src/gromacs/mdlib/ns.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 #ifdef GMX_THREAD_SHM_FDECOMP
  41 #include <pthread.h>
  42 #endif
  43
  44 #include <math.h>
  45 #include <string.h>
  46 #include "sysstuff.h"
  47 #include "smalloc.h"
  48 #include "macros.h"
  49 #include "maths.h"
  50 #include "vec.h"
  51 #include "network.h"
  52 #include "nsgrid.h"
  53 #include "force.h"
  54 #include "nonbonded.h"
  55 #include "ns.h"
  56 #include "pbc.h"
  57 #include "names.h"
  58 #include "gmx_fatal.h"
  59 #include "nrnb.h"
  60 #include "txtdump.h"
  61 #include "mtop_util.h"
  62
  63 #include "domdec.h"
  64
  65
  66 /*
  67  *    E X C L U S I O N   H A N D L I N G
  68  */
  69
  70 #ifdef DEBUG
  71 static void SETEXCL_(t_excl e[],atom_id i,atom_id j)
  72 {   e[j] = e[j] | (1<<i); }
  73 static void RMEXCL_(t_excl e[],atom_id i,atom_id j)
  74 { e[j]=e[j] & ~(1<<i); }
  75 static gmx_bool ISEXCL_(t_excl e[],atom_id i,atom_id j)
  76 { return (gmx_bool)(e[j] & (1<<i)); }
  77 static gmx_bool NOTEXCL_(t_excl e[],atom_id i,atom_id j)
  78 {  return !(ISEXCL(e,i,j)); }
  79 #else
  80 #define SETEXCL(e,i,j) (e)[((atom_id) (j))] |= (1<<((atom_id) (i)))
  81 #define RMEXCL(e,i,j)  (e)[((atom_id) (j))] &= (~(1<<((atom_id) (i))))
  82 #define ISEXCL(e,i,j)  (gmx_bool) ((e)[((atom_id) (j))] & (1<<((atom_id) (i))))
  83 #define NOTEXCL(e,i,j) !(ISEXCL(e,i,j))
  84 #endif
  85
  86 /************************************************
  87  *
  88  *  U T I L I T I E S    F O R    N S
  89  *
  90  ************************************************/
  91
  92 static void reallocate_nblist(t_nblist *nl)
  93 {
  94     if (gmx_debug_at)
  95     {
  96         fprintf(debug,"reallocating neigborlist il_code=%d, maxnri=%d\n",
  97                 nl->il_code,nl->maxnri);
  98     }
  99     srenew(nl->iinr,   nl->maxnri);
 100     if (nl->enlist == enlistCG_CG)
 101     {
 102         srenew(nl->iinr_end,nl->maxnri);
 103     }
 104     srenew(nl->gid,    nl->maxnri);
 105     srenew(nl->shift,  nl->maxnri);
 106     srenew(nl->jindex, nl->maxnri+1);
 107 }
 108
 109 /* ivdw/icoul are used to determine the type of interaction, so we
 110  * can set an innerloop index here. The obvious choice for this would have
 111  * been the vdwtype/coultype values in the forcerecord, but unfortunately
 112  * those types are braindead - for instance both Buckingham and normal
 113  * Lennard-Jones use the same value (evdwCUT), and a separate gmx_boolean variable
 114  * to determine which interaction is used. There is further no special value
 115  * for 'no interaction'. For backward compatibility with old TPR files we won't
 116  * change this in the 3.x series, so when calling this routine you should use:
 117  *
 118  * icoul=0 no coulomb interaction
 119  * icoul=1 cutoff standard coulomb
 120  * icoul=2 reaction-field coulomb
 121  * icoul=3 tabulated coulomb
 122  *
 123  * ivdw=0 no vdw interaction
 124  * ivdw=1 standard L-J interaction
 125  * ivdw=2 Buckingham
 126  * ivdw=3 tabulated vdw.
 127  *
 128  * Kind of ugly, but it works.
 129  */
 130 static void init_nblist(t_nblist *nl_sr,t_nblist *nl_lr,
 131                         int maxsr,int maxlr,
 132                         int ivdw, int icoul,
 133                         gmx_bool bfree, int enlist)
 134 {
 135     t_nblist *nl;
 136     int      homenr;
 137     int      i,nn;
 138
 139     int inloop[20] =
 140     {
 141         eNR_NBKERNEL_NONE,
 142         eNR_NBKERNEL010,
 143         eNR_NBKERNEL020,
 144         eNR_NBKERNEL030,
 145         eNR_NBKERNEL100,
 146         eNR_NBKERNEL110,
 147         eNR_NBKERNEL120,
 148         eNR_NBKERNEL130,
 149         eNR_NBKERNEL200,
 150         eNR_NBKERNEL210,
 151         eNR_NBKERNEL220,
 152         eNR_NBKERNEL230,
 153         eNR_NBKERNEL300,
 154         eNR_NBKERNEL310,
 155         eNR_NBKERNEL320,
 156         eNR_NBKERNEL330,
 157         eNR_NBKERNEL400,
 158         eNR_NBKERNEL410,
 159         eNR_NBKERNEL_NONE,
 160         eNR_NBKERNEL430
 161     };
 162
 163     for(i=0; (i<2); i++)
 164     {
 165         nl     = (i == 0) ? nl_sr : nl_lr;
 166         homenr = (i == 0) ? maxsr : maxlr;
 167
 168         if (nl == NULL)
 169         {
 170             continue;
 171         }
 172
 173         /* Set coul/vdw in neighborlist, and for the normal loops we determine
 174          * an index of which one to call.
 175          */
 176         nl->ivdw  = ivdw;
 177         nl->icoul = icoul;
 178         nl->free_energy = bfree;
 179
 180         if (bfree)
 181         {
 182             nl->enlist  = enlistATOM_ATOM;
 183             nl->il_code = eNR_NBKERNEL_FREE_ENERGY;
 184         }
 185         else
 186         {
 187             nl->enlist = enlist;
 188
 189             nn = inloop[4*icoul + ivdw];
 190
 191             /* solvent loops follow directly after the corresponding
 192             * ordinary loops, in the order:
 193             *
 194             * SPC, SPC-SPC, TIP4p, TIP4p-TIP4p
 195             *
 196             */
 197             switch (enlist) {
 198             case enlistATOM_ATOM:
 199             case enlistCG_CG:
 200                 break;
 201             case enlistSPC_ATOM:     nn += 1; break;
 202             case enlistSPC_SPC:      nn += 2; break;
 203             case enlistTIP4P_ATOM:   nn += 3; break;
 204             case enlistTIP4P_TIP4P:  nn += 4; break;
 205             }
 206
 207             nl->il_code = nn;
 208         }
 209
 210         if (debug)
 211             fprintf(debug,"Initiating neighbourlist type %d for %s interactions,\nwith %d SR, %d LR atoms.\n",
 212                     nl->il_code,ENLISTTYPE(enlist),maxsr,maxlr);
 213
 214         /* maxnri is influenced by the number of shifts (maximum is 8)
 215          * and the number of energy groups.
 216          * If it is not enough, nl memory will be reallocated during the run.
 217          * 4 seems to be a reasonable factor, which only causes reallocation
 218          * during runs with tiny and many energygroups.
 219          */
 220         nl->maxnri      = homenr*4;
 221         nl->maxnrj      = 0;
 222         nl->maxlen      = 0;
 223         nl->nri         = -1;
 224         nl->nrj         = 0;
 225         nl->iinr        = NULL;
 226         nl->gid         = NULL;
 227         nl->shift       = NULL;
 228         nl->jindex      = NULL;
 229         reallocate_nblist(nl);
 230         nl->jindex[0] = 0;
 231 #ifdef GMX_THREAD_SHM_FDECOMP
 232         nl->counter = 0;
 233         snew(nl->mtx,1);
 234         pthread_mutex_init(nl->mtx,NULL);
 235 #endif
 236     }
 237 }
 238
 239 void init_neighbor_list(FILE *log,t_forcerec *fr,int homenr)
 240 {
 241    /* Make maxlr tunable! (does not seem to be a big difference though)
 242     * This parameter determines the number of i particles in a long range
 243     * neighbourlist. Too few means many function calls, too many means
 244     * cache trashing.
 245     */
 246    int maxsr,maxsr_wat,maxlr,maxlr_wat;
 247    int icoul,icoulf,ivdw;
 248    int solvent;
 249    int enlist_def,enlist_w,enlist_ww;
 250    int i;
 251    t_nblists *nbl;
 252
 253    /* maxsr     = homenr-fr->nWatMol*3; */
 254    maxsr     = homenr;
 255
 256    if (maxsr < 0)
 257    {
 258      gmx_fatal(FARGS,"%s, %d: Negative number of short range atoms.\n"
 259                  "Call your Gromacs dealer for assistance.",__FILE__,__LINE__);
 260    }
 261    /* This is just for initial allocation, so we do not reallocate
 262     * all the nlist arrays many times in a row.
 263     * The numbers seem very accurate, but they are uncritical.
