src/gromacs/mdlib/force.c

   1 /* -*- mode: c; tab-width: 4; indent-tabs-mode: nil; c-basic-offset: 4; c-file-style: "stroustrup"; -*-
   2  *
   3  *
   4  *                This source code is part of
   5  *
   6  *                 G   R   O   M   A   C   S
   7  *
   8  *          GROningen MAchine for Chemical Simulations
   9  *
  10  *                        VERSION 3.2.0
  11  * Written by David van der Spoel, Erik Lindahl, Berk Hess, and others.
  12  * Copyright (c) 1991-2000, University of Groningen, The Netherlands.
  13  * Copyright (c) 2001-2004, The GROMACS development team,
  14  * check out http://www.gromacs.org for more information.
  15
  16  * This program is free software; you can redistribute it and/or
  17  * modify it under the terms of the GNU General Public License
  18  * as published by the Free Software Foundation; either version 2
  19  * of the License, or (at your option) any later version.
  20  *
  21  * If you want to redistribute modifications, please consider that
  22  * scientific software is very special. Version control is crucial -
  23  * bugs must be traceable. We will be happy to consider code for
  24  * inclusion in the official distribution, but derived work must not
  25  * be called official GROMACS. Details are found in the README & COPYING
  26  * files - if they are missing, get the official version at www.gromacs.org.
  27  *
  28  * To help us fund GROMACS development, we humbly ask that you cite
  29  * the papers on the package - you can find them in the top README file.
  30  *
  31  * For more info, check our website at http://www.gromacs.org
  32  *
  33  * And Hey:
  34  * GROwing Monsters And Cloning Shrimps
  35  */
  36 #ifdef HAVE_CONFIG_H
  37 #include <config.h>
  38 #endif
  39
  40 #include <math.h>
  41 #include <string.h>
  42 #include <assert.h>
  43 #include "sysstuff.h"
  44 #include "typedefs.h"
  45 #include "macros.h"
  46 #include "smalloc.h"
  47 #include "macros.h"
  48 #include "physics.h"
  49 #include "force.h"
  50 #include "nonbonded.h"
  51 #include "names.h"
  52 #include "network.h"
  53 #include "pbc.h"
  54 #include "ns.h"
  55 #include "nrnb.h"
  56 #include "bondf.h"
  57 #include "mshift.h"
  58 #include "txtdump.h"
  59 #include "coulomb.h"
  60 #include "pme.h"
  61 #include "mdrun.h"
  62 #include "domdec.h"
  63 #include "partdec.h"
  64 #include "qmmm.h"
  65 #include "gmx_omp_nthreads.h"
  66
  67
  68 void ns(FILE              *fp,
  69         t_forcerec        *fr,
  70         rvec               x[],
  71         matrix             box,
  72         gmx_groups_t      *groups,
  73         t_grpopts         *opts,
  74         gmx_localtop_t    *top,
  75         t_mdatoms         *md,
  76         t_commrec         *cr,
  77         t_nrnb            *nrnb,
  78         real              *lambda,
  79         real              *dvdlambda,
  80         gmx_grppairener_t *grppener,
  81         gmx_bool           bFillGrid,
  82         gmx_bool           bDoLongRangeNS)
  83 {
  84     char   *ptr;
  85     int     nsearch;
  86
  87
  88     if (!fr->ns.nblist_initialized)
  89     {
  90         init_neighbor_list(fp, fr, md->homenr);
  91     }
  92
  93     if (fr->bTwinRange)
  94     {
  95         fr->nlr = 0;
  96     }
  97
  98     nsearch = search_neighbours(fp, fr, x, box, top, groups, cr, nrnb, md,
  99                                 lambda, dvdlambda, grppener,
 100                                 bFillGrid, bDoLongRangeNS, TRUE);
 101     if (debug)
 102     {
 103         fprintf(debug, "nsearch = %d\n", nsearch);
 104     }
 105
 106     /* Check whether we have to do dynamic load balancing */
 107     /*if ((nsb->nstDlb > 0) && (mod(step,nsb->nstDlb) == 0))
 108        count_nb(cr,nsb,&(top->blocks[ebCGS]),nns,fr->nlr,
 109        &(top->idef),opts->ngener);
 110      */
 111     if (fr->ns.dump_nl > 0)
 112     {
 113         dump_nblist(fp, cr, fr, fr->ns.dump_nl);
 114     }
 115 }
 116
 117 static void reduce_thread_forces(int n, rvec *f,
 118                                  tensor vir,
 119                                  real *Vcorr,
 120                                  int efpt_ind, real *dvdl,
 121                                  int nthreads, f_thread_t *f_t)
 122 {
 123     int t, i;
 124
 125     /* This reduction can run over any number of threads */
 126 #pragma omp parallel for num_threads(gmx_omp_nthreads_get(emntBonded)) private(t) schedule(static)
 127     for (i = 0; i < n; i++)
 128     {
 129         for (t = 1; t < nthreads; t++)
 130         {
 131             rvec_inc(f[i], f_t[t].f[i]);
 132         }
 133     }
 134     for (t = 1; t < nthreads; t++)
 135     {
 136         *Vcorr += f_t[t].Vcorr;
 137         *dvdl  += f_t[t].dvdl[efpt_ind];
 138         m_add(vir, f_t[t].vir, vir);
 139     }
 140 }
 141
 142 void do_force_lowlevel(FILE       *fplog,   gmx_large_int_t step,
 143                        t_forcerec *fr,      t_inputrec *ir,
