src/gromacs/mdlib/force.cpp

   1 /*
   2  * This file is part of the GROMACS molecular simulation package.
   3  *
   4  * Copyright (c) 1991-2000, University of Groningen, The Netherlands.
   5  * Copyright (c) 2001-2004, The GROMACS development team.
   6  * Copyright (c) 2013,2014,2015,2016,2017,2018, by the GROMACS development team, led by
   7  * Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl,
   8  * and including many others, as listed in the AUTHORS file in the
   9  * top-level source directory and at http://www.gromacs.org.
  10  *
  11  * GROMACS is free software; you can redistribute it and/or
  12  * modify it under the terms of the GNU Lesser General Public License
  13  * as published by the Free Software Foundation; either version 2.1
  14  * of the License, or (at your option) any later version.
  15  *
  16  * GROMACS is distributed in the hope that it will be useful,
  17  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  18  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  19  * Lesser General Public License for more details.
  20  *
  21  * You should have received a copy of the GNU Lesser General Public
  22  * License along with GROMACS; if not, see
  23  * http://www.gnu.org/licenses, or write to the Free Software Foundation,
  24  * Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA.
  25  *
  26  * If you want to redistribute modifications to GROMACS, please
  27  * consider that scientific software is very special. Version
  28  * control is crucial - bugs must be traceable. We will be happy to
  29  * consider code for inclusion in the official distribution, but
  30  * derived work must not be called official GROMACS. Details are found
  31  * in the README & COPYING files - if they are missing, get the
  32  * official version at http://www.gromacs.org.
  33  *
  34  * To help us fund GROMACS development, we humbly ask that you cite
  35  * the research papers on the package. Check out http://www.gromacs.org.
  36  */
  37 #include "gmxpre.h"
  38
  39 #include "force.h"
  40
  41 #include "config.h"
  42
  43 #include <cassert>
  44 #include <cmath>
  45 #include <cstring>
  46
  47 #include "gromacs/domdec/domdec.h"
  48 #include "gromacs/domdec/domdec_struct.h"
  49 #include "gromacs/ewald/ewald.h"
  50 #include "gromacs/ewald/long-range-correction.h"
  51 #include "gromacs/ewald/pme.h"
  52 #include "gromacs/gmxlib/network.h"
  53 #include "gromacs/gmxlib/nrnb.h"
  54 #include "gromacs/gmxlib/nonbonded/nonbonded.h"
  55 #include "gromacs/listed-forces/listed-forces.h"
  56 #include "gromacs/math/vec.h"
  57 #include "gromacs/math/vecdump.h"
  58 #include "gromacs/mdlib/force_flags.h"
  59 #include "gromacs/mdlib/forcerec-threading.h"
  60 #include "gromacs/mdlib/mdrun.h"
  61 #include "gromacs/mdlib/ns.h"
  62 #include "gromacs/mdlib/qmmm.h"
  63 #include "gromacs/mdlib/rf_util.h"
  64 #include "gromacs/mdlib/wall.h"
  65 #include "gromacs/mdtypes/commrec.h"
  66 #include "gromacs/mdtypes/enerdata.h"
  67 #include "gromacs/mdtypes/forceoutput.h"
  68 #include "gromacs/mdtypes/forcerec.h"
  69 #include "gromacs/mdtypes/inputrec.h"
  70 #include "gromacs/mdtypes/md_enums.h"
  71 #include "gromacs/pbcutil/ishift.h"
  72 #include "gromacs/pbcutil/mshift.h"
  73 #include "gromacs/pbcutil/pbc.h"
  74 #include "gromacs/timing/wallcycle.h"
  75 #include "gromacs/utility/cstringutil.h"
  76 #include "gromacs/utility/exceptions.h"
  77 #include "gromacs/utility/fatalerror.h"
  78 #include "gromacs/utility/smalloc.h"
  79
  80 void ns(FILE               *fp,
  81         t_forcerec         *fr,
  82         matrix              box,
  83         const gmx_groups_t *groups,
  84         gmx_localtop_t     *top,
  85         const t_mdatoms    *md,
  86         const t_commrec    *cr,
  87         t_nrnb             *nrnb,
  88         gmx_bool            bFillGrid)
  89 {
  90     int     nsearch;
  91
  92
  93     if (!fr->ns->nblist_initialized)
  94     {
  95         init_neighbor_list(fp, fr, md->homenr);
  96     }
  97
  98     nsearch = search_neighbours(fp, fr, box, top, groups, cr, nrnb, md,
  99                                 bFillGrid);
 100     if (debug)
 101     {
 102         fprintf(debug, "nsearch = %d\n", nsearch);
 103     }
 104
 105     /* Check whether we have to do dynamic load balancing */
 106     /*if ((nsb->nstDlb > 0) && (mod(step,nsb->nstDlb) == 0))
 107        count_nb(cr,nsb,&(top->blocks[ebCGS]),nns,fr->nlr,
 108        &(top->idef),opts->ngener);
 109      */
 110     if (fr->ns->dump_nl > 0)
 111     {
 112         dump_nblist(fp, cr, fr, fr->ns->dump_nl);
 113     }
 114 }
 115
 116 static void clearEwaldThreadOutput(ewald_corr_thread_t *ewc_t)