 264     */
 265    maxsr_wat = min(fr->nWatMol,(homenr+2)/3);
 266    if (fr->bTwinRange)
 267    {
 268        maxlr     = 50;
 269        maxlr_wat = min(maxsr_wat,maxlr);
 270    }
 271    else
 272    {
 273      maxlr = maxlr_wat = 0;
 274    }
 275
 276    /* Determine the values for icoul/ivdw. */
 277    /* Start with GB */
 278    if(fr->bGB)
 279    {
 280        icoul=4;
 281    }
 282    else if (fr->bcoultab)
 283    {
 284        icoul = 3;
 285    }
 286    else if (EEL_RF(fr->eeltype))
 287    {
 288        icoul = 2;
 289    }
 290    else
 291    {
 292        icoul = 1;
 293    }
 294
 295    if (fr->bvdwtab)
 296    {
 297        ivdw = 3;
 298    }
 299    else if (fr->bBHAM)
 300    {
 301        ivdw = 2;
 302    }
 303    else
 304    {
 305        ivdw = 1;
 306    }
 307
 308    fr->ns.bCGlist = (getenv("GMX_NBLISTCG") != 0);
 309    if (!fr->ns.bCGlist)
 310    {
 311        enlist_def = enlistATOM_ATOM;
 312    }
 313    else
 314    {
 315        enlist_def = enlistCG_CG;
 316        if (log != NULL)
 317        {
 318            fprintf(log,"\nUsing charge-group - charge-group neighbor lists and kernels\n\n");
 319        }
 320    }
 321
 322    if (fr->solvent_opt == esolTIP4P) {
 323        enlist_w  = enlistTIP4P_ATOM;
 324        enlist_ww = enlistTIP4P_TIP4P;
 325    } else {
 326        enlist_w  = enlistSPC_ATOM;
 327        enlist_ww = enlistSPC_SPC;
 328    }
 329
 330    for(i=0; i<fr->nnblists; i++)
 331    {
 332        nbl = &(fr->nblists[i]);
 333        init_nblist(&nbl->nlist_sr[eNL_VDWQQ],&nbl->nlist_lr[eNL_VDWQQ],
 334                    maxsr,maxlr,ivdw,icoul,FALSE,enlist_def);
 335        init_nblist(&nbl->nlist_sr[eNL_VDW],&nbl->nlist_lr[eNL_VDW],
 336                    maxsr,maxlr,ivdw,0,FALSE,enlist_def);
 337        init_nblist(&nbl->nlist_sr[eNL_QQ],&nbl->nlist_lr[eNL_QQ],
 338                    maxsr,maxlr,0,icoul,FALSE,enlist_def);
 339        init_nblist(&nbl->nlist_sr[eNL_VDWQQ_WATER],&nbl->nlist_lr[eNL_VDWQQ_WATER],
 340                    maxsr_wat,maxlr_wat,ivdw,icoul, FALSE,enlist_w);
 341        init_nblist(&nbl->nlist_sr[eNL_QQ_WATER],&nbl->nlist_lr[eNL_QQ_WATER],
 342                    maxsr_wat,maxlr_wat,0,icoul, FALSE,enlist_w);
 343        init_nblist(&nbl->nlist_sr[eNL_VDWQQ_WATERWATER],&nbl->nlist_lr[eNL_VDWQQ_WATERWATER],
 344                    maxsr_wat,maxlr_wat,ivdw,icoul, FALSE,enlist_ww);
 345        init_nblist(&nbl->nlist_sr[eNL_QQ_WATERWATER],&nbl->nlist_lr[eNL_QQ_WATERWATER],
 346                    maxsr_wat,maxlr_wat,0,icoul, FALSE,enlist_ww);
 347
 348        if (fr->efep != efepNO)
 349        {
 350            if (fr->bEwald)
 351            {
 352                icoulf = 5;
 353            }
 354            else
 355            {
 356                icoulf = icoul;
 357            }
 358
 359            init_nblist(&nbl->nlist_sr[eNL_VDWQQ_FREE],&nbl->nlist_lr[eNL_VDWQQ_FREE],
 360                        maxsr,maxlr,ivdw,icoulf,TRUE,enlistATOM_ATOM);
 361            init_nblist(&nbl->nlist_sr[eNL_VDW_FREE],&nbl->nlist_lr[eNL_VDW_FREE],
 362                        maxsr,maxlr,ivdw,0,TRUE,enlistATOM_ATOM);
 363            init_nblist(&nbl->nlist_sr[eNL_QQ_FREE],&nbl->nlist_lr[eNL_QQ_FREE],
 364                        maxsr,maxlr,0,icoulf,TRUE,enlistATOM_ATOM);
 365        }
 366    }
 367    /* QMMM MM list */
 368    if (fr->bQMMM && fr->qr->QMMMscheme != eQMMMschemeoniom)
 369    {
 370        init_nblist(&fr->QMMMlist,NULL,
 371                    maxsr,maxlr,0,icoul,FALSE,enlistATOM_ATOM);
 372    }
 373
 374    fr->ns.nblist_initialized=TRUE;
 375 }
 376
 377 static void reset_nblist(t_nblist *nl)
 378 {
 379      nl->nri       = -1;
 380      nl->nrj       = 0;
 381      nl->maxlen    = 0;
 382      if (nl->jindex)
 383      {
 384          nl->jindex[0] = 0;
 385      }
 386 }
 387
 388 static void reset_neighbor_list(t_forcerec *fr,gmx_bool bLR,int nls,int eNL)
 389 {
 390     int n,i;
 391
 392     if (bLR)
 393     {
 394         reset_nblist(&(fr->nblists[nls].nlist_lr[eNL]));
 395     }
 396     else
 397     {
 398         for(n=0; n<fr->nnblists; n++)
 399         {
 400             for(i=0; i<eNL_NR; i++)
 401             {
 402                 reset_nblist(&(fr->nblists[n].nlist_sr[i]));
 403             }
 404         }
 405         if (fr->bQMMM)
 406         {
 407             /* only reset the short-range nblist */
 408             reset_nblist(&(fr->QMMMlist));
 409         }
 410     }
 411 }
 412
 413
 414
 415
 416 static inline void new_i_nblist(t_nblist *nlist,
 417                                 gmx_bool bLR,atom_id i_atom,int shift,int gid)
 418 {
 419     int    i,k,nri,nshift;
 420
 421     nri = nlist->nri;
 422
 423     /* Check whether we have to increase the i counter */
 424     if ((nri == -1) ||
 425         (nlist->iinr[nri]  != i_atom) ||
 426         (nlist->shift[nri] != shift) ||
 427         (nlist->gid[nri]   != gid))
 428     {
 429         /* This is something else. Now see if any entries have
 430          * been added in the list of the previous atom.
 431          */
 432         if ((nri == -1) ||
 433             ((nlist->jindex[nri+1] > nlist->jindex[nri]) &&
 434              (nlist->gid[nri] != -1)))
 435         {
 436             /* If so increase the counter */
 437             nlist->nri++;
 438             nri++;
 439             if (nlist->nri >= nlist->maxnri)
 440             {
 441                 nlist->maxnri += over_alloc_large(nlist->nri);
 442                 reallocate_nblist(nlist);
 443             }
 444         }
 445         /* Set the number of neighbours and the atom number */
 446         nlist->jindex[nri+1] = nlist->jindex[nri];
 447         nlist->iinr[nri]     = i_atom;
 448         nlist->gid[nri]      = gid;
 449         nlist->shift[nri]    = shift;
 450     }
 451 }
 452
 453 static inline void close_i_nblist(t_nblist *nlist)
 454 {
 455     int nri = nlist->nri;
 456     int len;
 457
 458     if (nri >= 0)
 459     {
 460         nlist->jindex[nri+1] = nlist->nrj;
 461
 462         len=nlist->nrj -  nlist->jindex[nri];
 463
 464         /* nlist length for water i molecules is treated statically
 465          * in the innerloops
 466          */
 467         if (len > nlist->maxlen)
 468         {
 469             nlist->maxlen = len;
 470         }
 471     }
 472 }
 473
 474 static inline void close_nblist(t_nblist *nlist)
 475 {
 476     /* Only close this nblist when it has been initialized.
 477      * Avoid the creation of i-lists with no j-particles.
 478      */
 479     if (nlist->nrj == 0)
 480     {
 481         /* Some assembly kernels do not support empty lists,
 482          * make sure here that we don't generate any empty lists.
