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, 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     state_.setCoordValue(grid_, coordValue);
 124
 125     std::vector<double, AlignedAllocator<double>>& probWeightNeighbor = alignedTempWorkSpace_;
 126
 127     /* If the convolved force is needed or this is a sampling step,
 128      * the bias in the current neighborhood needs to be up-to-date
 129      * and the probablity weights need to be calculated.
 130      */
 131     const bool        isSampleCoordStep = params_.isSampleCoordStep(step);
 132     const bool        moveUmbrella      = (isSampleCoordStep || step == 0);
 133     double            convolvedBias     = 0;
 134     const CoordState& coordState        = state_.coordState();
 135
 136     if (params_.convolveForce || moveUmbrella || isSampleCoordStep)
 137     {
 138         if (params_.skipUpdates())
 139         {
 140             state_.doSkippedUpdatesInNeighborhood(params_, grid_);
 141         }
 142         convolvedBias = state_.updateProbabilityWeightsAndConvolvedBias(
 143                 dimParams_, grid_, moveUmbrella ? neighborLambdaEnergies : ArrayRef<const double>{},
 144                 &probWeightNeighbor);
 145
 146         if (isSampleCoordStep)
 147         {
 148             updateForceCorrelationGrid(probWeightNeighbor, neighborLambdaDhdl, t);
 149
 150             state_.sampleCoordAndPmf(dimParams_, grid_, probWeightNeighbor, convolvedBias);
 151         }
 152         /* Set the umbrella grid point (for the lambda axis) to the
 153          * current grid point. */
 154         if (params_.convolveForce && grid_.hasLambdaAxis())
 155         {
 156             state_.setUmbrellaGridpointToGridpoint();
 157         }
 158     }
 159
 160     /* Set the bias force and get the potential contribution from this bias.
 161      * The potential jump occurs at different times depending on how
 162      * the force is applied (and how the potential is normalized).
 163      * For the convolved force it happens when the bias is updated,
 164      * for the umbrella when the umbrella is moved.
 165      */
 166     *potentialJump = 0;
 167     double potential;
 168     if (params_.convolveForce)
 169     {
 170         state_.calcConvolvedForce(dimParams_, grid_, probWeightNeighbor,
 171                                   moveUmbrella ? neighborLambdaDhdl : ArrayRef<const double>{},
 172                                   tempForce_, 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_, grid_, coordState.umbrellaGridpoint(),
 184                 moveUmbrella ? neighborLambdaDhdl : ArrayRef<const double>{}, biasForce_);
 185
 186         /* Moving the umbrella results in a force correction and
 187          * a new potential. The umbrella center is sampled as often as
 188          * the coordinate so we know the probability weights needed
 189          * for moving the umbrella are up-to-date.
 190          */
 191         if (moveUmbrella)
 192         {
 193             const bool onlySampleUmbrellaGridpoint = false;
 194             double     newPotential = state_.moveUmbrella(dimParams_, grid_, probWeightNeighbor,
 195                                                       neighborLambdaDhdl, biasForce_, step, seed,
 196                                                       params_.biasIndex, onlySampleUmbrellaGridpoint);
 197             *potentialJump          = newPotential - potential;
 198         }
 199     }
 200
 201     /* Update the free energy estimates and bias and other history dependent method parameters */
 202     if (params_.isUpdateFreeEnergyStep(step))
 203     {
 204         state_.updateFreeEnergyAndAddSamplesToHistogram(dimParams_, grid_, params_, commRecord, ms,
 205                                                         t, step, fplog, &updateList_);
 206
 207         if (params_.convolveForce)
 208         {
 209             /* The update results in a potential jump, so we need the new convolved potential. */
 210             double newPotential = -calcConvolvedBias(coordState.coordValue()) * params_.invBeta;
 211             *potentialJump      = newPotential - potential;
 212         }
 213     }
 214     /* If there is a lambda axis it is still controlled using an umbrella even if the force
 215      * is convolved in the other dimensions. */
 216     if (moveUmbrella && params_.convolveForce && grid_.hasLambdaAxis())
 217     {
 218         const bool onlySampleUmbrellaGridpoint = true;
 219         state_.moveUmbrella(dimParams_, grid_, probWeightNeighbor, neighborLambdaDhdl, biasForce_,
 220                             step, seed, params_.biasIndex, onlySampleUmbrellaGridpoint);
 221     }
 222
 223     /* Return the potential. */
 224     *awhPotential = potential;
 225
 226     /* Check the sampled histograms and potentially warn user if something is suspicious */
 227     warnForHistogramAnomalies(t, step, fplog);
 228
 229     return biasForce_;
 230 }
 231
 232 /*! \brief
 233  * Count the total number of samples / sample weight over all grid points.
 234  *
 235  * \param[in] pointState  The state of the points in a bias.
 236  * \returns the total sample count.