 117 {
 118     ewc_t->Vcorr_q        = 0;
 119     ewc_t->Vcorr_lj       = 0;
 120     ewc_t->dvdl[efptCOUL] = 0;
 121     ewc_t->dvdl[efptVDW]  = 0;
 122     clear_mat(ewc_t->vir_q);
 123     clear_mat(ewc_t->vir_lj);
 124 }
 125
 126 static void reduceEwaldThreadOuput(int nthreads, ewald_corr_thread_t *ewc_t)
 127 {
 128     ewald_corr_thread_t &dest = ewc_t[0];
 129
 130     for (int t = 1; t < nthreads; t++)
 131     {
 132         dest.Vcorr_q        += ewc_t[t].Vcorr_q;
 133         dest.Vcorr_lj       += ewc_t[t].Vcorr_lj;
 134         dest.dvdl[efptCOUL] += ewc_t[t].dvdl[efptCOUL];
 135         dest.dvdl[efptVDW]  += ewc_t[t].dvdl[efptVDW];
 136         m_add(dest.vir_q,  ewc_t[t].vir_q,  dest.vir_q);
 137         m_add(dest.vir_lj, ewc_t[t].vir_lj, dest.vir_lj);
 138     }
 139 }
 140
 141 void do_force_lowlevel(t_forcerec           *fr,
 142                        const t_inputrec     *ir,
 143                        const t_idef         *idef,
 144                        const t_commrec      *cr,
 145                        const gmx_multisim_t *ms,
 146                        t_nrnb               *nrnb,
 147                        gmx_wallcycle_t       wcycle,
 148                        const t_mdatoms      *md,
 149                        rvec                  x[],
 150                        history_t            *hist,
 151                        rvec                 *forceForUseWithShiftForces,
 152                        gmx::ForceWithVirial *forceWithVirial,
 153                        gmx_enerdata_t       *enerd,
 154                        t_fcdata             *fcd,
 155                        matrix                box,
 156                        t_lambda             *fepvals,
 157                        real                 *lambda,
 158                        const t_graph        *graph,
 159                        const t_blocka       *excl,
 160                        rvec                  mu_tot[],
 161                        int                   flags,
 162                        float                *cycles_pme)
 163 {
 164     int         i, j;
 165     int         donb_flags;
 166     int         pme_flags;
 167     t_pbc       pbc;
 168     real        dvdl_dum[efptNR], dvdl_nb[efptNR];
 169
 170 #if GMX_MPI
 171     double  t0 = 0.0, t1, t2, t3; /* time measurement for coarse load balancing */
 172 #endif
 173
 174     set_pbc(&pbc, fr->ePBC, box);
 175
 176     /* reset free energy components */
 177     for (i = 0; i < efptNR; i++)
 178     {
 179         dvdl_nb[i]  = 0;
 180         dvdl_dum[i] = 0;
 181     }
 182
 183     /* do QMMM first if requested */
 184     if (fr->bQMMM)
 185     {
 186         enerd->term[F_EQM] = calculate_QMMM(cr, forceForUseWithShiftForces, fr);
 187     }
 188
 189     /* Call the short range functions all in one go. */
 190
 191 #if GMX_MPI
 192     /*#define TAKETIME ((cr->npmenodes) && (fr->timesteps < 12))*/
 193 #define TAKETIME FALSE
 194     if (TAKETIME)
 195     {
 196         MPI_Barrier(cr->mpi_comm_mygroup);
 197         t0 = MPI_Wtime();
 198     }
 199 #endif
 200
 201     if (ir->nwall)
 202     {
 203         /* foreign lambda component for walls */
 204         real dvdl_walls = do_walls(ir, fr, box, md, x, forceForUseWithShiftForces, lambda[efptVDW],
 205                                    enerd->grpp.ener[egLJSR], nrnb);
 206         enerd->dvdl_lin[efptVDW] += dvdl_walls;
 207     }
 208
 209     /* We only do non-bonded calculation with group scheme here, the verlet
 210      * calls are done from do_force_cutsVERLET(). */
 211     if (fr->cutoff_scheme == ecutsGROUP && (flags & GMX_FORCE_NONBONDED))
 212     {
 213         donb_flags = 0;
 214         /* Add short-range interactions */
 215         donb_flags |= GMX_NONBONDED_DO_SR;
 216
 217         /* Currently all group scheme kernels always calculate (shift-)forces */
 218         if (flags & GMX_FORCE_FORCES)
 219         {
 220             donb_flags |= GMX_NONBONDED_DO_FORCE;
 221         }
 222         if (flags & GMX_FORCE_VIRIAL)
 223         {
 224             donb_flags |= GMX_NONBONDED_DO_SHIFTFORCE;
 225         }
 226         if (flags & GMX_FORCE_ENERGY)
 227         {
 228             donb_flags |= GMX_NONBONDED_DO_POTENTIAL;
 229         }
 230
 231         wallcycle_sub_start(wcycle, ewcsNONBONDED);
 232         do_nonbonded(fr, x, forceForUseWithShiftForces, md, excl,
 233                      &enerd->grpp, nrnb,
 234                      lambda, dvdl_nb, -1, -1, donb_flags);
 235
 236         /* If we do foreign lambda and we have soft-core interactions
 237          * we have to recalculate the (non-linear) energies contributions.