 483          * With the current ns code this branch is taken in two cases:
 484          * No i-particles at all: nri=-1 here
 485          * There are i-particles, but no j-particles; nri=0 here
 486          */
 487         nlist->nri = 0;
 488     }
 489     else
 490     {
 491         /* Close list number nri by incrementing the count */
 492         nlist->nri++;
 493     }
 494 }
 495
 496 static inline void close_neighbor_list(t_forcerec *fr,gmx_bool bLR,int nls,int eNL,
 497                                        gmx_bool bMakeQMMMnblist)
 498 {
 499     int n,i;
 500
 501     if (bMakeQMMMnblist) {
 502         if (!bLR)
 503         {
 504             close_nblist(&(fr->QMMMlist));
 505         }
 506     }
 507     else
 508     {
 509         if (bLR)
 510         {
 511             close_nblist(&(fr->nblists[nls].nlist_lr[eNL]));
 512         }
 513         else
 514         {
 515             for(n=0; n<fr->nnblists; n++)
 516             {
 517                 for(i=0; (i<eNL_NR); i++)
 518                 {
 519                     close_nblist(&(fr->nblists[n].nlist_sr[i]));
 520                 }
 521             }
 522         }
 523     }
 524 }
 525
 526 static inline void add_j_to_nblist(t_nblist *nlist,atom_id j_atom,gmx_bool bLR)
 527 {
 528     int nrj=nlist->nrj;
 529
 530     if (nlist->nrj >= nlist->maxnrj)
 531     {
 532         nlist->maxnrj = over_alloc_small(nlist->nrj + 1);
 533         if (gmx_debug_at)
 534             fprintf(debug,"Increasing %s nblist %s j size to %d\n",
 535                     bLR ? "LR" : "SR",nrnb_str(nlist->il_code),nlist->maxnrj);
 536
 537         srenew(nlist->jjnr,nlist->maxnrj);
 538     }
 539
 540     nlist->jjnr[nrj] = j_atom;
 541     nlist->nrj ++;
 542 }
 543
 544 static inline void add_j_to_nblist_cg(t_nblist *nlist,
 545                                       atom_id j_start,int j_end,
 546                                       t_excl *bexcl,gmx_bool i_is_j,
 547                                       gmx_bool bLR)
 548 {
 549     int nrj=nlist->nrj;
 550     int j;
 551
 552     if (nlist->nrj >= nlist->maxnrj)
 553     {
 554         nlist->maxnrj = over_alloc_small(nlist->nrj + 1);
 555         if (gmx_debug_at)
 556             fprintf(debug,"Increasing %s nblist %s j size to %d\n",
 557                     bLR ? "LR" : "SR",nrnb_str(nlist->il_code),nlist->maxnrj);
 558
 559         srenew(nlist->jjnr    ,nlist->maxnrj);
 560         srenew(nlist->jjnr_end,nlist->maxnrj);
 561         srenew(nlist->excl    ,nlist->maxnrj*MAX_CGCGSIZE);
 562     }
 563
 564     nlist->jjnr[nrj]     = j_start;
 565     nlist->jjnr_end[nrj] = j_end;
 566
 567     if (j_end - j_start > MAX_CGCGSIZE)
 568     {
 569         gmx_fatal(FARGS,"The charge-group - charge-group neighborlist do not support charge groups larger than %d, found a charge group of size %d",MAX_CGCGSIZE,j_end-j_start);
 570     }
 571
 572     /* Set the exclusions */
 573     for(j=j_start; j<j_end; j++)
 574     {
 575         nlist->excl[nrj*MAX_CGCGSIZE + j - j_start] = bexcl[j];
 576     }
 577     if (i_is_j)
 578     {
 579         /* Avoid double counting of intra-cg interactions */
 580         for(j=1; j<j_end-j_start; j++)
 581         {
 582             nlist->excl[nrj*MAX_CGCGSIZE + j] |= (1<<j) - 1;
 583         }
 584     }
 585
 586     nlist->nrj ++;
 587 }
 588
 589 typedef void
 590 put_in_list_t(gmx_bool              bHaveVdW[],
 591               int               ngid,
 592               t_mdatoms *       md,
 593               int               icg,
 594               int               jgid,
 595               int               nj,
 596               atom_id           jjcg[],
 597               atom_id           index[],
 598               t_excl            bExcl[],
 599               int               shift,
 600               t_forcerec *      fr,
 601               gmx_bool              bLR,
 602               gmx_bool              bDoVdW,
 603               gmx_bool              bDoCoul);
 604
 605 static void
 606 put_in_list_at(gmx_bool              bHaveVdW[],
 607                int               ngid,
 608                t_mdatoms *       md,
 609                int               icg,
 610                int               jgid,
 611                int               nj,
 612                atom_id           jjcg[],
 613                atom_id           index[],
 614                t_excl            bExcl[],
 615                int               shift,
 616                t_forcerec *      fr,
 617                gmx_bool              bLR,
 618                gmx_bool              bDoVdW,
 619                gmx_bool              bDoCoul)
 620 {
 621     /* The a[] index has been removed,
 622      * to put it back in i_atom should be a[i0] and jj should be a[jj].
 623      */
 624     t_nblist *   vdwc;
 625     t_nblist *   vdw;
 626     t_nblist *   coul;
 627     t_nblist *   vdwc_free  = NULL;
 628     t_nblist *   vdw_free   = NULL;
 629     t_nblist *   coul_free  = NULL;
 630     t_nblist *   vdwc_ww    = NULL;
 631     t_nblist *   coul_ww    = NULL;
 632
 633     int             i,j,jcg,igid,gid,nbl_ind,ind_ij;
 634     atom_id   jj,jj0,jj1,i_atom;
 635     int       i0,nicg,len;
 636
 637     int       *cginfo;
 638     int       *type,*typeB;
 639     real      *charge,*chargeB;
 640     real      qi,qiB,qq,rlj;
 641     gmx_bool      bFreeEnergy,bFree,bFreeJ,bNotEx,*bPert;
 642     gmx_bool      bDoVdW_i,bDoCoul_i,bDoCoul_i_sol;
 643     int       iwater,jwater;
 644     t_nblist  *nlist;
 645
 646     /* Copy some pointers */
 647     cginfo  = fr->cginfo;
 648     charge  = md->chargeA;
 649     chargeB = md->chargeB;
 650     type    = md->typeA;
 651     typeB   = md->typeB;
 652     bPert   = md->bPerturbed;
 653
 654     /* Get atom range */
 655     i0     = index[icg];
 656     nicg   = index[icg+1]-i0;
 657
 658     /* Get the i charge group info */
 659     igid   = GET_CGINFO_GID(cginfo[icg]);
 660     iwater = GET_CGINFO_SOLOPT(cginfo[icg]);
 661
 662     bFreeEnergy = FALSE;
 663     if (md->nPerturbed)
 664     {
 665         /* Check if any of the particles involved are perturbed.
 666          * If not we can do the cheaper normal put_in_list
 667          * and use more solvent optimization.
 668          */
 669         for(i=0; i<nicg; i++)
 670         {
 671             bFreeEnergy |= bPert[i0+i];
 672         }
 673         /* Loop over the j charge groups */
 674         for(j=0; (j<nj && !bFreeEnergy); j++)
 675         {
 676             jcg = jjcg[j];
 677             jj0 = index[jcg];
 678             jj1 = index[jcg+1];
 679             /* Finally loop over the atoms in the j-charge group */
 680             for(jj=jj0; jj<jj1; jj++)
 681             {
 682                 bFreeEnergy |= bPert[jj];
 683             }
 684         }
 685     }
 686
 687     /* Unpack pointers to neighbourlist structs */
 688     if (fr->nnblists == 1)
 689     {
 690         nbl_ind = 0;
 691     }
 692     else
 693     {
 694         nbl_ind = fr->gid2nblists[GID(igid,jgid,ngid)];
 695     }
 696     if (bLR)
 697     {
 698         nlist = fr->nblists[nbl_ind].nlist_lr;
 699     }
 700     else
 701     {
 702         nlist = fr->nblists[nbl_ind].nlist_sr;
 703     }
 704
 705     if (iwater != esolNO)
 706     {
 707         vdwc = &nlist[eNL_VDWQQ_WATER];
 708         vdw  = &nlist[eNL_VDW];
 709         coul = &nlist[eNL_QQ_WATER];
 710 #ifndef DISABLE_WATERWATER_NLIST
 711         vdwc_ww = &nlist[eNL_VDWQQ_WATERWATER];
 712         coul_ww = &nlist[eNL_QQ_WATERWATER];
 713 #endif
 714     }
 715     else
 716     {
 717         vdwc = &nlist[eNL_VDWQQ];
 718         vdw  = &nlist[eNL_VDW];
 719         coul = &nlist[eNL_QQ];
 720     }
 721
 722     if (!bFreeEnergy)
 723     {
 724         if (iwater != esolNO)
 725         {
 726             /* Loop over the atoms in the i charge group */
 727             i_atom  = i0;
 728             gid     = GID(igid,jgid,ngid);
 729             /* Create new i_atom for each energy group */
 730             if (bDoCoul && bDoVdW)
 731             {
 732                 new_i_nblist(vdwc,bLR,i_atom,shift,gid);
 733 #ifndef DISABLE_WATERWATER_NLIST
 734                 new_i_nblist(vdwc_ww,bLR,i_atom,shift,gid);
 735 #endif
 736             }
 737             if (bDoVdW)
 738             {
 739                 new_i_nblist(vdw,bLR,i_atom,shift,gid);
 740             }
 741             if (bDoCoul)
 742             {
 743                 new_i_nblist(coul,bLR,i_atom,shift,gid);
 744 #ifndef DISABLE_WATERWATER_NLIST
 745                 new_i_nblist(coul_ww,bLR,i_atom,shift,gid);
 746 #endif
 747             }
 748           /* Loop over the j charge groups */
 749             for(j=0; (j<nj); j++)
 750             {
 751                 jcg=jjcg[j];
 752
 753                 if (jcg == icg)
 754                 {
 755                     continue;
 756                 }
 757
 758                 jj0 = index[jcg];
 759                 jwater = GET_CGINFO_SOLOPT(cginfo[jcg]);
 760
 761                 if (iwater == esolSPC && jwater == esolSPC)
 762                 {
 763                     /* Interaction between two SPC molecules */
 764                     if (!bDoCoul)
 765                     {
 766                         /* VdW only - only first atoms in each water interact */
 767                         add_j_to_nblist(vdw,jj0,bLR);
 768                     }
 769                     else
 770                     {
 771 #ifdef DISABLE_WATERWATER_NLIST
 772                         /* Add entries for the three atoms - only do VdW if we need to */
 773                         if (!bDoVdW)
 774                         {
 775                             add_j_to_nblist(coul,jj0,bLR);
 776                         }
 777                         else
 778                         {
 779                             add_j_to_nblist(vdwc,jj0,bLR);
 780                         }
 781                         add_j_to_nblist(coul,jj0+1,bLR);
 782                         add_j_to_nblist(coul,jj0+2,bLR);
 783 #else
 784                         /* One entry for the entire water-water interaction */
 785                         if (!bDoVdW)
 786                         {
 787                             add_j_to_nblist(coul_ww,jj0,bLR);
 788                         }
 789                         else
 790                         {
 791                             add_j_to_nblist(vdwc_ww,jj0,bLR);
 792                         }
 793 #endif
 794                     }
 795                 }
 796                 else if (iwater == esolTIP4P && jwater == esolTIP4P)
 797                 {
 798                     /* Interaction between two TIP4p molecules */
 799                     if (!bDoCoul)
 800                     {
 801                         /* VdW only - only first atoms in each water interact */
 802                         add_j_to_nblist(vdw,jj0,bLR);
 803                     }
 804                     else
 805                     {
 806 #ifdef DISABLE_WATERWATER_NLIST
 807                         /* Add entries for the four atoms - only do VdW if we need to */
 808                         if (bDoVdW)
 809                         {
 810                             add_j_to_nblist(vdw,jj0,bLR);
 811                         }
 812                         add_j_to_nblist(coul,jj0+1,bLR);
 813                         add_j_to_nblist(coul,jj0+2,bLR);
 814                         add_j_to_nblist(coul,jj0+3,bLR);
 815 #else
 816                         /* One entry for the entire water-water interaction */
 817                         if (!bDoVdW)
 818                         {
 819                             add_j_to_nblist(coul_ww,jj0,bLR);
 820                         }
 821                         else
 822                         {
 823                             add_j_to_nblist(vdwc_ww,jj0,bLR);
 824                         }
 825 #endif
 826                     }
 827                 }
 828                 else
 829                 {
 830                     /* j charge group is not water, but i is.