 144                        t_idef     *idef,    t_commrec  *cr,
 145                        t_nrnb     *nrnb,    gmx_wallcycle_t wcycle,
 146                        t_mdatoms  *md,
 147                        t_grpopts  *opts,
 148                        rvec       x[],      history_t  *hist,
 149                        rvec       f[],
 150                        rvec       f_longrange[],
 151                        gmx_enerdata_t *enerd,
 152                        t_fcdata   *fcd,
 153                        gmx_mtop_t     *mtop,
 154                        gmx_localtop_t *top,
 155                        gmx_genborn_t *born,
 156                        t_atomtypes *atype,
 157                        gmx_bool       bBornRadii,
 158                        matrix     box,
 159                        t_lambda   *fepvals,
 160                        real       *lambda,
 161                        t_graph    *graph,
 162                        t_blocka   *excl,
 163                        rvec       mu_tot[],
 164                        int        flags,
 165                        float      *cycles_pme)
 166 {
 167     int         i, j, status;
 168     int         donb_flags;
 169     gmx_bool    bDoEpot, bSepDVDL, bSB;
 170     int         pme_flags;
 171     matrix      boxs;
 172     rvec        box_size;
 173     real        Vsr, Vlr, Vcorr = 0;
 174     t_pbc       pbc;
 175     real        dvdgb;
 176     char        buf[22];
 177     double      clam_i, vlam_i;
 178     real        dvdl_dum[efptNR], dvdl, dvdl_nb[efptNR], lam_i[efptNR];
 179     real        dvdlsum;
 180
 181 #ifdef GMX_MPI
 182     double  t0 = 0.0, t1, t2, t3; /* time measurement for coarse load balancing */
 183 #endif
 184
 185 #define PRINT_SEPDVDL(s, v, dvdlambda) if (bSepDVDL) {fprintf(fplog, sepdvdlformat, s, v, dvdlambda); }
 186
 187
 188     set_pbc(&pbc, fr->ePBC, box);
 189
 190     /* reset free energy components */
 191     for (i = 0; i < efptNR; i++)
 192     {
 193         dvdl_nb[i]  = 0;
 194         dvdl_dum[i] = 0;
 195     }
 196
 197     /* Reset box */
 198     for (i = 0; (i < DIM); i++)
 199     {
 200         box_size[i] = box[i][i];
 201     }
 202
 203     bSepDVDL = (fr->bSepDVDL && do_per_step(step, ir->nstlog));
 204     debug_gmx();
 205
 206     /* do QMMM first if requested */
 207     if (fr->bQMMM)
 208     {
 209         enerd->term[F_EQM] = calculate_QMMM(cr, x, f, fr, md);
 210     }
 211
 212     if (bSepDVDL)
 213     {
 214         fprintf(fplog, "Step %s: non-bonded V and dVdl for node %d:\n",
 215                 gmx_step_str(step, buf), cr->nodeid);
 216     }
 217
 218     /* Call the short range functions all in one go. */
 219
 220 #ifdef GMX_MPI
 221     /*#define TAKETIME ((cr->npmenodes) && (fr->timesteps < 12))*/
 222 #define TAKETIME FALSE
 223     if (TAKETIME)
 224     {
 225         MPI_Barrier(cr->mpi_comm_mygroup);
 226         t0 = MPI_Wtime();
 227     }
 228 #endif
 229
 230     if (ir->nwall)
 231     {
 232         /* foreign lambda component for walls */
 233         dvdl = do_walls(ir, fr, box, md, x, f, lambda[efptVDW],
 234                         enerd->grpp.ener[egLJSR], nrnb);
 235         PRINT_SEPDVDL("Walls", 0.0, dvdl);
 236         enerd->dvdl_lin[efptVDW] += dvdl;
 237     }
 238
 239     /* If doing GB, reset dvda and calculate the Born radii */
 240     if (ir->implicit_solvent)
 241     {
 242         wallcycle_sub_start(wcycle, ewcsNONBONDED);
 243
 244         for (i = 0; i < born->nr; i++)
 245         {
 246             fr->dvda[i] = 0;
 247         }
 248
 249         if (bBornRadii)
 250         {
 251             calc_gb_rad(cr, fr, ir, top, atype, x, &(fr->gblist), born, md, nrnb);
 252         }
 253
 254         wallcycle_sub_stop(wcycle, ewcsNONBONDED);
 255     }
 256
 257     where();
 258     /* We only do non-bonded calculation with group scheme here, the verlet
 259      * calls are done from do_force_cutsVERLET(). */
 260     if (fr->cutoff_scheme == ecutsGROUP && (flags & GMX_FORCE_NONBONDED))
 261     {
 262         donb_flags = 0;
 263         /* Add short-range interactions */
 264         donb_flags |= GMX_NONBONDED_DO_SR;
 265
 266         if (flags & GMX_FORCE_FORCES)
 267         {
 268             donb_flags |= GMX_NONBONDED_DO_FORCE;
 269         }
 270         if (flags & GMX_FORCE_ENERGY)
 271         {
 272             donb_flags |= GMX_NONBONDED_DO_POTENTIAL;
 273         }
 274         if (flags & GMX_FORCE_DO_LR)
 275         {
 276             donb_flags |= GMX_NONBONDED_DO_LR;
 277         }
 278
 279         wallcycle_sub_start(wcycle, ewcsNONBONDED);
 280         do_nonbonded(cr, fr, x, f, f_longrange, md, excl,
 281                      &enerd->grpp, box_size, nrnb,
 282                      lambda, dvdl_nb, -1, -1, donb_flags);
 283
 284         /* If we do foreign lambda and we have soft-core interactions
 285          * we have to recalculate the (non-linear) energies contributions.