 238          */
 239         if (fepvals->n_lambda > 0 && (flags & GMX_FORCE_DHDL) && fepvals->sc_alpha != 0)
 240         {
 241             for (i = 0; i < enerd->n_lambda; i++)
 242             {
 243                 real lam_i[efptNR];
 244
 245                 for (j = 0; j < efptNR; j++)
 246                 {
 247                     lam_i[j] = (i == 0 ? lambda[j] : fepvals->all_lambda[j][i-1]);
 248                 }
 249                 reset_foreign_enerdata(enerd);
 250                 do_nonbonded(fr, x, forceForUseWithShiftForces, md, excl,
 251                              &(enerd->foreign_grpp), nrnb,
 252                              lam_i, dvdl_dum, -1, -1,
 253                              (donb_flags & ~GMX_NONBONDED_DO_FORCE) | GMX_NONBONDED_DO_FOREIGNLAMBDA);
 254                 sum_epot(&(enerd->foreign_grpp), enerd->foreign_term);
 255                 enerd->enerpart_lambda[i] += enerd->foreign_term[F_EPOT];
 256             }
 257         }
 258         wallcycle_sub_stop(wcycle, ewcsNONBONDED);
 259     }
 260
 261 #if GMX_MPI
 262     if (TAKETIME)
 263     {
 264         t1          = MPI_Wtime();
 265         fr->t_fnbf += t1-t0;
 266     }
 267 #endif
 268
 269     if (fepvals->sc_alpha != 0)
 270     {
 271         enerd->dvdl_nonlin[efptVDW] += dvdl_nb[efptVDW];
 272     }
 273     else
 274     {
 275         enerd->dvdl_lin[efptVDW] += dvdl_nb[efptVDW];
 276     }
 277
 278     if (fepvals->sc_alpha != 0)
 279
 280     /* even though coulomb part is linear, we already added it, beacuse we
 281        need to go through the vdw calculation anyway */
 282     {
 283         enerd->dvdl_nonlin[efptCOUL] += dvdl_nb[efptCOUL];
 284     }
 285     else
 286     {
 287         enerd->dvdl_lin[efptCOUL] += dvdl_nb[efptCOUL];
 288     }
 289
 290     if (debug)
 291     {
 292         pr_rvecs(debug, 0, "fshift after SR", fr->fshift, SHIFTS);
 293     }
 294
 295     /* Shift the coordinates. Must be done before listed forces and PPPM,
 296      * but is also necessary for SHAKE and update, therefore it can NOT
 297      * go when no listed forces have to be evaluated.
 298      *
 299      * The shifting and PBC code is deliberately not timed, since with
 300      * the Verlet scheme it only takes non-zero time with triclinic
 301      * boxes, and even then the time is around a factor of 100 less
 302      * than the next smallest counter.
 303      */
 304
 305
 306     /* Here sometimes we would not need to shift with NBFonly,
 307      * but we do so anyhow for consistency of the returned coordinates.