 831                      * Add entries to the water-other_atom lists; the geometry of the water
 832                      * molecule doesn't matter - that is taken care of in the nonbonded kernel,
 833                      * so we don't care if it is SPC or TIP4P...
 834                      */
 835
 836                     jj1 = index[jcg+1];
 837
 838                     if (!bDoVdW)
 839                     {
 840                         for(jj=jj0; (jj<jj1); jj++)
 841                         {
 842                             if (charge[jj] != 0)
 843                             {
 844                                 add_j_to_nblist(coul,jj,bLR);
 845                             }
 846                         }
 847                     }
 848                     else if (!bDoCoul)
 849                     {
 850                         for(jj=jj0; (jj<jj1); jj++)
 851                         {
 852                             if (bHaveVdW[type[jj]])
 853                             {
 854                                 add_j_to_nblist(vdw,jj,bLR);
 855                             }
 856                         }
 857                     }
 858                     else
 859                     {
 860                         /* _charge_ _groups_ interact with both coulomb and LJ */
 861                         /* Check which atoms we should add to the lists!       */
 862                         for(jj=jj0; (jj<jj1); jj++)
 863                         {
 864                             if (bHaveVdW[type[jj]])
 865                             {
 866                                 if (charge[jj] != 0)
 867                                 {
 868                                     add_j_to_nblist(vdwc,jj,bLR);
 869                                 }
 870                                 else
 871                                 {
 872                                     add_j_to_nblist(vdw,jj,bLR);
 873                                 }
 874                             }
 875                             else if (charge[jj] != 0)
 876                             {
 877                                 add_j_to_nblist(coul,jj,bLR);
 878                             }
 879                         }
 880                     }
 881                 }
 882             }
 883             close_i_nblist(vdw);
 884             close_i_nblist(coul);
 885             close_i_nblist(vdwc);
 886 #ifndef DISABLE_WATERWATER_NLIST
 887             close_i_nblist(coul_ww);
 888             close_i_nblist(vdwc_ww);
 889 #endif
 890         }
 891         else
 892         {
 893             /* no solvent as i charge group */
 894             /* Loop over the atoms in the i charge group */
 895             for(i=0; i<nicg; i++)
 896             {
 897                 i_atom  = i0+i;
 898                 gid     = GID(igid,jgid,ngid);
 899                 qi      = charge[i_atom];
 900
 901                 /* Create new i_atom for each energy group */
 902                 if (bDoVdW && bDoCoul)
 903                 {
 904                     new_i_nblist(vdwc,bLR,i_atom,shift,gid);
 905                 }
 906                 if (bDoVdW)
 907                 {
 908                     new_i_nblist(vdw,bLR,i_atom,shift,gid);
 909                 }
 910                 if (bDoCoul)
 911                 {
 912                     new_i_nblist(coul,bLR,i_atom,shift,gid);
 913                 }
 914                 bDoVdW_i  = (bDoVdW  && bHaveVdW[type[i_atom]]);
 915                 bDoCoul_i = (bDoCoul && qi!=0);
 916
 917                 if (bDoVdW_i || bDoCoul_i)
 918                 {
 919                     /* Loop over the j charge groups */
 920                     for(j=0; (j<nj); j++)
 921                     {
 922                         jcg=jjcg[j];
 923
 924                         /* Check for large charge groups */
 925                         if (jcg == icg)
 926                         {
 927                             jj0 = i0 + i + 1;
 928                         }
 929                         else
 930                         {
 931                             jj0 = index[jcg];
 932                         }
 933
 934                         jj1=index[jcg+1];
 935                         /* Finally loop over the atoms in the j-charge group */
 936                         for(jj=jj0; jj<jj1; jj++)
 937                         {
 938                             bNotEx = NOTEXCL(bExcl,i,jj);
 939
 940                             if (bNotEx)
 941                             {
 942                                 if (!bDoVdW_i)
 943                                 {
 944                                     if (charge[jj] != 0)
 945                                     {
 946                                         add_j_to_nblist(coul,jj,bLR);
 947                                     }
 948                                 }
 949                                 else if (!bDoCoul_i)
 950                                 {
 951                                     if (bHaveVdW[type[jj]])
 952                                     {
 953                                         add_j_to_nblist(vdw,jj,bLR);
 954                                     }
 955                                 }
 956                                 else
 957                                 {
 958                                     if (bHaveVdW[type[jj]])
 959                                     {
 960                                         if (charge[jj] != 0)
 961                                         {
 962                                             add_j_to_nblist(vdwc,jj,bLR);
 963                                         }
 964                                         else
 965                                         {
 966                                             add_j_to_nblist(vdw,jj,bLR);
 967                                         }
 968                                     }
 969                                     else if (charge[jj] != 0)
 970                                     {
 971                                         add_j_to_nblist(coul,jj,bLR);
 972                                     }
 973                                 }
 974                             }
 975                         }
 976                     }
 977                 }
 978                 close_i_nblist(vdw);
 979                 close_i_nblist(coul);
 980                 close_i_nblist(vdwc);
 981             }
 982         }
 983     }
 984     else
 985     {
 986         /* we are doing free energy */
 987         vdwc_free = &nlist[eNL_VDWQQ_FREE];
 988         vdw_free  = &nlist[eNL_VDW_FREE];
 989         coul_free = &nlist[eNL_QQ_FREE];
 990         /* Loop over the atoms in the i charge group */
 991         for(i=0; i<nicg; i++)
 992         {
 993             i_atom  = i0+i;
 994             gid     = GID(igid,jgid,ngid);
 995             qi      = charge[i_atom];
 996             qiB     = chargeB[i_atom];
 997
 998             /* Create new i_atom for each energy group */
 999             if (bDoVdW && bDoCoul)
1000                 new_i_nblist(vdwc,bLR,i_atom,shift,gid);
1001             if (bDoVdW)
1002                 new_i_nblist(vdw,bLR,i_atom,shift,gid);
1003             if (bDoCoul)
1004                 new_i_nblist(coul,bLR,i_atom,shift,gid);
1005
1006             new_i_nblist(vdw_free,bLR,i_atom,shift,gid);
1007             new_i_nblist(coul_free,bLR,i_atom,shift,gid);
1008             new_i_nblist(vdwc_free,bLR,i_atom,shift,gid);
1009
1010             bDoVdW_i  = (bDoVdW  &&
1011                          (bHaveVdW[type[i_atom]] || bHaveVdW[typeB[i_atom]]));
1012             bDoCoul_i = (bDoCoul && (qi!=0 || qiB!=0));
1013             /* For TIP4P the first atom does not have a charge,
1014              * but the last three do. So we should still put an atom
1015              * without LJ but with charge in the water-atom neighborlist
1016              * for a TIP4p i charge group.