 286          */
 287         if (fepvals->n_lambda > 0 && (flags & GMX_FORCE_DHDL) && fepvals->sc_alpha != 0)
 288         {
 289             for (i = 0; i < enerd->n_lambda; i++)
 290             {
 291                 for (j = 0; j < efptNR; j++)
 292                 {
 293                     lam_i[j] = (i == 0 ? lambda[j] : fepvals->all_lambda[j][i-1]);
 294                 }
 295                 reset_foreign_enerdata(enerd);
 296                 do_nonbonded(cr, fr, x, f, f_longrange, md, excl,
 297                              &(enerd->foreign_grpp), box_size, nrnb,
 298                              lam_i, dvdl_dum, -1, -1,
 299                              (donb_flags & ~GMX_NONBONDED_DO_FORCE) | GMX_NONBONDED_DO_FOREIGNLAMBDA);
 300                 sum_epot(&ir->opts, &(enerd->foreign_grpp), enerd->foreign_term);
 301                 enerd->enerpart_lambda[i] += enerd->foreign_term[F_EPOT];
 302             }
 303         }
 304         wallcycle_sub_stop(wcycle, ewcsNONBONDED);
 305         where();
 306     }
 307
 308     /* If we are doing GB, calculate bonded forces and apply corrections
 309      * to the solvation forces */
 310     /* MRS: Eventually, many need to include free energy contribution here! */
 311     if (ir->implicit_solvent)
 312     {
 313         wallcycle_sub_start(wcycle, ewcsBONDED);
 314         calc_gb_forces(cr, md, born, top, atype, x, f, fr, idef,
 315                        ir->gb_algorithm, ir->sa_algorithm, nrnb, bBornRadii, &pbc, graph, enerd);
 316         wallcycle_sub_stop(wcycle, ewcsBONDED);
 317     }
 318
 319 #ifdef GMX_MPI
 320     if (TAKETIME)
 321     {
 322         t1          = MPI_Wtime();
 323         fr->t_fnbf += t1-t0;
 324     }
 325 #endif
 326
 327     if (fepvals->sc_alpha != 0)
 328     {
 329         enerd->dvdl_nonlin[efptVDW] += dvdl_nb[efptVDW];
 330     }
 331     else
 332     {
 333         enerd->dvdl_lin[efptVDW] += dvdl_nb[efptVDW];
 334     }
 335
 336     if (fepvals->sc_alpha != 0)
 337
 338     /* even though coulomb part is linear, we already added it, beacuse we
 339        need to go through the vdw calculation anyway */
 340     {
 341         enerd->dvdl_nonlin[efptCOUL] += dvdl_nb[efptCOUL];
 342     }
 343     else
 344     {
 345         enerd->dvdl_lin[efptCOUL] += dvdl_nb[efptCOUL];
 346     }
 347
 348     Vsr = 0;
 349     if (bSepDVDL)
 350     {
 351         for (i = 0; i < enerd->grpp.nener; i++)
 352         {
 353             Vsr +=
 354                 (fr->bBHAM ?
 355                  enerd->grpp.ener[egBHAMSR][i] :
 356                  enerd->grpp.ener[egLJSR][i])
 357                 + enerd->grpp.ener[egCOULSR][i] + enerd->grpp.ener[egGB][i];
 358         }
 359         dvdlsum = dvdl_nb[efptVDW] + dvdl_nb[efptCOUL];
 360         PRINT_SEPDVDL("VdW and Coulomb SR particle-p.", Vsr, dvdlsum);
 361     }
 362     debug_gmx();
 363
 364
 365     if (debug)
 366     {
 367         pr_rvecs(debug, 0, "fshift after SR", fr->fshift, SHIFTS);
 368     }
 369
 370     /* Shift the coordinates. Must be done before bonded forces and PPPM,
 371      * but is also necessary for SHAKE and update, therefore it can NOT
 372      * go when no bonded forces have to be evaluated.