 308      */
 309     if (graph)
 310     {
 311         shift_self(graph, box, x);
 312         if (TRICLINIC(box))
 313         {
 314             inc_nrnb(nrnb, eNR_SHIFTX, 2*graph->nnodes);
 315         }
 316         else
 317         {
 318             inc_nrnb(nrnb, eNR_SHIFTX, graph->nnodes);
 319         }
 320     }
 321     /* Check whether we need to do listed interactions or correct for exclusions */
 322     if (fr->bMolPBC &&
 323         ((flags & GMX_FORCE_LISTED)
 324          || EEL_RF(fr->ic->eeltype) || EEL_FULL(fr->ic->eeltype) || EVDW_PME(fr->ic->vdwtype)))
 325     {
 326         /* TODO There are no electrostatics methods that require this
 327            transformation, when using the Verlet scheme, so update the
 328            above conditional. */
 329         /* Since all atoms are in the rectangular or triclinic unit-cell,
 330          * only single box vector shifts (2 in x) are required.
 331          */
 332         set_pbc_dd(&pbc, fr->ePBC, DOMAINDECOMP(cr) ? cr->dd->nc : nullptr,
 333                    TRUE, box);
 334     }
 335
 336     do_force_listed(wcycle, box, ir->fepvals, cr, ms,
 337                     idef, x, hist,
 338                     forceForUseWithShiftForces, forceWithVirial,
 339                     fr, &pbc, graph, enerd, nrnb, lambda, md, fcd,
 340                     DOMAINDECOMP(cr) ? cr->dd->globalAtomIndices.data() : nullptr,
 341                     flags);
 342
 343
 344     *cycles_pme = 0;
 345
 346     /* Do long-range electrostatics and/or LJ-PME, including related short-range
 347      * corrections.
 348      */
 349     if (EEL_FULL(fr->ic->eeltype) || EVDW_PME(fr->ic->vdwtype))
 350     {
 351         int  status            = 0;
 352         real Vlr_q             = 0, Vlr_lj = 0;
 353
 354         /* We reduce all virial, dV/dlambda and energy contributions, except
 355          * for the reciprocal energies (Vlr_q, Vlr_lj) into the same struct.
 356          */
 357         ewald_corr_thread_t &ewaldOutput = fr->ewc_t[0];
 358         clearEwaldThreadOutput(&ewaldOutput);
 359
 360         if (EEL_PME_EWALD(fr->ic->eeltype) || EVDW_PME(fr->ic->vdwtype))
 361         {
 362             /* With the Verlet scheme exclusion forces are calculated
 363              * in the non-bonded kernel.
 364              */
 365             /* The TPI molecule does not have exclusions with the rest
 366              * of the system and no intra-molecular PME grid
 367              * contributions will be calculated in
 368              * gmx_pme_calc_energy.
 369              */
 370             if ((ir->cutoff_scheme == ecutsGROUP && fr->n_tpi == 0) ||
 371                 ir->ewald_geometry != eewg3D ||
 372                 ir->epsilon_surface != 0)
 373             {
 374                 int nthreads, t;
 375
 376                 wallcycle_sub_start(wcycle, ewcsEWALD_CORRECTION);
 377
 378                 if (fr->n_tpi > 0)
 379                 {
 380                     gmx_fatal(FARGS, "TPI with PME currently only works in a 3D geometry with tin-foil boundary conditions");
 381                 }
 382
 383                 nthreads = fr->nthread_ewc;
 384 #pragma omp parallel for num_threads(nthreads) schedule(static)
 385                 for (t = 0; t < nthreads; t++)
 386                 {
 387                     try
 388                     {
 389                         ewald_corr_thread_t &ewc_t = fr->ewc_t[t];
 390                         if (t > 0)
 391                         {
 392                             clearEwaldThreadOutput(&ewc_t);
 393                         }
 394
 395                         /* Threading is only supported with the Verlet cut-off
 396                          * scheme and then only single particle forces (no
 397                          * exclusion forces) are calculated, so we can store
 398                          * the forces in the normal, single forceWithVirial->force_ array.