1017              * For SPC type water the first atom has LJ and charge,
1018              * so there is no such problem.
1019              */
1020             if (iwater == esolNO)
1021             {
1022                 bDoCoul_i_sol = bDoCoul_i;
1023             }
1024             else
1025             {
1026                 bDoCoul_i_sol = bDoCoul;
1027             }
1028
1029             if (bDoVdW_i || bDoCoul_i_sol)
1030             {
1031                 /* Loop over the j charge groups */
1032                 for(j=0; (j<nj); j++)
1033                 {
1034                     jcg=jjcg[j];
1035
1036                     /* Check for large charge groups */
1037                     if (jcg == icg)
1038                     {
1039                         jj0 = i0 + i + 1;
1040                     }
1041                     else
1042                     {
1043                         jj0 = index[jcg];
1044                     }
1045
1046                     jj1=index[jcg+1];
1047                     /* Finally loop over the atoms in the j-charge group */
1048                     bFree = bPert[i_atom];
1049                     for(jj=jj0; (jj<jj1); jj++)
1050                     {
1051                         bFreeJ = bFree || bPert[jj];
1052                         /* Complicated if, because the water H's should also
1053                          * see perturbed j-particles
1054                          */
1055                         if (iwater==esolNO || i==0 || bFreeJ)
1056                         {
1057                             bNotEx = NOTEXCL(bExcl,i,jj);
1058
1059                             if (bNotEx)
1060                             {
1061                                 if (bFreeJ)
1062                                 {
1063                                     if (!bDoVdW_i)
1064                                     {
1065                                         if (charge[jj]!=0 || chargeB[jj]!=0)
1066                                         {
1067                                             add_j_to_nblist(coul_free,jj,bLR);
1068                                         }
1069                                     }
1070                                     else if (!bDoCoul_i)
1071                                     {
1072                                         if (bHaveVdW[type[jj]] || bHaveVdW[typeB[jj]])
1073                                         {
1074                                             add_j_to_nblist(vdw_free,jj,bLR);
1075                                         }
1076                                     }
1077                                     else
1078                                     {
1079                                         if (bHaveVdW[type[jj]] || bHaveVdW[typeB[jj]])
1080                                         {
1081                                             if (charge[jj]!=0 || chargeB[jj]!=0)
1082                                             {
1083                                                 add_j_to_nblist(vdwc_free,jj,bLR);
1084                                             }
1085                                             else
1086                                             {
1087                                                 add_j_to_nblist(vdw_free,jj,bLR);
1088                                             }
1089                                         }
1090                                         else if (charge[jj]!=0 || chargeB[jj]!=0)
1091                                             add_j_to_nblist(coul_free,jj,bLR);
1092                                     }
1093                                 }
1094                                 else if (!bDoVdW_i)
1095                                 {
1096                                     /* This is done whether or not bWater is set */
1097                                     if (charge[jj] != 0)
1098                                     {
1099                                         add_j_to_nblist(coul,jj,bLR);
1100                                     }
1101                                 }
1102                                 else if (!bDoCoul_i_sol)
1103                                 {
1104                                     if (bHaveVdW[type[jj]])
1105                                     {
1106                                         add_j_to_nblist(vdw,jj,bLR);
1107                                     }
1108                                 }
1109                                 else
1110                                 {
1111                                     if (bHaveVdW[type[jj]])
1112                                     {
1113                                         if (charge[jj] != 0)
1114                                         {
1115                                             add_j_to_nblist(vdwc,jj,bLR);
1116                                         }
1117                                         else
1118                                         {
1119                                             add_j_to_nblist(vdw,jj,bLR);
1120                                         }
1121                                     }
1122                                     else if (charge[jj] != 0)
1123                                     {
1124                                         add_j_to_nblist(coul,jj,bLR);
1125                                     }
1126                                 }
1127                             }
1128                         }
1129                     }
1130                 }
1131             }
1132             close_i_nblist(vdw);
1133             close_i_nblist(coul);
1134             close_i_nblist(vdwc);
1135             close_i_nblist(vdw_free);
1136             close_i_nblist(coul_free);
1137             close_i_nblist(vdwc_free);
1138         }
1139     }
1140 }
1141
1142 static void
1143 put_in_list_qmmm(gmx_bool              bHaveVdW[],
1144                  int               ngid,
1145                  t_mdatoms *       md,
1146                  int               icg,
1147                  int               jgid,
1148                  int               nj,
1149                  atom_id           jjcg[],
1150                  atom_id           index[],
1151                  t_excl            bExcl[],
1152                  int               shift,
1153                  t_forcerec *      fr,
1154                  gmx_bool              bLR,
1155                  gmx_bool              bDoVdW,
1156                  gmx_bool              bDoCoul)
1157 {
1158     t_nblist *   coul;
1159     int           i,j,jcg,igid,gid;
1160     atom_id   jj,jj0,jj1,i_atom;
1161     int       i0,nicg;
1162     gmx_bool      bNotEx;
1163
1164     /* Get atom range */
1165     i0     = index[icg];
1166     nicg   = index[icg+1]-i0;
1167
1168     /* Get the i charge group info */
1169     igid   = GET_CGINFO_GID(fr->cginfo[icg]);
1170
1171     coul = &fr->QMMMlist;
1172
1173     /* Loop over atoms in the ith charge group */
1174     for (i=0;i<nicg;i++)
1175     {
1176         i_atom = i0+i;
1177         gid    = GID(igid,jgid,ngid);
1178         /* Create new i_atom for each energy group */
1179         new_i_nblist(coul,bLR,i_atom,shift,gid);
1180
1181         /* Loop over the j charge groups */
1182         for (j=0;j<nj;j++)
1183         {
1184             jcg=jjcg[j];
1185
1186             /* Charge groups cannot have QM and MM atoms simultaneously */
1187             if (jcg!=icg)
1188             {
1189                 jj0 = index[jcg];
1190                 jj1 = index[jcg+1];
1191                 /* Finally loop over the atoms in the j-charge group */
1192                 for(jj=jj0; jj<jj1; jj++)
1193                 {
1194                     bNotEx = NOTEXCL(bExcl,i,jj);
1195                     if(bNotEx)
1196                         add_j_to_nblist(coul,jj,bLR);
1197                 }
1198             }
1199         }
1200         close_i_nblist(coul);
1201     }
1202 }
1203
1204 static void
1205 put_in_list_cg(gmx_bool              bHaveVdW[],
1206                int               ngid,
1207                t_mdatoms *       md,
1208                int               icg,
1209                int               jgid,
1210                int               nj,
1211                atom_id           jjcg[],
1212                atom_id           index[],
1213                t_excl            bExcl[],
1214                int               shift,
1215                t_forcerec *      fr,
1216                gmx_bool              bLR,
1217                gmx_bool              bDoVdW,
1218                gmx_bool              bDoCoul)
1219 {
1220     int          cginfo;
1221     int          igid,gid,nbl_ind;
1222     t_nblist *   vdwc;
1223     int          j,jcg;
1224
1225     cginfo = fr->cginfo[icg];
1226
1227     igid = GET_CGINFO_GID(cginfo);
1228     gid  = GID(igid,jgid,ngid);
1229
1230     /* Unpack pointers to neighbourlist structs */
1231     if (fr->nnblists == 1)
1232     {
1233         nbl_ind = 0;
1234     }
1235     else
1236     {
1237         nbl_ind = fr->gid2nblists[gid];
1238     }
1239     if (bLR)
1240     {
1241         vdwc = &fr->nblists[nbl_ind].nlist_lr[eNL_VDWQQ];
1242     }
1243     else
1244     {
1245         vdwc = &fr->nblists[nbl_ind].nlist_sr[eNL_VDWQQ];
1246     }
1247
1248     /* Make a new neighbor list for charge group icg.
1249      * Currently simply one neighbor list is made with LJ and Coulomb.
1250      * If required, zero interactions could be removed here
1251      * or in the force loop.
1252      */
1253     new_i_nblist(vdwc,bLR,index[icg],shift,gid);
1254     vdwc->iinr_end[vdwc->nri] = index[icg+1];
1255
1256     for(j=0; (j<nj); j++)
1257     {
1258         jcg = jjcg[j];
1259         /* Skip the icg-icg pairs if all self interactions are excluded */
1260         if (!(jcg == icg && GET_CGINFO_EXCL_INTRA(cginfo)))
1261         {
1262             /* Here we add the j charge group jcg to the list,
1263              * exclusions are also added to the list.