 237  */
 238 static int64_t countSamples(const std::vector<PointState>& pointState)
 239 {
 240     double numSamples = 0;
 241     for (const PointState& point : pointState)
 242     {
 243         numSamples += point.weightSumTot();
 244     }
 245
 246     return gmx::roundToInt64(numSamples);
 247 }
 248
 249 /*! \brief
 250  * Check if the state (loaded from checkpoint) and the run are consistent.
 251  *
 252  * When the state and the run setup are inconsistent, an exception is thrown.
 253  *
 254  * \param[in] params  The parameters of the bias.
 255  * \param[in] state   The state of the bias.
 256  */
 257 static void ensureStateAndRunConsistency(const BiasParams& params, const BiasState& state)
 258 {
 259     int64_t numSamples = countSamples(state.points());
 260     int64_t numUpdatesFromSamples =
 261             numSamples / (params.numSamplesUpdateFreeEnergy_ * params.numSharedUpdate);
 262     int64_t numUpdatesExpected = state.histogramSize().numUpdates();
 263     if (numUpdatesFromSamples != numUpdatesExpected)
 264     {
 265         std::string mesg = gmx::formatString(
 266                 "The number of AWH updates in the checkpoint file (%" PRId64
 267                 ") does not match the total number of AWH samples divided by the number of samples "
 268                 "per update for %d sharing AWH bias(es) (%" PRId64 "/%d=%" PRId64 ")",
 269                 numUpdatesExpected, params.numSharedUpdate, numSamples,
 270                 params.numSamplesUpdateFreeEnergy_ * params.numSharedUpdate, numUpdatesFromSamples);
 271         mesg += " Maybe you changed AWH parameters.";
 272         /* Unfortunately we currently do not store the number of simulations
 273          * sharing the bias or the state to checkpoint. But we can hint at
 274          * a mismatch in the number of sharing simulations.
 275          */
 276         if (numUpdatesFromSamples % state.histogramSize().numUpdates() == 0)
 277         {
 278             mesg += gmx::formatString(
 279                     " Or the run you continued from used %" PRId64
 280                     " sharing simulations, whereas you now specified %d sharing simulations.",
 281                     numUpdatesFromSamples / state.histogramSize().numUpdates(), params.numSharedUpdate);
 282         }
 283         GMX_THROW(InvalidInputError(mesg));
 284     }
 285 }
 286
 287 void Bias::restoreStateFromHistory(const AwhBiasHistory* biasHistory, const t_commrec* cr)
 288 {
 289     GMX_RELEASE_ASSERT(thisRankDoesIO_ == MASTER(cr),
 290                        "The master rank should do I/O, the other ranks should not");
 291
 292     if (MASTER(cr))
 293     {
 294         GMX_RELEASE_ASSERT(biasHistory != nullptr,
 295                            "On the master rank we need a valid history object to restore from");
 296         state_.restoreFromHistory(*biasHistory, grid_);
 297
 298         /* Ensure that the state is consistent with our current run setup,
 299          * since the user can have changed AWH parameters or the number
 300          * of simulations sharing the bias.
 301          */
 302         ensureStateAndRunConsistency(params_, state_);
 303
 304         if (forceCorrelationGrid_ != nullptr)
 305         {
 306             forceCorrelationGrid_->restoreStateFromHistory(biasHistory->forceCorrelationGrid);
 307         }
 308     }
 309
 310     if (PAR(cr))
 311     {
 312         state_.broadcast(cr);
 313     }
 314 }
 315
 316 void Bias::initHistoryFromState(AwhBiasHistory* biasHistory) const
 317 {
 318     GMX_RELEASE_ASSERT(biasHistory != nullptr, "Need a valid biasHistory");
 319
 320     state_.initHistoryFromState(biasHistory);
 321
 322     if (forceCorrelationGrid_ != nullptr)
 323     {
 324         biasHistory->forceCorrelationGrid = initCorrelationGridHistoryFromState(forceCorrelationGrid());
 325     }
 326 }
 327
 328 void Bias::updateHistory(AwhBiasHistory* biasHistory) const
 329 {
 330     GMX_RELEASE_ASSERT(biasHistory != nullptr, "Need a valid biasHistory");
 331
 332     state_.updateHistory(biasHistory, grid_);
 333
 334     if (forceCorrelationGrid_ != nullptr)
 335     {
 336         updateCorrelationGridHistory(&biasHistory->forceCorrelationGrid, forceCorrelationGrid());
 337     }
 338 }
 339
 340 Bias::Bias(int                            biasIndexInCollection,
 341            const AwhParams&               awhParams,
 342            const AwhBiasParams&           awhBiasParams,
 343            const std::vector<DimParams>&  dimParamsInit,
 344            double                         beta,
 345            double                         mdTimeStep,
 346            int                            numSharingSimulations,
 347            const std::string&             biasInitFilename,
 348            ThisRankWillDoIO               thisRankWillDoIO,
 349            BiasParams::DisableUpdateSkips disableUpdateSkips) :
 350     dimParams_(dimParamsInit),
 351     grid_(dimParamsInit, awhBiasParams.dimParams),
 352     params_(awhParams,
 353             awhBiasParams,
 354             dimParams_,
 355             beta,
 356             mdTimeStep,
 357             disableUpdateSkips,
 358             numSharingSimulations,
 359             grid_.axis(),
 360             biasIndexInCollection),
 361     state_(awhBiasParams, params_.initialHistogramSize, dimParams_, grid_),
 362     thisRankDoesIO_(thisRankWillDoIO == ThisRankWillDoIO::Yes),
 363     biasForce_(ndim()),
 364
 365     tempForce_(ndim()),
 366     numWarningsIssued_(0)
 367 {
 368     /* For a global update updateList covers all points, so reserve that */
 369     updateList_.reserve(grid_.numPoints());
 370
 371     state_.initGridPointState(awhBiasParams, dimParams_, grid_, params_, biasInitFilename, awhParams.numBias);
 372
 373     if (thisRankDoesIO_)
 374     {
 375         /* Set up the force correlation object. */
 376
 377         /* We let the correlation init function set its parameters
 378          * to something useful for now.