 373      */
 374
 375     /* Here sometimes we would not need to shift with NBFonly,
 376      * but we do so anyhow for consistency of the returned coordinates.
 377      */
 378     if (graph)
 379     {
 380         shift_self(graph, box, x);
 381         if (TRICLINIC(box))
 382         {
 383             inc_nrnb(nrnb, eNR_SHIFTX, 2*graph->nnodes);
 384         }
 385         else
 386         {
 387             inc_nrnb(nrnb, eNR_SHIFTX, graph->nnodes);
 388         }
 389     }
 390     /* Check whether we need to do bondeds or correct for exclusions */
 391     if (fr->bMolPBC &&
 392         ((flags & GMX_FORCE_BONDED)
 393          || EEL_RF(fr->eeltype) || EEL_FULL(fr->eeltype)))
 394     {
 395         /* Since all atoms are in the rectangular or triclinic unit-cell,
 396          * only single box vector shifts (2 in x) are required.
 397          */
 398         set_pbc_dd(&pbc, fr->ePBC, cr->dd, TRUE, box);
 399     }
 400     debug_gmx();
 401
 402     if (flags & GMX_FORCE_BONDED)
 403     {
 404         wallcycle_sub_start(wcycle, ewcsBONDED);
 405         calc_bonds(fplog, cr->ms,
 406                    idef, x, hist, f, fr, &pbc, graph, enerd, nrnb, lambda, md, fcd,
 407                    DOMAINDECOMP(cr) ? cr->dd->gatindex : NULL, atype, born,
 408                    flags,
 409                    fr->bSepDVDL && do_per_step(step, ir->nstlog), step);
 410
 411         /* Check if we have to determine energy differences
 412          * at foreign lambda's.
 413          */
 414         if (fepvals->n_lambda > 0 && (flags & GMX_FORCE_DHDL) &&
 415             idef->ilsort != ilsortNO_FE)
 416         {
 417             if (idef->ilsort != ilsortFE_SORTED)
 418             {
 419                 gmx_incons("The bonded interactions are not sorted for free energy");
 420             }
 421             for (i = 0; i < enerd->n_lambda; i++)
 422             {
 423                 reset_foreign_enerdata(enerd);
 424                 for (j = 0; j < efptNR; j++)
 425                 {
 426                     lam_i[j] = (i == 0 ? lambda[j] : fepvals->all_lambda[j][i-1]);
 427                 }
 428                 calc_bonds_lambda(fplog, idef, x, fr, &pbc, graph, &(enerd->foreign_grpp), enerd->foreign_term, nrnb, lam_i, md,
 429                                   fcd, DOMAINDECOMP(cr) ? cr->dd->gatindex : NULL);
 430                 sum_epot(&ir->opts, &(enerd->foreign_grpp), enerd->foreign_term);
 431                 enerd->enerpart_lambda[i] += enerd->foreign_term[F_EPOT];
 432             }
 433         }
 434         debug_gmx();
 435
 436         wallcycle_sub_stop(wcycle, ewcsBONDED);
 437     }
 438
 439     where();
 440
 441     *cycles_pme = 0;
 442     if (EEL_FULL(fr->eeltype))
 443     {
 444         bSB = (ir->nwall == 2);
 445         if (bSB)
 446         {
 447             copy_mat(box, boxs);
 448             svmul(ir->wall_ewald_zfac, boxs[ZZ], boxs[ZZ]);
 449             box_size[ZZ] *= ir->wall_ewald_zfac;
 450         }
 451
 452         clear_mat(fr->vir_el_recip);
 453
 454         if (fr->bEwald)
 455         {
 456             Vcorr = 0;
 457             dvdl  = 0;
 458
 459             /* With the Verlet scheme exclusion forces are calculated
 460              * in the non-bonded kernel.
 461              */
 462             /* The TPI molecule does not have exclusions with the rest
 463              * of the system and no intra-molecular PME grid contributions
 464              * will be calculated in gmx_pme_calc_energy.