1264              */
1265             add_j_to_nblist_cg(vdwc,index[jcg],index[jcg+1],bExcl,icg==jcg,bLR);
1266         }
1267     }
1268
1269     close_i_nblist(vdwc);
1270 }
1271
1272 static void setexcl(atom_id start,atom_id end,t_blocka *excl,gmx_bool b,
1273                     t_excl bexcl[])
1274 {
1275     atom_id i,k;
1276
1277     if (b)
1278     {
1279         for(i=start; i<end; i++)
1280         {
1281             for(k=excl->index[i]; k<excl->index[i+1]; k++)
1282             {
1283                 SETEXCL(bexcl,i-start,excl->a[k]);
1284             }
1285         }
1286     }
1287     else
1288     {
1289         for(i=start; i<end; i++)
1290         {
1291             for(k=excl->index[i]; k<excl->index[i+1]; k++)
1292             {
1293                 RMEXCL(bexcl,i-start,excl->a[k]);
1294             }
1295         }
1296     }
1297 }
1298
1299 int calc_naaj(int icg,int cgtot)
1300 {
1301     int naaj;
1302
1303     if ((cgtot % 2) == 1)
1304     {
1305         /* Odd number of charge groups, easy */
1306         naaj = 1 + (cgtot/2);
1307     }
1308     else if ((cgtot % 4) == 0)
1309     {
1310     /* Multiple of four is hard */
1311         if (icg < cgtot/2)
1312         {
1313             if ((icg % 2) == 0)
1314             {
1315                 naaj=1+(cgtot/2);
1316             }
1317             else
1318             {
1319                 naaj=cgtot/2;
1320             }
1321         }
1322         else
1323         {
1324             if ((icg % 2) == 1)
1325             {
1326                 naaj=1+(cgtot/2);
1327             }
1328             else
1329             {
1330                 naaj=cgtot/2;
1331             }
1332         }
1333     }
1334     else
1335     {
1336         /* cgtot/2 = odd */
1337         if ((icg % 2) == 0)
1338         {
1339             naaj=1+(cgtot/2);
1340         }
1341         else
1342         {
1343             naaj=cgtot/2;
1344         }
1345     }
1346 #ifdef DEBUG
1347     fprintf(log,"naaj=%d\n",naaj);
1348 #endif
1349
1350     return naaj;
1351 }
1352
1353 /************************************************
1354  *
1355  *  S I M P L E      C O R E     S T U F F
1356  *
1357  ************************************************/
1358
1359 static real calc_image_tric(rvec xi,rvec xj,matrix box,
1360                             rvec b_inv,int *shift)
1361 {
1362     /* This code assumes that the cut-off is smaller than
1363      * a half times the smallest diagonal element of the box.
1364      */
1365     const real h25=2.5;
1366     real dx,dy,dz;
1367     real r2;
1368     int  tx,ty,tz;
1369
1370     /* Compute diff vector */
1371     dz = xj[ZZ] - xi[ZZ];
1372     dy = xj[YY] - xi[YY];
1373     dx = xj[XX] - xi[XX];
1374
1375   /* Perform NINT operation, using trunc operation, therefore
1376    * we first add 2.5 then subtract 2 again
1377    */
1378     tz = dz*b_inv[ZZ] + h25;
1379     tz -= 2;
1380     dz -= tz*box[ZZ][ZZ];
1381     dy -= tz*box[ZZ][YY];
1382     dx -= tz*box[ZZ][XX];
1383
1384     ty = dy*b_inv[YY] + h25;
1385     ty -= 2;
1386     dy -= ty*box[YY][YY];
1387     dx -= ty*box[YY][XX];
1388
1389     tx = dx*b_inv[XX]+h25;
1390     tx -= 2;
1391     dx -= tx*box[XX][XX];
1392
1393     /* Distance squared */
1394     r2 = (dx*dx) + (dy*dy) + (dz*dz);
1395
1396     *shift = XYZ2IS(tx,ty,tz);
1397
1398     return r2;
1399 }
1400
1401 static real calc_image_rect(rvec xi,rvec xj,rvec box_size,
1402                             rvec b_inv,int *shift)
1403 {
1404     const real h15=1.5;
1405     real ddx,ddy,ddz;
1406     real dx,dy,dz;
1407     real r2;
1408     int  tx,ty,tz;
1409
1410     /* Compute diff vector */
1411     dx = xj[XX] - xi[XX];
1412     dy = xj[YY] - xi[YY];
1413     dz = xj[ZZ] - xi[ZZ];
1414
1415     /* Perform NINT operation, using trunc operation, therefore
1416      * we first add 1.5 then subtract 1 again
1417      */
1418     tx = dx*b_inv[XX] + h15;
1419     ty = dy*b_inv[YY] + h15;
1420     tz = dz*b_inv[ZZ] + h15;
1421     tx--;
1422     ty--;
1423     tz--;
1424
1425     /* Correct diff vector for translation */
1426     ddx = tx*box_size[XX] - dx;
1427     ddy = ty*box_size[YY] - dy;
1428     ddz = tz*box_size[ZZ] - dz;
1429
1430     /* Distance squared */
1431     r2 = (ddx*ddx) + (ddy*ddy) + (ddz*ddz);
1432
1433     *shift = XYZ2IS(tx,ty,tz);
1434
1435     return r2;
1436 }
1437
1438 static void add_simple(t_ns_buf *nsbuf,int nrj,atom_id cg_j,
1439                        gmx_bool bHaveVdW[],int ngid,t_mdatoms *md,
1440                        int icg,int jgid,t_block *cgs,t_excl bexcl[],
1441                        int shift,t_forcerec *fr,put_in_list_t *put_in_list)
1442 {
1443     if (nsbuf->nj + nrj > MAX_CG)
1444     {
1445         put_in_list(bHaveVdW,ngid,md,icg,jgid,nsbuf->ncg,nsbuf->jcg,
1446                     cgs->index,bexcl,shift,fr,FALSE,TRUE,TRUE);
1447         /* Reset buffer contents */
1448         nsbuf->ncg = nsbuf->nj = 0;
1449     }
1450     nsbuf->jcg[nsbuf->ncg++] = cg_j;
1451     nsbuf->nj += nrj;
1452 }
1453
1454 static void ns_inner_tric(rvec x[],int icg,int *i_egp_flags,
1455                           int njcg,atom_id jcg[],
1456                           matrix box,rvec b_inv,real rcut2,
1457                           t_block *cgs,t_ns_buf **ns_buf,
1458                           gmx_bool bHaveVdW[],int ngid,t_mdatoms *md,
1459                           t_excl bexcl[],t_forcerec *fr,
1460                           put_in_list_t *put_in_list)
1461 {
1462     int      shift;
1463     int      j,nrj,jgid;
1464     int      *cginfo=fr->cginfo;
1465     atom_id  cg_j,*cgindex;
1466     t_ns_buf *nsbuf;
1467
1468     cgindex = cgs->index;
1469     shift   = CENTRAL;
1470     for(j=0; (j<njcg); j++)
1471     {
1472         cg_j   = jcg[j];
1473         nrj    = cgindex[cg_j+1]-cgindex[cg_j];
1474         if (calc_image_tric(x[icg],x[cg_j],box,b_inv,&shift) < rcut2)
1475         {
1476             jgid  = GET_CGINFO_GID(cginfo[cg_j]);
1477             if (!(i_egp_flags[jgid] & EGP_EXCL))
1478             {
1479                 add_simple(&ns_buf[jgid][shift],nrj,cg_j,
1480                            bHaveVdW,ngid,md,icg,jgid,cgs,bexcl,shift,fr,
1481                            put_in_list);
1482             }
1483         }
1484     }
1485 }
1486
1487 static void ns_inner_rect(rvec x[],int icg,int *i_egp_flags,
1488                           int njcg,atom_id jcg[],
1489                           gmx_bool bBox,rvec box_size,rvec b_inv,real rcut2,
1490                           t_block *cgs,t_ns_buf **ns_buf,
1491                           gmx_bool bHaveVdW[],int ngid,t_mdatoms *md,
1492                           t_excl bexcl[],t_forcerec *fr,
1493                           put_in_list_t *put_in_list)
1494 {
1495     int      shift;
1496     int      j,nrj,jgid;
1497     int      *cginfo=fr->cginfo;
1498     atom_id  cg_j,*cgindex;
1499     t_ns_buf *nsbuf;
1500
1501     cgindex = cgs->index;
1502     if (bBox)
1503     {
1504         shift = CENTRAL;
1505         for(j=0; (j<njcg); j++)
1506         {
1507             cg_j   = jcg[j];
1508             nrj    = cgindex[cg_j+1]-cgindex[cg_j];
1509             if (calc_image_rect(x[icg],x[cg_j],box_size,b_inv,&shift) < rcut2)
1510             {
1511                 jgid  = GET_CGINFO_GID(cginfo[cg_j]);
1512                 if (!(i_egp_flags[jgid] & EGP_EXCL))
1513                 {
1514                     add_simple(&ns_buf[jgid][shift],nrj,cg_j,
1515                                bHaveVdW,ngid,md,icg,jgid,cgs,bexcl,shift,fr,
1516                                put_in_list);
1517                 }
1518             }
1519         }
1520     }
1521     else
1522     {
1523         for(j=0; (j<njcg); j++)
1524         {
1525             cg_j   = jcg[j];
1526             nrj    = cgindex[cg_j+1]-cgindex[cg_j];
1527             if ((rcut2 == 0) || (distance2(x[icg],x[cg_j]) < rcut2)) {
1528                 jgid  = GET_CGINFO_GID(cginfo[cg_j]);
1529                 if (!(i_egp_flags[jgid] & EGP_EXCL))
1530                 {
1531                     add_simple(&ns_buf[jgid][CENTRAL],nrj,cg_j,
1532                                bHaveVdW,ngid,md,icg,jgid,cgs,bexcl,CENTRAL,fr,
1533                                put_in_list);
1534                 }
1535             }
1536         }
1537     }
1538 }
1539
1540 /* ns_simple_core needs to be adapted for QMMM still 2005 */
1541
1542 static int ns_simple_core(t_forcerec *fr,
1543                           gmx_localtop_t *top,
1544                           t_mdatoms *md,
1545                           matrix box,rvec box_size,
1546                           t_excl bexcl[],atom_id *aaj,
1547                           int ngid,t_ns_buf **ns_buf,
1548                           put_in_list_t *put_in_list,gmx_bool bHaveVdW[])
1549 {
1550     int      naaj,k;
1551     real     rlist2;
1552     int      nsearch,icg,jcg,igid,i0,nri,nn;