 399                          */
 400                         ewald_LRcorrection(md->homenr, cr, nthreads, t, fr, ir,
 401                                            md->chargeA, md->chargeB,
 402                                            md->sqrt_c6A, md->sqrt_c6B,
 403                                            md->sigmaA, md->sigmaB,
 404                                            md->sigma3A, md->sigma3B,
 405                                            md->nChargePerturbed || md->nTypePerturbed,
 406                                            ir->cutoff_scheme != ecutsVERLET,
 407                                            excl, x, box, mu_tot,
 408                                            ir->ewald_geometry,
 409                                            ir->epsilon_surface,
 410                                            as_rvec_array(forceWithVirial->force_.data()),
 411                                            ewc_t.vir_q, ewc_t.vir_lj,
 412                                            &ewc_t.Vcorr_q, &ewc_t.Vcorr_lj,
 413                                            lambda[efptCOUL], lambda[efptVDW],
 414                                            &ewc_t.dvdl[efptCOUL], &ewc_t.dvdl[efptVDW]);
 415                     }
 416                     GMX_CATCH_ALL_AND_EXIT_WITH_FATAL_ERROR;
 417                 }
 418                 if (nthreads > 1)
 419                 {
 420                     reduceEwaldThreadOuput(nthreads, fr->ewc_t);
 421                 }
 422                 wallcycle_sub_stop(wcycle, ewcsEWALD_CORRECTION);
 423             }
 424
 425             if (EEL_PME_EWALD(fr->ic->eeltype) && fr->n_tpi == 0)
 426             {
 427                 /* This is not in a subcounter because it takes a
 428                    negligible and constant-sized amount of time */
 429                 ewaldOutput.Vcorr_q +=
 430                     ewald_charge_correction(cr, fr, lambda[efptCOUL], box,
 431                                             &ewaldOutput.dvdl[efptCOUL],
 432                                             ewaldOutput.vir_q);
 433             }
 434
 435             if ((EEL_PME(fr->ic->eeltype) || EVDW_PME(fr->ic->vdwtype)) &&
 436                 thisRankHasDuty(cr, DUTY_PME) && (pme_run_mode(fr->pmedata) == PmeRunMode::CPU))
 437             {
 438                 /* Do reciprocal PME for Coulomb and/or LJ. */
 439                 assert(fr->n_tpi >= 0);
 440                 if (fr->n_tpi == 0 || (flags & GMX_FORCE_STATECHANGED))
 441                 {
 442                     pme_flags = GMX_PME_SPREAD | GMX_PME_SOLVE;
 443
 444                     if (flags & GMX_FORCE_FORCES)
 445                     {
 446                         pme_flags |= GMX_PME_CALC_F;
 447                     }
 448                     if (flags & GMX_FORCE_VIRIAL)
 449                     {
 450                         pme_flags |= GMX_PME_CALC_ENER_VIR;
 451                     }
 452                     if (fr->n_tpi > 0)
 453                     {
 454                         /* We don't calculate f, but we do want the potential */
 455                         pme_flags |= GMX_PME_CALC_POT;
 456                     }
 457
 458                     /* With domain decomposition we close the CPU side load
 459                      * balancing region here, because PME does global
 460                      * communication that acts as a global barrier.
 461                      */
 462                     if (DOMAINDECOMP(cr))
 463                     {
 464                         ddCloseBalanceRegionCpu(cr->dd);
 465                     }
 466
 467                     wallcycle_start(wcycle, ewcPMEMESH);
 468                     status = gmx_pme_do(fr->pmedata,
 469                                         0, md->homenr - fr->n_tpi,
 470                                         x,
 471                                         as_rvec_array(forceWithVirial->force_.data()),
 472                                         md->chargeA, md->chargeB,
 473                                         md->sqrt_c6A, md->sqrt_c6B,
 474                                         md->sigmaA, md->sigmaB,
 475                                         box, cr,
 476                                         DOMAINDECOMP(cr) ? dd_pme_maxshift_x(cr->dd) : 0,
 477                                         DOMAINDECOMP(cr) ? dd_pme_maxshift_y(cr->dd) : 0,
 478                                         nrnb, wcycle,
 479                                         ewaldOutput.vir_q, ewaldOutput.vir_lj,
 480                                         &Vlr_q, &Vlr_lj,
 481                                         lambda[efptCOUL], lambda[efptVDW],
 482                                         &ewaldOutput.dvdl[efptCOUL],
 483                                         &ewaldOutput.dvdl[efptVDW],
 484                                         pme_flags);
 485                     *cycles_pme = wallcycle_stop(wcycle, ewcPMEMESH);
 486                     if (status != 0)
 487                     {
 488                         gmx_fatal(FARGS, "Error %d in reciprocal PME routine", status);
 489                     }
 490
 491                     /* We should try to do as little computation after
 492                      * this as possible, because parallel PME synchronizes
 493                      * the nodes, so we want all load imbalance of the
 494                      * rest of the force calculation to be before the PME
 495                      * call.  DD load balancing is done on the whole time
 496                      * of the force call (without PME).
 497                      */
 498                 }
 499                 if (fr->n_tpi > 0)
 500                 {
 501                     if (EVDW_PME(ir->vdwtype))
 502                     {
 503
 504                         gmx_fatal(FARGS, "Test particle insertion not implemented with LJ-PME");
 505                     }
 506                     /* Determine the PME grid energy of the test molecule
 507                      * with the PME grid potential of the other charges.