 465              */
 466             if ((ir->cutoff_scheme == ecutsGROUP && fr->n_tpi == 0) ||
 467                 ir->ewald_geometry != eewg3D ||
 468                 ir->epsilon_surface != 0)
 469             {
 470                 int nthreads, t;
 471
 472                 wallcycle_sub_start(wcycle, ewcsEWALD_CORRECTION);
 473
 474                 if (fr->n_tpi > 0)
 475                 {
 476                     gmx_fatal(FARGS, "TPI with PME currently only works in a 3D geometry with tin-foil boundary conditions");
 477                 }
 478
 479                 nthreads = gmx_omp_nthreads_get(emntBonded);
 480 #pragma omp parallel for num_threads(nthreads) schedule(static)
 481                 for (t = 0; t < nthreads; t++)
 482                 {
 483                     int     s, e, i;
 484                     rvec   *fnv;
 485                     tensor *vir;
 486                     real   *Vcorrt, *dvdlt;
 487                     if (t == 0)
 488                     {
 489                         fnv    = fr->f_novirsum;
 490                         vir    = &fr->vir_el_recip;
 491                         Vcorrt = &Vcorr;
 492                         dvdlt  = &dvdl;
 493                     }
 494                     else
 495                     {
 496                         fnv    = fr->f_t[t].f;
 497                         vir    = &fr->f_t[t].vir;
 498                         Vcorrt = &fr->f_t[t].Vcorr;
 499                         dvdlt  = &fr->f_t[t].dvdl[efptCOUL];
 500                         for (i = 0; i < fr->natoms_force; i++)
 501                         {
 502                             clear_rvec(fnv[i]);
 503                         }
 504                         clear_mat(*vir);
 505                     }
 506                     *dvdlt  = 0;
 507                     *Vcorrt =
 508                         ewald_LRcorrection(fplog,
 509                                            fr->excl_load[t], fr->excl_load[t+1],
 510                                            cr, t, fr,
 511                                            md->chargeA,
 512                                            md->nChargePerturbed ? md->chargeB : NULL,
 513                                            ir->cutoff_scheme != ecutsVERLET,
 514                                            excl, x, bSB ? boxs : box, mu_tot,
 515                                            ir->ewald_geometry,
 516                                            ir->epsilon_surface,
 517                                            fnv, *vir,
 518                                            lambda[efptCOUL], dvdlt);
 519                 }
 520                 if (nthreads > 1)
 521                 {
 522                     reduce_thread_forces(fr->natoms_force, fr->f_novirsum,
 523                                          fr->vir_el_recip,
 524                                          &Vcorr, efptCOUL, &dvdl,
 525                                          nthreads, fr->f_t);
 526                 }
 527
 528                 wallcycle_sub_stop(wcycle, ewcsEWALD_CORRECTION);
 529             }
 530
 531             if (fr->n_tpi == 0)
 532             {
 533                 Vcorr += ewald_charge_correction(cr, fr, lambda[efptCOUL], box,
 534                                                  &dvdl, fr->vir_el_recip);
 535             }
 536
 537             PRINT_SEPDVDL("Ewald excl./charge/dip. corr.", Vcorr, dvdl);
 538             enerd->dvdl_lin[efptCOUL] += dvdl;
 539         }
 540
 541         status = 0;
 542         Vlr    = 0;
 543         dvdl   = 0;
 544         switch (fr->eeltype)
 545         {
 546             case eelPME:
 547             case eelPMESWITCH:
 548             case eelPMEUSER:
 549             case eelPMEUSERSWITCH:
 550             case eelP3M_AD:
 551                 if (cr->duty & DUTY_PME)
 552                 {
 553                     assert(fr->n_tpi >= 0);
 554                     if (fr->n_tpi == 0 || (flags & GMX_FORCE_STATECHANGED))
 555                     {
 556                         pme_flags = GMX_PME_SPREAD_Q | GMX_PME_SOLVE;
 557                         if (flags & GMX_FORCE_FORCES)
 558                         {
 559                             pme_flags |= GMX_PME_CALC_F;
 560                         }
 561                         if (flags & (GMX_FORCE_VIRIAL | GMX_FORCE_ENERGY))
 562                         {
 563                             pme_flags |= GMX_PME_CALC_ENER_VIR;
 564                         }
 565                         if (fr->n_tpi > 0)
 566                         {
 567                             /* We don't calculate f, but we do want the potential */
 568                             pme_flags |= GMX_PME_CALC_POT;
 569                         }
 570                         wallcycle_start(wcycle, ewcPMEMESH);
 571                         status = gmx_pme_do(fr->pmedata,
 572                                             md->start, md->homenr - fr->n_tpi,
 573                                             x, fr->f_novirsum,
 574                                             md->chargeA, md->chargeB,
 575                                             bSB ? boxs : box, cr,
 576                                             DOMAINDECOMP(cr) ? dd_pme_maxshift_x(cr->dd) : 0,
 577                                             DOMAINDECOMP(cr) ? dd_pme_maxshift_y(cr->dd) : 0,
 578                                             nrnb, wcycle,
 579                                             fr->vir_el_recip, fr->ewaldcoeff,
 580                                             &Vlr, lambda[efptCOUL], &dvdl,
 581                                             pme_flags);
 582                         *cycles_pme = wallcycle_stop(wcycle, ewcPMEMESH);
 583
 584                         /* We should try to do as little computation after
 585                          * this as possible, because parallel PME synchronizes
 586                          * the nodes, so we want all load imbalance of the rest
 587                          * of the force calculation to be before the PME call.
 588                          * DD load balancing is done on the whole time of
 589                          * the force call (without PME).
 590                          */
 591                     }
 592                     if (fr->n_tpi > 0)
 593                     {
 594                         /* Determine the PME grid energy of the test molecule
 595                          * with the PME grid potential of the other charges.