1553     int      *cginfo;
1554     t_ns_buf *nsbuf;
1555     /* atom_id  *i_atoms; */
1556     t_block  *cgs=&(top->cgs);
1557     t_blocka *excl=&(top->excls);
1558     rvec     b_inv;
1559     int      m;
1560     gmx_bool     bBox,bTriclinic;
1561     int      *i_egp_flags;
1562
1563     rlist2 = sqr(fr->rlist);
1564
1565     bBox = (fr->ePBC != epbcNONE);
1566     if (bBox)
1567     {
1568         for(m=0; (m<DIM); m++)
1569         {
1570             b_inv[m] = divide_err(1.0,box_size[m]);
1571         }
1572         bTriclinic = TRICLINIC(box);
1573     }
1574     else
1575     {
1576         bTriclinic = FALSE;
1577     }
1578
1579     cginfo = fr->cginfo;
1580
1581     nsearch=0;
1582     for (icg=fr->cg0; (icg<fr->hcg); icg++)
1583     {
1584         /*
1585           i0        = cgs->index[icg];
1586           nri       = cgs->index[icg+1]-i0;
1587           i_atoms   = &(cgs->a[i0]);
1588           i_eg_excl = fr->eg_excl + ngid*md->cENER[*i_atoms];
1589           setexcl(nri,i_atoms,excl,TRUE,bexcl);
1590         */
1591         igid = GET_CGINFO_GID(cginfo[icg]);
1592         i_egp_flags = fr->egp_flags + ngid*igid;
1593         setexcl(cgs->index[icg],cgs->index[icg+1],excl,TRUE,bexcl);
1594
1595         naaj=calc_naaj(icg,cgs->nr);
1596         if (bTriclinic)
1597         {
1598             ns_inner_tric(fr->cg_cm,icg,i_egp_flags,naaj,&(aaj[icg]),
1599                           box,b_inv,rlist2,cgs,ns_buf,
1600                           bHaveVdW,ngid,md,bexcl,fr,put_in_list);
1601         }
1602         else
1603         {
1604             ns_inner_rect(fr->cg_cm,icg,i_egp_flags,naaj,&(aaj[icg]),
1605                           bBox,box_size,b_inv,rlist2,cgs,ns_buf,
1606                           bHaveVdW,ngid,md,bexcl,fr,put_in_list);
1607         }
1608         nsearch += naaj;
1609
1610         for(nn=0; (nn<ngid); nn++)
1611         {
1612             for(k=0; (k<SHIFTS); k++)
1613             {
1614                 nsbuf = &(ns_buf[nn][k]);
1615                 if (nsbuf->ncg > 0)
1616                 {
1617                     put_in_list(bHaveVdW,ngid,md,icg,nn,nsbuf->ncg,nsbuf->jcg,
1618                                 cgs->index,bexcl,k,fr,FALSE,TRUE,TRUE);
1619                     nsbuf->ncg=nsbuf->nj=0;
1620                 }
1621             }
1622         }
1623         /* setexcl(nri,i_atoms,excl,FALSE,bexcl); */
1624         setexcl(cgs->index[icg],cgs->index[icg+1],excl,FALSE,bexcl);
1625     }
1626     close_neighbor_list(fr,FALSE,-1,-1,FALSE);
1627
1628     return nsearch;
1629 }
1630
1631 /************************************************
1632  *
1633  *    N S 5     G R I D     S T U F F
1634  *
1635  ************************************************/
1636
1637 static inline void get_dx(int Nx,real gridx,real rc2,int xgi,real x,
1638                           int *dx0,int *dx1,real *dcx2)
1639 {
1640     real dcx,tmp;
1641     int  xgi0,xgi1,i;
1642
1643     if (xgi < 0)
1644     {
1645         *dx0 = 0;
1646         xgi0 = -1;
1647         *dx1 = -1;
1648         xgi1 = 0;
1649     }
1650     else if (xgi >= Nx)
1651     {
1652         *dx0 = Nx;
1653         xgi0 = Nx-1;
1654         *dx1 = Nx-1;
1655         xgi1 = Nx;
1656     }
1657     else
1658     {
1659         dcx2[xgi] = 0;
1660         *dx0 = xgi;
1661         xgi0 = xgi-1;
1662         *dx1 = xgi;
1663         xgi1 = xgi+1;
1664     }
1665
1666     for(i=xgi0; i>=0; i--)
1667     {
1668         dcx = (i+1)*gridx-x;
1669         tmp = dcx*dcx;
1670         if (tmp >= rc2)
1671             break;
1672         *dx0 = i;
1673         dcx2[i] = tmp;
1674     }
1675     for(i=xgi1; i<Nx; i++)
1676     {
1677         dcx = i*gridx-x;
1678         tmp = dcx*dcx;
1679         if (tmp >= rc2)
1680         {
1681             break;
1682         }
1683         *dx1 = i;
1684         dcx2[i] = tmp;
1685     }
1686 }
1687
1688 static inline void get_dx_dd(int Nx,real gridx,real rc2,int xgi,real x,
1689                              int ncpddc,int shift_min,int shift_max,
1690                              int *g0,int *g1,real *dcx2)
1691 {
1692     real dcx,tmp;
1693     int  g_min,g_max,shift_home;
1694
1695     if (xgi < 0)
1696     {
1697         g_min = 0;
1698         g_max = Nx - 1;
1699         *g0   = 0;
1700         *g1   = -1;
1701     }
1702     else if (xgi >= Nx)
1703     {
1704         g_min = 0;
1705         g_max = Nx - 1;
1706         *g0   = Nx;
1707         *g1   = Nx - 1;
1708     }
1709     else
1710     {
1711         if (ncpddc == 0)
1712         {
1713             g_min = 0;
1714             g_max = Nx - 1;
1715         }
1716         else
1717         {
1718             if (xgi < ncpddc)
1719             {
1720                 shift_home = 0;
1721             }
1722             else
1723             {
1724                 shift_home = -1;
1725             }
1726             g_min = (shift_min == shift_home ? 0          : ncpddc);
1727             g_max = (shift_max == shift_home ? ncpddc - 1 : Nx - 1);
1728         }
1729         if (shift_min > 0)
1730         {
1731             *g0 = g_min;
1732             *g1 = g_min - 1;
1733         }
1734         else if (shift_max < 0)
1735         {
1736             *g0 = g_max + 1;
1737             *g1 = g_max;
1738         }
1739         else
1740         {
1741             *g0 = xgi;
1742             *g1 = xgi;
1743             dcx2[xgi] = 0;
1744         }
1745     }
1746
1747     while (*g0 > g_min)
1748     {
1749         /* Check one grid cell down */
1750         dcx = ((*g0 - 1) + 1)*gridx - x;
1751         tmp = dcx*dcx;
1752         if (tmp >= rc2)
1753         {
1754             break;
1755         }
1756         (*g0)--;
1757         dcx2[*g0] = tmp;
1758     }
1759
1760     while (*g1 < g_max)
1761     {
1762         /* Check one grid cell up */
1763         dcx = (*g1 + 1)*gridx - x;
1764         tmp = dcx*dcx;
1765         if (tmp >= rc2)
1766         {
1767             break;
1768         }
1769         (*g1)++;
1770         dcx2[*g1] = tmp;
1771     }
1772 }
1773
1774
1775 #define sqr(x) ((x)*(x))
1776 #define calc_dx2(XI,YI,ZI,y) (sqr(XI-y[XX]) + sqr(YI-y[YY]) + sqr(ZI-y[ZZ]))
1777 #define calc_cyl_dx2(XI,YI,y) (sqr(XI-y[XX]) + sqr(YI-y[YY]))
1778 /****************************************************
1779  *
1780  *    F A S T   N E I G H B O R  S E A R C H I N G
1781  *
1782  *    Optimized neighboursearching routine using grid
1783  *    at least 1x1x1, see GROMACS manual
1784  *
1785  ****************************************************/
1786
1787 static void do_longrange(t_commrec *cr,gmx_localtop_t *top,t_forcerec *fr,
1788                          int ngid,t_mdatoms *md,int icg,
1789                          int jgid,int nlr,
1790                          atom_id lr[],t_excl bexcl[],int shift,
1791                          rvec x[],rvec box_size,t_nrnb *nrnb,
1792                          real lambda,real *dvdlambda,
1793                          gmx_grppairener_t *grppener,
1794                          gmx_bool bDoVdW,gmx_bool bDoCoul,
1795                          gmx_bool bEvaluateNow,put_in_list_t *put_in_list,
1796                          gmx_bool bHaveVdW[],
1797                          gmx_bool bDoForces,rvec *f)
1798 {
1799     int n,i;
1800     t_nblist *nl;
1801
1802     for(n=0; n<fr->nnblists; n++)
1803     {
1804         for(i=0; (i<eNL_NR); i++)
1805         {
1806             nl = &fr->nblists[n].nlist_lr[i];
1807             if ((nl->nri > nl->maxnri-32) || bEvaluateNow)
1808             {
1809                 close_neighbor_list(fr,TRUE,n,i,FALSE);
1810                 /* Evaluate the energies and forces */
1811                 do_nonbonded(cr,fr,x,f,md,NULL,
1812                              grppener->ener[fr->bBHAM ? egBHAMLR : egLJLR],
1813                              grppener->ener[egCOULLR],
1814                                                          grppener->ener[egGB],box_size,
1815                              nrnb,lambda,dvdlambda,n,i,
1816                              GMX_DONB_LR | GMX_DONB_FORCES);
1817
1818                 reset_neighbor_list(fr,TRUE,n,i);
1819             }
1820         }
1821     }
1822
1823     if (!bEvaluateNow)
1824     {
1825         /* Put the long range particles in a list */
1826         /* do_longrange is never called for QMMM  */
1827         put_in_list(bHaveVdW,ngid,md,icg,jgid,nlr,lr,top->cgs.index,
1828                     bexcl,shift,fr,TRUE,bDoVdW,bDoCoul);
1829     }
1830 }
1831
1832 static void get_cutoff2(t_forcerec *fr,gmx_bool bDoLongRange,
1833                         real *rvdw2,real *rcoul2,
1834                         real *rs2,real *rm2,real *rl2)
1835 {
1836     *rs2 = sqr(fr->rlist);
1837     if (bDoLongRange && fr->bTwinRange)
1838     {
1839         /* The VdW and elec. LR cut-off's could be different,
1840          * so we can not simply set them to rlistlong.