 508                      */
 509                     gmx_pme_calc_energy(fr->pmedata, fr->n_tpi,
 510                                         x + md->homenr - fr->n_tpi,
 511                                         md->chargeA + md->homenr - fr->n_tpi,
 512                                         &Vlr_q);
 513                 }
 514             }
 515         }
 516
 517         if (!EEL_PME(fr->ic->eeltype) && EEL_PME_EWALD(fr->ic->eeltype))
 518         {
 519             Vlr_q = do_ewald(ir, x, as_rvec_array(forceWithVirial->force_.data()),
 520                              md->chargeA, md->chargeB,
 521                              box, cr, md->homenr,
 522                              ewaldOutput.vir_q, fr->ic->ewaldcoeff_q,
 523                              lambda[efptCOUL], &ewaldOutput.dvdl[efptCOUL],
 524                              fr->ewald_table);
 525         }
 526
 527         /* Note that with separate PME nodes we get the real energies later */
 528         forceWithVirial->addVirialContribution(ewaldOutput.vir_q);
 529         forceWithVirial->addVirialContribution(ewaldOutput.vir_lj);
 530         enerd->dvdl_lin[efptCOUL] += ewaldOutput.dvdl[efptCOUL];
 531         enerd->dvdl_lin[efptVDW]  += ewaldOutput.dvdl[efptVDW];
 532         enerd->term[F_COUL_RECIP]  = Vlr_q + ewaldOutput.Vcorr_q;
 533         enerd->term[F_LJ_RECIP]    = Vlr_lj + ewaldOutput.Vcorr_lj;
 534
 535         if (debug)
 536         {
 537             fprintf(debug, "Vlr_q = %g, Vcorr_q = %g, Vlr_corr_q = %g\n",
 538                     Vlr_q, ewaldOutput.Vcorr_q, enerd->term[F_COUL_RECIP]);
 539             pr_rvecs(debug, 0, "vir_el_recip after corr", ewaldOutput.vir_q, DIM);
 540             pr_rvecs(debug, 0, "fshift after LR Corrections", fr->fshift, SHIFTS);
 541             fprintf(debug, "Vlr_lj: %g, Vcorr_lj = %g, Vlr_corr_lj = %g\n",
 542                     Vlr_lj, ewaldOutput.Vcorr_lj, enerd->term[F_LJ_RECIP]);
 543             pr_rvecs(debug, 0, "vir_lj_recip after corr", ewaldOutput.vir_lj, DIM);
 544         }
 545     }
 546     else
 547     {
 548         /* Is there a reaction-field exclusion correction needed?
 549          * With the Verlet scheme, exclusion forces are calculated
 550          * in the non-bonded kernel.
 551          */
 552         if (ir->cutoff_scheme != ecutsVERLET && EEL_RF(fr->ic->eeltype))
 553         {
 554             real dvdl_rf_excl      = 0;
 555             enerd->term[F_RF_EXCL] =
 556                 RF_excl_correction(fr, graph, md, excl, DOMAINDECOMP(cr),
 557                                    x, forceForUseWithShiftForces,
 558                                    fr->fshift, &pbc, lambda[efptCOUL], &dvdl_rf_excl);
 559
 560             enerd->dvdl_lin[efptCOUL] += dvdl_rf_excl;
 561         }
 562     }
 563
 564     if (debug)
 565     {
 566         print_nrnb(debug, nrnb);
 567     }
 568
 569 #if GMX_MPI
 570     if (TAKETIME)
 571     {
 572         t2 = MPI_Wtime();
 573         MPI_Barrier(cr->mpi_comm_mygroup);
 574         t3          = MPI_Wtime();
 575         fr->t_wait += t3-t2;
 576         if (fr->timesteps == 11)
 577         {
 578             char buf[22];
 579             fprintf(stderr, "* PP load balancing info: rank %d, step %s, rel wait time=%3.0f%% , load string value: %7.2f\n",
 580                     cr->nodeid, gmx_step_str(fr->timesteps, buf),
 581                     100*fr->t_wait/(fr->t_wait+fr->t_fnbf),
 582                     (fr->t_fnbf+fr->t_wait)/fr->t_fnbf);
 583         }
 584         fr->timesteps++;
 585     }
 586 #endif
 587
 588     if (debug)
 589     {
 590         pr_rvecs(debug, 0, "fshift after bondeds", fr->fshift, SHIFTS);
 591     }
 592
 593 }
 594
 595 void init_enerdata(int ngener, int n_lambda, gmx_enerdata_t *enerd)
 596 {
 597     int i, n2;
 598
 599     for (i = 0; i < F_NRE; i++)
 600     {
 601         enerd->term[i]         = 0;
 602         enerd->foreign_term[i] = 0;
 603     }
 604
 605
 606     for (i = 0; i < efptNR; i++)
 607     {
 608         enerd->dvdl_lin[i]     = 0;
 609         enerd->dvdl_nonlin[i]  = 0;
 610     }
 611
 612     n2 = ngener*ngener;
 613     if (debug)
 614     {