 596                          */
 597                         gmx_pme_calc_energy(fr->pmedata, fr->n_tpi,
 598                                             x + md->homenr - fr->n_tpi,
 599                                             md->chargeA + md->homenr - fr->n_tpi,
 600                                             &Vlr);
 601                     }
 602                     PRINT_SEPDVDL("PME mesh", Vlr, dvdl);
 603                 }
 604                 break;
 605             case eelEWALD:
 606                 Vlr = do_ewald(fplog, FALSE, ir, x, fr->f_novirsum,
 607                                md->chargeA, md->chargeB,
 608                                box_size, cr, md->homenr,
 609                                fr->vir_el_recip, fr->ewaldcoeff,
 610                                lambda[efptCOUL], &dvdl, fr->ewald_table);
 611                 PRINT_SEPDVDL("Ewald long-range", Vlr, dvdl);
 612                 break;
 613             default:
 614                 gmx_fatal(FARGS, "No such electrostatics method implemented %s",
 615                           eel_names[fr->eeltype]);
 616         }
 617         if (status != 0)
 618         {
 619             gmx_fatal(FARGS, "Error %d in long range electrostatics routine %s",
 620                       status, EELTYPE(fr->eeltype));
 621         }
 622         /* Note that with separate PME nodes we get the real energies later */
 623         enerd->dvdl_lin[efptCOUL] += dvdl;
 624         enerd->term[F_COUL_RECIP]  = Vlr + Vcorr;
 625         if (debug)
 626         {
 627             fprintf(debug, "Vlr = %g, Vcorr = %g, Vlr_corr = %g\n",
 628                     Vlr, Vcorr, enerd->term[F_COUL_RECIP]);
 629             pr_rvecs(debug, 0, "vir_el_recip after corr", fr->vir_el_recip, DIM);
 630             pr_rvecs(debug, 0, "fshift after LR Corrections", fr->fshift, SHIFTS);
 631         }
 632     }
 633     else
 634     {
 635         if (EEL_RF(fr->eeltype))
 636         {
 637             /* With the Verlet scheme exclusion forces are calculated
 638              * in the non-bonded kernel.
 639              */
 640             if (ir->cutoff_scheme != ecutsVERLET && fr->eeltype != eelRF_NEC)
 641             {
 642                 dvdl                   = 0;
 643                 enerd->term[F_RF_EXCL] =
 644                     RF_excl_correction(fplog, fr, graph, md, excl, x, f,
 645                                        fr->fshift, &pbc, lambda[efptCOUL], &dvdl);
 646             }
 647
 648             enerd->dvdl_lin[efptCOUL] += dvdl;
 649             PRINT_SEPDVDL("RF exclusion correction",
 650                           enerd->term[F_RF_EXCL], dvdl);
 651         }
 652     }
 653     where();
 654     debug_gmx();
 655
 656     if (debug)
 657     {
 658         print_nrnb(debug, nrnb);
 659     }
 660     debug_gmx();
 661
 662 #ifdef GMX_MPI
 663     if (TAKETIME)
 664     {
 665         t2 = MPI_Wtime();
 666         MPI_Barrier(cr->mpi_comm_mygroup);
 667         t3          = MPI_Wtime();
 668         fr->t_wait += t3-t2;
 669         if (fr->timesteps == 11)
 670         {
 671             fprintf(stderr, "* PP load balancing info: node %d, step %s, rel wait time=%3.0f%% , load string value: %7.2f\n",
 672                     cr->nodeid, gmx_step_str(fr->timesteps, buf),
 673                     100*fr->t_wait/(fr->t_wait+fr->t_fnbf),
 674                     (fr->t_fnbf+fr->t_wait)/fr->t_fnbf);
 675         }
 676         fr->timesteps++;
 677     }
 678 #endif
 679
 680     if (debug)
 681     {
 682         pr_rvecs(debug, 0, "fshift after bondeds", fr->fshift, SHIFTS);
 683     }
 684
 685 }
 686
 687 void init_enerdata(int ngener, int n_lambda, gmx_enerdata_t *enerd)
 688 {
 689     int i, n2;
 690
 691     for (i = 0; i < F_NRE; i++)
 692     {
 693         enerd->term[i]         = 0;
 694         enerd->foreign_term[i] = 0;
 695     }
 696
 697
 698     for (i = 0; i < efptNR; i++)
 699     {
 700         enerd->dvdl_lin[i]     = 0;
 701         enerd->dvdl_nonlin[i]  = 0;
 702     }
 703
 704     n2 = ngener*ngener;
 705     if (debug)
 706     {
 707         fprintf(debug, "Creating %d sized group matrix for energies\n", n2);
 708     }
 709     enerd->grpp.nener         = n2;
 710     enerd->foreign_grpp.nener = n2;
 711     for (i = 0; (i < egNR); i++)
 712     {
