)
 255 {
 256     double numSamples = 0;
 257     for (const PointState& point : pointState)
 258     {
 259         numSamples += point.weightSumTot();
 260     }
 261
 262     return gmx::roundToInt64(numSamples);
 263 }
 264
 265 /*! \brief
 266  * Check if the state (loaded from checkpoint) and the run are consistent.
 267  *
 268  * When the state and the run setup are inconsistent, an exception is thrown.
 269  *
 270  * \param[in] params  The parameters of the bias.
 271  * \param[in] state   The state of the bias.
 272  */
 273 static void ensureStateAndRunConsistency(const BiasParams& params, const BiasState& state)
 274 {
 275     int64_t numSamples = countSamples(state.points());
 276     int64_t numUpdatesFromSamples =
 277             numSamples / (params.numSamplesUpdateFreeEnergy_ * params.numSharedUpdate);
 278     int64_t numUpdatesExpected = state.histogramSize().numUpdates();
 279     if (numUpdatesFromSamples != numUpdatesExpected)
 280     {
 281         std::string mesg = gmx::formatString(
 282                 "The number of AWH updates in the checkpoint file (%" PRId64
 283                 ") does not match the total number of AWH samples divided by the number of samples "
 284                 "per update for %d sharing AWH bias(es) (%" PRId64 "/%d=%" PRId64 ")",
 285                 numUpdatesExpected,
 286                 params.numSharedUpdate,
 287                 numSamples,
 288                 params.numSamplesUpdateFreeEnergy_ * params.numSharedUpdate,
 289                 numUpdatesFromSamples);
 290         mesg += " Maybe you changed AWH parameters.";
 291         /* Unfortunately we currently do not store the number of simulations
 292          * sharing the bias or the state to checkpoint. But we can hint at
 293          * a mismatch in the number of sharing simulations.
 294          */
 295         if (numUpdatesFromSamples % state.histogramSize().numUpdates() == 0)
 296         {
 297             mesg += gmx::formatString(
 298                     " Or the run you continued from used %" PRId64
 299                     " sharing simulations, whereas you now specified %d sharing simulations.",
 300                     numUpdatesFromSamples / state.histogramSize().numUpdates(),
 301                     params.numSharedUpdate);
 302         }
 303         GMX_THROW(InvalidInputError(mesg));
 304     }
 305 }
 306
 307 void Bias::restoreStateFromHistory(const AwhBiasHistory* biasHistory, const t_commrec* cr)
 308 {
 309     GMX_RELEASE_ASSERT(thisRankDoesIO_ == MASTER(cr),
 310                        "The master rank should do I/O, the other ranks should not");
 311
 312     if (MASTER(cr))
 313     {
 314         GMX_RELEASE_ASSERT(biasHistory != nullptr,
 315                            "On the master rank we need a valid history object to restore from");
 316         state_.restoreFromHistory(*biasHistory, grid_);
 317
 318         /* Ensure that the state is consistent with our current run setup,
 319          * since the user can have changed AWH parameters or the number
 320          * of simulations sharing the bias.
 321          */
 322         ensureStateAndRunConsistency(params_, state_);
 323
 324         if (forceCorrelationGrid_ != nullptr)
 325         {
 326             forceCorrelationGrid_->restoreStateFromHistory(biasHistory->forceCorrelationGrid);
 327         }
 328     }
 329
 330     if (PAR(cr))
 331     {
 332         state_.broadcast(cr);
 333     }
 334 }
 335
 336 void Bias::initHistoryFromState(AwhBiasHistory* biasHistory) const
 337 {
 338     GMX_RELEASE_ASSERT(biasHistory != nullptr, "Need a valid biasHistory");
 339
 340     state_.initHistoryFromState(biasHistory);
 341
 342     if (forceCorrelationGrid_ != nullptr)
 343     {
 344         biasHistory->forceCorrelationGrid = initCorrelationGridHistoryFromState(forceCorrelationGrid());
 345     }
 346 }
 347
 348 void Bias::updateHistory(AwhBiasHistory* biasHistory) const
 349 {
 350     GMX_RELEASE_ASSERT(biasHistory != nullptr, "Need a valid biasHistory");
 351
 352     state_.updateHistory(biasHistory, grid_);
 353
 354     if (forceCorrelationGrid_ != nullptr)
 355     {
 356         updateCorrelationGridHistory(&biasHistory->forceCorrelationGrid, forceCorrelationGrid());
 357     }
 358 }
 359
 360 Bias::Bias(int                            biasIndexInCollection,
 361            const AwhParams&               awhParams,
 362            const AwhBiasParams&           awhBiasParams,
 363            ArrayRef<const DimParams>      dimParamsInit,
 364            double                         beta,
 365            double                         mdTimeStep,
 366            int                            numSharingSimulations,
 367            const std::string&             biasInitFilename,
 368            ThisRankWillDoIO               thisRankWillDoIO,
 369            BiasParams::DisableUpdateSkips disableUpdateSkips) :
 370     dimParams_(dimParamsInit.begin(), dimParamsInit.end()),
 371     grid_(dimParamsInit, awhBiasParams.dimParams()),
 372     params_(awhParams,
 373             awhBiasParams,
 374             dimParams_,
 375             beta,
 376             mdTimeStep,
 377             disableUpdateSkips,
 378             numSharingSimulations,
 379             grid_.axis(),
 380             biasIndexInCollection),
 381     state_(awhBiasParams, params_.initialHistogramSize, dimParams_, grid_),
 382     thisRankDoesIO_(thisRankWillDoIO == ThisRankWillDoIO::Yes),
 383     biasForce_(ndim()),
 384
 385     tempForce_(ndim()),
 386     numWarningsIssued_(0)
 387 {
 388     /* For a global update updateList covers all points, so reserve that */
 389     updateList_.reserve(grid_.numPoints());
 390
 391     state_.initGridPointState(
 392             awhBiasParams, dimParams_, grid_, params_, biasInitFilename, awhParams.numBias());
 393
 394     if (thisRankDoesIO_)
 395     {
 396         /* Set up the force correlation object. */
 397
 398         /* We let the correlation init function set its parameters
 399          * to something useful for now.