 615         fprintf(debug, "Creating %d sized group matrix for energies\n", n2);
 616     }
 617     enerd->grpp.nener         = n2;
 618     enerd->foreign_grpp.nener = n2;
 619     for (i = 0; (i < egNR); i++)
 620     {
 621         snew(enerd->grpp.ener[i], n2);
 622         snew(enerd->foreign_grpp.ener[i], n2);
 623     }
 624
 625     if (n_lambda)
 626     {
 627         enerd->n_lambda = 1 + n_lambda;
 628         snew(enerd->enerpart_lambda, enerd->n_lambda);
 629     }
 630     else
 631     {
 632         enerd->n_lambda = 0;
 633     }
 634 }
 635
 636 void destroy_enerdata(gmx_enerdata_t *enerd)
 637 {
 638     int i;
 639
 640     for (i = 0; (i < egNR); i++)
 641     {
 642         sfree(enerd->grpp.ener[i]);
 643     }
 644
 645     for (i = 0; (i < egNR); i++)
 646     {
 647         sfree(enerd->foreign_grpp.ener[i]);
 648     }
 649
 650     if (enerd->n_lambda)
 651     {
 652         sfree(enerd->enerpart_lambda);
 653     }
 654 }
 655
 656 static real sum_v(int n, const real v[])
 657 {
 658     real t;
 659     int  i;
 660
 661     t = 0.0;
 662     for (i = 0; (i < n); i++)
 663     {
 664         t = t + v[i];
 665     }
 666
 667     return t;
 668 }
 669
 670 void sum_epot(gmx_grppairener_t *grpp, real *epot)
 671 {
 672     int i;
 673
 674     /* Accumulate energies */
 675     epot[F_COUL_SR]  = sum_v(grpp->nener, grpp->ener[egCOULSR]);
 676     epot[F_LJ]       = sum_v(grpp->nener, grpp->ener[egLJSR]);
 677     epot[F_LJ14]     = sum_v(grpp->nener, grpp->ener[egLJ14]);
 678     epot[F_COUL14]   = sum_v(grpp->nener, grpp->ener[egCOUL14]);
 679
 680 /* lattice part of LR doesnt belong to any group
 681  * and has been added earlier
 682  */
 683     epot[F_BHAM]     = sum_v(grpp->nener, grpp->ener[egBHAMSR]);
 684
 685     epot[F_EPOT] = 0;
 686     for (i = 0; (i < F_EPOT); i++)
 687     {
 688         if (i != F_DISRESVIOL && i != F_ORIRESDEV)
 689         {
 690             epot[F_EPOT] += epot[i];
 691         }
 692     }
 693 }
 694
 695 void sum_dhdl(gmx_enerdata_t *enerd, gmx::ArrayRef<const real> lambda, t_lambda *fepvals)
 696 {
 697     int    index;
 698     double dlam;
 699
 700     enerd->dvdl_lin[efptVDW] += enerd->term[F_DVDL_VDW];  /* include dispersion correction */
 701     enerd->term[F_DVDL]       = 0.0;
 702     for (int i = 0; i < efptNR; i++)
 703     {
 704         if (fepvals->separate_dvdl[i])
 705         {
 706             /* could this be done more readably/compactly? */
 707             switch (i)
 708             {
 709                 case (efptMASS):
 710                     index = F_DKDL;
 711                     break;
 712                 case (efptCOUL):
 713                     index = F_DVDL_COUL;
 714                     break;
 715                 case (efptVDW):
 716                     index = F_DVDL_VDW;
 717                     break;
 718                 case (efptBONDED):
 719                     index = F_DVDL_BONDED;
 720                     break;
 721                 case (efptRESTRAINT):
 722                     index = F_DVDL_RESTRAINT;
 723                     break;
 724                 default:
 725                     index = F_DVDL;
 726                     break;
 727             }
 728             enerd->term[index] = enerd->dvdl_lin[i] + enerd->dvdl_nonlin[i];
 729             if (debug)
 730             {
 731                 fprintf(debug, "dvdl-%s[%2d]: %f: non-linear %f + linear %f\n",
 732                         efpt_names[i], i, enerd->term[index], enerd->dvdl_nonlin[i], enerd->dvdl_lin[i]);
 733             }
 734         }
 735         else
 736         {
 737             enerd->term[F_DVDL] += enerd->dvdl_lin[i] + enerd->dvdl_nonlin[i];
 738             if (debug)
 739             {
 740                 fprintf(debug, "dvd-%sl[%2d]: %f: non-linear %f + linear %f\n",
 741                         efpt_names[0], i, enerd->term[F_DVDL], enerd->dvdl_nonlin[i], enerd->dvdl_lin[i]);
 742             }
 743         }
 744     }
 745
 746     /* Notes on the foreign lambda free energy difference evaluation:
 747      * Adding the potential and ekin terms that depend linearly on lambda
 748      * as delta lam * dvdl to the energy differences is exact.