 379          */
 380         double blockLength = 0;
 381         /* Construct the force correlation object. */
 382         forceCorrelationGrid_ = std::make_unique<CorrelationGrid>(
 383                 state_.points().size(), ndim(), blockLength,
 384                 CorrelationGrid::BlockLengthMeasure::Time, awhParams.nstSampleCoord * mdTimeStep);
 385
 386         writer_ = std::make_unique<BiasWriter>(*this);
 387     }
 388 }
 389
 390 void Bias::printInitializationToLog(FILE* fplog) const
 391 {
 392     if (fplog != nullptr && forceCorrelationGrid_ != nullptr)
 393     {
 394         std::string prefix = gmx::formatString("\nawh%d:", params_.biasIndex + 1);
 395
 396         fprintf(fplog,
 397                 "%s initial force correlation block length = %g %s"
 398                 "%s force correlation number of blocks = %d",
 399                 prefix.c_str(), forceCorrelationGrid().getBlockLength(),
 400                 forceCorrelationGrid().blockLengthMeasure == CorrelationGrid::BlockLengthMeasure::Weight
 401                         ? ""
 402                         : "ps",
 403                 prefix.c_str(), forceCorrelationGrid().getNumBlocks());
 404     }
 405 }
 406
 407 void Bias::updateForceCorrelationGrid(gmx::ArrayRef<const double> probWeightNeighbor,
 408                                       ArrayRef<const double>      neighborLambdaDhdl,
 409                                       double                      t)
 410 {
 411     if (forceCorrelationGrid_ == nullptr)
 412     {
 413         return;
 414     }
 415
 416     const std::vector<int>& neighbor = grid_.point(state_.coordState().gridpointIndex()).neighbor;
 417
 418     gmx::ArrayRef<double> forceFromNeighbor = tempForce_;
 419     for (size_t n = 0; n < neighbor.size(); n++)
 420     {
 421         double weightNeighbor = probWeightNeighbor[n];
 422         int    indexNeighbor  = neighbor[n];
 423
 424         /* Add the force data of this neighbor point. Note: the sum of these forces is the convolved force.
 425
 426            We actually add the force normalized by beta which has the units of 1/length. This means that the
 427            resulting correlation time integral is directly in units of friction time/length^2 which is really what
 428            we're interested in. */
 429         state_.calcUmbrellaForceAndPotential(dimParams_, grid_, indexNeighbor, neighborLambdaDhdl,
 430                                              forceFromNeighbor);
 431
 432         /* Note: we might want to give a whole list of data to add instead and have this loop in the data adding function */
 433         forceCorrelationGrid_->addData(indexNeighbor, weightNeighbor, forceFromNeighbor, t);
 434     }
 435 }
 436
 437 /* Return the number of data blocks that have been prepared for writing. */
 438 int Bias::numEnergySubblocksToWrite() const
 439 {
 440     GMX_RELEASE_ASSERT(writer_ != nullptr, "Should only request data from an initialized writer");
 441
 442     return writer_->numBlocks();
 443 }
 444
 445 /* Write bias data blocks to energy subblocks. */
 446 int Bias::writeToEnergySubblocks(t_enxsubblock* subblock) const
 447 {
 448     GMX_RELEASE_ASSERT(writer_ != nullptr, "Should only request data from an initialized writer");
 449
 450     return writer_->writeToEnergySubblocks(*this, subblock);
 451 }
 452
 453 bool Bias::isFepLambdaDimension(int dim) const
 454 {
 455     GMX_ASSERT(dim < ndim(), "Requested dimension out of range.");
 456
 457     return dimParams_[dim].isFepLambdaDimension();
 458 }
 459
 460 bool Bias::isSampleCoordStep(const int64_t step) const
 461 {
 462     return params_.isSampleCoordStep(step);
 463 }
 464
 465
 466 } // namespace gmx