 400          */
 401         double blockLength = 0;
 402         /* Construct the force correlation object. */
 403         forceCorrelationGrid_ =
 404                 std::make_unique<CorrelationGrid>(state_.points().size(),
 405                                                   ndim(),
 406                                                   blockLength,
 407                                                   CorrelationGrid::BlockLengthMeasure::Time,
 408                                                   awhParams.nstSampleCoord() * mdTimeStep);
 409
 410         writer_ = std::make_unique<BiasWriter>(*this);
 411     }
 412 }
 413
 414 void Bias::printInitializationToLog(FILE* fplog) const
 415 {
 416     if (fplog != nullptr && forceCorrelationGrid_ != nullptr)
 417     {
 418         std::string prefix = gmx::formatString("\nawh%d:", params_.biasIndex + 1);
 419
 420         fprintf(fplog,
 421                 "%s initial force correlation block length = %g %s"
 422                 "%s force correlation number of blocks = %d",
 423                 prefix.c_str(),
 424                 forceCorrelationGrid().getBlockLength(),
 425                 forceCorrelationGrid().blockLengthMeasure == CorrelationGrid::BlockLengthMeasure::Weight
 426                         ? ""
 427                         : "ps",
 428                 prefix.c_str(),
 429                 forceCorrelationGrid().getNumBlocks());
 430     }
 431 }
 432
 433 void Bias::updateForceCorrelationGrid(gmx::ArrayRef<const double> probWeightNeighbor,
 434                                       ArrayRef<const double>      neighborLambdaDhdl,
 435                                       double                      t)
 436 {
 437     if (forceCorrelationGrid_ == nullptr)
 438     {
 439         return;
 440     }
 441
 442     const std::vector<int>& neighbor = grid_.point(state_.coordState().gridpointIndex()).neighbor;
 443
 444     gmx::ArrayRef<double> forceFromNeighbor = tempForce_;
 445     for (size_t n = 0; n < neighbor.size(); n++)
 446     {
 447         double weightNeighbor = probWeightNeighbor[n];
 448         int    indexNeighbor  = neighbor[n];
 449
 450         /* Add the force data of this neighbor point. Note: the sum of these forces is the convolved force.
 451
 452            We actually add the force normalized by beta which has the units of 1/length. This means that the
 453            resulting correlation time integral is directly in units of friction time/length^2 which is really what
 454            we're interested in. */
 455         state_.calcUmbrellaForceAndPotential(
 456                 dimParams_, grid_, indexNeighbor, neighborLambdaDhdl, forceFromNeighbor);
 457
 458         /* Note: we might want to give a whole list of data to add instead and have this loop in the data adding function */
 459         forceCorrelationGrid_->addData(indexNeighbor, weightNeighbor, forceFromNeighbor, t);
 460     }
 461 }
 462
 463 /* Return the number of data blocks that have been prepared for writing. */
 464 int Bias::numEnergySubblocksToWrite() const
 465 {
 466     GMX_RELEASE_ASSERT(writer_ != nullptr, "Should only request data from an initialized writer");
 467
 468     return writer_->numBlocks();
 469 }
 470
 471 /* Write bias data blocks to energy subblocks. */
 472 int Bias::writeToEnergySubblocks(t_enxsubblock* subblock) const
 473 {
 474     GMX_RELEASE_ASSERT(writer_ != nullptr, "Should only request data from an initialized writer");
 475
 476     return writer_->writeToEnergySubblocks(*this, subblock);
 477 }
 478
 479 bool Bias::isSampleCoordStep(const int64_t step) const
 480 {
 481     return params_.isSampleCoordStep(step);
 482 }
 483
 484
 485 } // namespace gmx