 749      * For the constraints this is not exact, but we have no other option
 750      * without literally changing the lengths and reevaluating the energies at each step.
 751      * (try to remedy this post 4.6 - MRS)
 752      */
 753     if (fepvals->separate_dvdl[efptBONDED])
 754     {
 755         enerd->term[F_DVDL_BONDED] += enerd->term[F_DVDL_CONSTR];
 756     }
 757     else
 758     {
 759         enerd->term[F_DVDL] += enerd->term[F_DVDL_CONSTR];
 760     }
 761     enerd->term[F_DVDL_CONSTR] = 0;
 762
 763     for (int i = 0; i < fepvals->n_lambda; i++)
 764     {
 765         /* note we are iterating over fepvals here!
 766            For the current lam, dlam = 0 automatically,
 767            so we don't need to add anything to the
 768            enerd->enerpart_lambda[0] */
 769
 770         /* we don't need to worry about dvdl_lin contributions to dE at
 771            current lambda, because the contributions to the current
 772            lambda are automatically zeroed */
 773
 774         for (gmx::index j = 0; j < lambda.size(); j++)
 775         {
 776             /* Note that this loop is over all dhdl components, not just the separated ones */
 777             dlam = (fepvals->all_lambda[j][i] - lambda[j]);
 778             enerd->enerpart_lambda[i+1] += dlam*enerd->dvdl_lin[j];
 779             if (debug)
 780             {
 781                 fprintf(debug, "enerdiff lam %g: (%15s), non-linear %f linear %f*%f\n",
 782                         fepvals->all_lambda[j][i], efpt_names[j],
 783                         (enerd->enerpart_lambda[i+1] - enerd->enerpart_lambda[0]),
 784                         dlam, enerd->dvdl_lin[j]);
 785             }
 786         }
 787     }
 788 }
 789
 790
 791 void reset_foreign_enerdata(gmx_enerdata_t *enerd)
 792 {
 793     int  i, j;
 794
 795     /* First reset all foreign energy components.  Foreign energies always called on
 796        neighbor search steps */
 797     for (i = 0; (i < egNR); i++)
 798     {
 799         for (j = 0; (j < enerd->grpp.nener); j++)
 800         {
 801             enerd->foreign_grpp.ener[i][j] = 0.0;
 802         }
 803     }
 804
 805     /* potential energy components */
 806     for (i = 0; (i <= F_EPOT); i++)
 807     {
 808         enerd->foreign_term[i] = 0.0;
 809     }
 810 }
 811
 812 void reset_enerdata(gmx_enerdata_t *enerd)
 813 {
 814     int      i, j;
 815
 816     /* First reset all energy components. */
 817     for (i = 0; (i < egNR); i++)
 818     {
 819         for (j = 0; (j < enerd->grpp.nener); j++)
 820         {
 821             enerd->grpp.ener[i][j] = 0.0;
 822         }
 823     }
 824     for (i = 0; i < efptNR; i++)
 825     {
 826         enerd->dvdl_lin[i]    = 0.0;
 827         enerd->dvdl_nonlin[i] = 0.0;
 828     }
 829
 830     /* Normal potential energy components */
 831     for (i = 0; (i <= F_EPOT); i++)
 832     {
 833         enerd->term[i] = 0.0;
 834     }
 835     enerd->term[F_DVDL]            = 0.0;
 836     enerd->term[F_DVDL_COUL]       = 0.0;
 837     enerd->term[F_DVDL_VDW]        = 0.0;
 838     enerd->term[F_DVDL_BONDED]     = 0.0;
 839     enerd->term[F_DVDL_RESTRAINT]  = 0.0;
 840     enerd->term[F_DKDL]            = 0.0;
 841     if (enerd->n_lambda > 0)
 842     {
 843         for (i = 0; i < enerd->n_lambda; i++)
 844         {
 845             enerd->enerpart_lambda[i] = 0.0;
 846         }
 847     }
 848     /* reset foreign energy data - separate function since we also call it elsewhere */
 849     reset_foreign_enerdata(enerd);
 850 }