 713         snew(enerd->grpp.ener[i], n2);
 714         snew(enerd->foreign_grpp.ener[i], n2);
 715     }
 716
 717     if (n_lambda)
 718     {
 719         enerd->n_lambda = 1 + n_lambda;
 720         snew(enerd->enerpart_lambda, enerd->n_lambda);
 721     }
 722     else
 723     {
 724         enerd->n_lambda = 0;
 725     }
 726 }
 727
 728 void destroy_enerdata(gmx_enerdata_t *enerd)
 729 {
 730     int i;
 731
 732     for (i = 0; (i < egNR); i++)
 733     {
 734         sfree(enerd->grpp.ener[i]);
 735     }
 736
 737     for (i = 0; (i < egNR); i++)
 738     {
 739         sfree(enerd->foreign_grpp.ener[i]);
 740     }
 741
 742     if (enerd->n_lambda)
 743     {
 744         sfree(enerd->enerpart_lambda);
 745     }
 746 }
 747
 748 static real sum_v(int n, real v[])
 749 {
 750     real t;
 751     int  i;
 752
 753     t = 0.0;
 754     for (i = 0; (i < n); i++)
 755     {
 756         t = t + v[i];
 757     }
 758
 759     return t;
 760 }
 761
 762 void sum_epot(t_grpopts *opts, gmx_grppairener_t *grpp, real *epot)
 763 {
 764     int i;
 765
 766     /* Accumulate energies */
 767     epot[F_COUL_SR]  = sum_v(grpp->nener, grpp->ener[egCOULSR]);
 768     epot[F_LJ]       = sum_v(grpp->nener, grpp->ener[egLJSR]);
 769     epot[F_LJ14]     = sum_v(grpp->nener, grpp->ener[egLJ14]);
 770     epot[F_COUL14]   = sum_v(grpp->nener, grpp->ener[egCOUL14]);
 771     epot[F_COUL_LR]  = sum_v(grpp->nener, grpp->ener[egCOULLR]);
 772     epot[F_LJ_LR]    = sum_v(grpp->nener, grpp->ener[egLJLR]);
 773     /* We have already added 1-2,1-3, and 1-4 terms to F_GBPOL */
 774     epot[F_GBPOL]   += sum_v(grpp->nener, grpp->ener[egGB]);
 775
 776 /* lattice part of LR doesnt belong to any group
 777  * and has been added earlier
 778  */
 779     epot[F_BHAM]     = sum_v(grpp->nener, grpp->ener[egBHAMSR]);
 780     epot[F_BHAM_LR]  = sum_v(grpp->nener, grpp->ener[egBHAMLR]);
 781
 782     epot[F_EPOT] = 0;
 783     for (i = 0; (i < F_EPOT); i++)
 784     {
 785         if (i != F_DISRESVIOL && i != F_ORIRESDEV)
 786         {
 787             epot[F_EPOT] += epot[i];
 788         }
 789     }
 790 }
 791
 792 void sum_dhdl(gmx_enerdata_t *enerd, real *lambda, t_lambda *fepvals)
 793 {
 794     int    i, j, index;
 795     double dlam;
 796
 797     enerd->dvdl_lin[efptVDW] += enerd->term[F_DVDL_VDW];  /* include dispersion correction */
 798     enerd->term[F_DVDL]       = 0.0;
 799     for (i = 0; i < efptNR; i++)
 800     {
 801         if (fepvals->separate_dvdl[i])
 802         {
 803             /* could this be done more readably/compactly? */
 804             switch (i)
 805             {
 806                 case (efptCOUL):
 807                     index = F_DVDL_COUL;
 808                     break;
 809                 case (efptVDW):
 810                     index = F_DVDL_VDW;
 811                     break;
 812                 case (efptBONDED):
 813                     index = F_DVDL_BONDED;
 814                     break;
 815                 case (efptRESTRAINT):
 816                     index = F_DVDL_RESTRAINT;
 817                     break;
 818                 case (efptMASS):
 819                     index = F_DKDL;
 820                     break;
 821                 default:
 822                     index = F_DVDL;
 823                     break;
 824             }
 825             enerd->term[index] = enerd->dvdl_lin[i] + enerd->dvdl_nonlin[i];
 826             if (debug)
 827             {
 828                 fprintf(debug, "dvdl-%s[%2d]: %f: non-linear %f + linear %f\n",
 829                         efpt_names[i], i, enerd->term[index], enerd->dvdl_nonlin[i], enerd->dvdl_lin[i]);
 830             }
 831         }
 832         else
 833         {
 834             enerd->term[F_DVDL] += enerd->dvdl_lin[i] + enerd->dvdl_nonlin[i];
 835             if (debug)
 836             {
 837                 fprintf(debug, "dvd-%sl[%2d]: %f: non-linear %f + linear %f\n",
 838                         efpt_names[0], i, enerd->term[F_DVDL], enerd->dvdl_nonlin[i], enerd->dvdl_lin[i]);
 839             }
 840         }
 841     }
 842
 843     /* Notes on the foreign lambda free energy difference evaluation:
 844      * Adding the potential and ekin terms that depend linearly on lambda
 845      * as delta lam * dvdl to the energy differences is exact.
 846      * For the constraints this is not exact, but we have no other option
 847      * without literally changing the lengths and reevaluating the energies at each step.
 848      * (try to remedy this post 4.6 - MRS)
 849      * For the non-bonded LR term we assume that the soft-core (if present)
 850      * no longer affects the energy beyond the short-range cut-off,
 851      * which is a very good approximation (except for exotic settings).
 852      * (investigate how to overcome this post 4.6 - MRS)
 853      */
 854
 855     for (i = 0; i < fepvals->n_lambda; i++)
 856     {                                         /* note we are iterating over fepvals here!
 857                                                  For the current lam, dlam = 0 automatically,
 858                                                  so we don't need to add anything to the
 859                                                  enerd->enerpart_lambda[0] */
 860
 861         /* we don't need to worry about dvdl contributions to the current lambda, because
 862            it's automatically zero */
 863
 864         /* first kinetic energy term */
 865         dlam = (fepvals->all_lambda[efptMASS][i] - lambda[efptMASS]);
 866
 867         enerd->enerpart_lambda[i+1] += enerd->term[F_DKDL]*dlam;
 868
 869         for (j = 0; j < efptNR; j++)
 870         {
 871             if (j == efptMASS)
 872             {
 873                 continue;
 874             }                            /* no other mass term to worry about */
 875
 876             dlam = (fepvals->all_lambda[j][i]-lambda[j]);
 877             enerd->enerpart_lambda[i+1] += dlam*enerd->dvdl_lin[j];
 878             if (debug)
 879             {
 880                 fprintf(debug, "enerdiff lam %g: (%15s), non-linear %f linear %f*%f\n",
 881                         fepvals->all_lambda[j][i], efpt_names[j],
 882                         (enerd->enerpart_lambda[i+1] - enerd->enerpart_lambda[0]),
 883                         dlam, enerd->dvdl_lin[j]);
 884             }
 885         }
 886     }
 887 }
 888
 889
 890 void reset_foreign_enerdata(gmx_enerdata_t *enerd)
 891 {
 892     int  i, j;
 893
 894     /* First reset all foreign energy components.  Foreign energies always called on
 895        neighbor search steps */
 896     for (i = 0; (i < egNR); i++)
 897     {
 898         for (j = 0; (j < enerd->grpp.nener); j++)
 899         {
 900             enerd->foreign_grpp.ener[i][j] = 0.0;
 901         }
 902     }
 903
 904     /* potential energy components */
 905     for (i = 0; (i <= F_EPOT); i++)
 906     {
 907         enerd->foreign_term[i] = 0.0;
 908     }
 909 }
 910
 911 void reset_enerdata(t_grpopts *opts,
 912                     t_forcerec *fr, gmx_bool bNS,
 913                     gmx_enerdata_t *enerd,
 914                     gmx_bool bMaster)
 915 {
 916     gmx_bool bKeepLR;
 917     int      i, j;
 918
 919     /* First reset all energy components, except for the long range terms
 920      * on the master at non neighbor search steps, since the long range
 921      * terms have already been summed at the last neighbor search step.
 922      */
 923     bKeepLR = (fr->bTwinRange && !bNS);
 924     for (i = 0; (i < egNR); i++)
 925     {
 926         if (!(bKeepLR && bMaster && (i == egCOULLR || i == egLJLR)))
 927         {
 928             for (j = 0; (j < enerd->grpp.nener); j++)
 929             {
 930                 enerd->grpp.ener[i][j] = 0.0;
 931             }
 932         }
 933     }
 934     for (i = 0; i < efptNR; i++)
 935     {
 936         enerd->dvdl_lin[i]    = 0.0;
 937         enerd->dvdl_nonlin[i] = 0.0;
 938     }
 939
 940     /* Normal potential energy components */
 941     for (i = 0; (i <= F_EPOT); i++)
 942     {
 943         enerd->term[i] = 0.0;
 944     }
 945     /* Initialize the dVdlambda term with the long range contribution */
 946     /* Initialize the dvdl term with the long range contribution */
 947     enerd->term[F_DVDL]            = 0.0;
 948     enerd->term[F_DVDL_COUL]       = 0.0;
 949     enerd->term[F_DVDL_VDW]        = 0.0;
 950     enerd->term[F_DVDL_BONDED]     = 0.0;
 951     enerd->term[F_DVDL_RESTRAINT]  = 0.0;
 952     enerd->term[F_DKDL]            = 0.0;
 953     if (enerd->n_lambda > 0)
 954     {
 955         for (i = 0; i < enerd->n_lambda; i++)
 956         {
 957             enerd->enerpart_lambda[i] = 0.0;
 958         }
 959     }
 960     /* reset foreign energy data - separate function since we also call it elsewhere */
 961     reset_foreign_enerdata(enerd);
 962 }