1841          */
1842         if (EVDW_MIGHT_BE_ZERO_AT_CUTOFF(fr->vdwtype) &&
1843             fr->rvdw > fr->rlist)
1844         {
1845             *rvdw2  = sqr(fr->rlistlong);
1846         }
1847         else
1848         {
1849             *rvdw2  = sqr(fr->rvdw);
1850         }
1851         if (EEL_MIGHT_BE_ZERO_AT_CUTOFF(fr->eeltype) &&
1852             fr->rcoulomb > fr->rlist)
1853         {
1854             *rcoul2 = sqr(fr->rlistlong);
1855         }
1856         else
1857         {
1858             *rcoul2 = sqr(fr->rcoulomb);
1859         }
1860     }
1861     else
1862     {
1863         /* Workaround for a gcc -O3 or -ffast-math problem */
1864         *rvdw2  = *rs2;
1865         *rcoul2 = *rs2;
1866     }
1867     *rm2 = min(*rvdw2,*rcoul2);
1868     *rl2 = max(*rvdw2,*rcoul2);
1869 }
1870
1871 static void init_nsgrid_lists(t_forcerec *fr,int ngid,gmx_ns_t *ns)
1872 {
1873     real rvdw2,rcoul2,rs2,rm2,rl2;
1874     int j;
1875
1876     get_cutoff2(fr,TRUE,&rvdw2,&rcoul2,&rs2,&rm2,&rl2);
1877
1878     /* Short range buffers */
1879     snew(ns->nl_sr,ngid);
1880     /* Counters */
1881     snew(ns->nsr,ngid);
1882     snew(ns->nlr_ljc,ngid);
1883     snew(ns->nlr_one,ngid);
1884
1885     if (rm2 > rs2)
1886     {
1887             /* Long range VdW and Coul buffers */
1888         snew(ns->nl_lr_ljc,ngid);
1889     }
1890     if (rl2 > rm2)
1891     {
1892         /* Long range VdW or Coul only buffers */
1893         snew(ns->nl_lr_one,ngid);
1894     }
1895     for(j=0; (j<ngid); j++) {
1896         snew(ns->nl_sr[j],MAX_CG);
1897         if (rm2 > rs2)
1898         {
1899             snew(ns->nl_lr_ljc[j],MAX_CG);
1900         }
1901         if (rl2 > rm2)
1902         {
1903             snew(ns->nl_lr_one[j],MAX_CG);
1904         }
1905     }
1906     if (debug)
1907     {
1908         fprintf(debug,
1909                 "ns5_core: rs2 = %g, rm2 = %g, rl2 = %g (nm^2)\n",
1910                 rs2,rm2,rl2);
1911     }
1912 }
1913
1914 static int nsgrid_core(FILE *log,t_commrec *cr,t_forcerec *fr,
1915                        matrix box,rvec box_size,int ngid,
1916                        gmx_localtop_t *top,
1917                        t_grid *grid,rvec x[],
1918                        t_excl bexcl[],gmx_bool *bExcludeAlleg,
1919                        t_nrnb *nrnb,t_mdatoms *md,
1920                        real lambda,real *dvdlambda,
1921                        gmx_grppairener_t *grppener,
1922                        put_in_list_t *put_in_list,
1923                        gmx_bool bHaveVdW[],
1924                        gmx_bool bDoLongRange,gmx_bool bDoForces,rvec *f,
1925                        gmx_bool bMakeQMMMnblist)
1926 {
1927     gmx_ns_t *ns;
1928     atom_id **nl_lr_ljc,**nl_lr_one,**nl_sr;
1929     int     *nlr_ljc,*nlr_one,*nsr;
1930     gmx_domdec_t *dd=NULL;
1931     t_block *cgs=&(top->cgs);
1932     int     *cginfo=fr->cginfo;
1933     /* atom_id *i_atoms,*cgsindex=cgs->index; */
1934     ivec    sh0,sh1,shp;
1935     int     cell_x,cell_y,cell_z;
1936     int     d,tx,ty,tz,dx,dy,dz,cj;
1937 #ifdef ALLOW_OFFDIAG_LT_HALFDIAG
1938     int     zsh_ty,zsh_tx,ysh_tx;
1939 #endif
1940     int     dx0,dx1,dy0,dy1,dz0,dz1;
1941     int     Nx,Ny,Nz,shift=-1,j,nrj,nns,nn=-1;
1942     real    gridx,gridy,gridz,grid_x,grid_y,grid_z;
1943     real    *dcx2,*dcy2,*dcz2;
1944     int     zgi,ygi,xgi;
1945     int     cg0,cg1,icg=-1,cgsnr,i0,igid,nri,naaj,max_jcg;
1946     int     jcg0,jcg1,jjcg,cgj0,jgid;
1947     int     *grida,*gridnra,*gridind;
1948     gmx_bool    rvdw_lt_rcoul,rcoul_lt_rvdw;
1949     rvec    xi,*cgcm,grid_offset;
1950     real    r2,rs2,rvdw2,rcoul2,rm2,rl2,XI,YI,ZI,dcx,dcy,dcz,tmp1,tmp2;
1951     int     *i_egp_flags;
1952     gmx_bool    bDomDec,bTriclinicX,bTriclinicY;
1953     ivec    ncpddc;
1954
1955     ns = &fr->ns;
1956
1957     bDomDec = DOMAINDECOMP(cr);
1958     if (bDomDec)
1959     {
1960         dd = cr->dd;
1961     }
1962
1963     bTriclinicX = ((YY < grid->npbcdim &&
1964                     (!bDomDec || dd->nc[YY]==1) && box[YY][XX] != 0) ||
1965                    (ZZ < grid->npbcdim &&
1966                     (!bDomDec || dd->nc[ZZ]==1) && box[ZZ][XX] != 0));
1967     bTriclinicY =  (ZZ < grid->npbcdim &&
1968                     (!bDomDec || dd->nc[ZZ]==1) && box[ZZ][YY] != 0);
1969
1970     cgsnr    = cgs->nr;
1971
1972     get_cutoff2(fr,bDoLongRange,&rvdw2,&rcoul2,&rs2,&rm2,&rl2);
1973
1974     rvdw_lt_rcoul = (rvdw2 >= rcoul2);
1975     rcoul_lt_rvdw = (rcoul2 >= rvdw2);
1976
1977     if (bMakeQMMMnblist)
1978     {
1979         rm2 = rl2;
1980         rs2 = rl2;
1981     }
1982
1983     nl_sr     = ns->nl_sr;
1984     nsr       = ns->nsr;
1985     nl_lr_ljc = ns->nl_lr_ljc;
1986     nl_lr_one = ns->nl_lr_one;
1987     nlr_ljc   = ns->nlr_ljc;
1988     nlr_one   = ns->nlr_one;
1989
1990     /* Unpack arrays */
1991     cgcm    = fr->cg_cm;
1992     Nx      = grid->n[XX];
1993     Ny      = grid->n[YY];
1994     Nz      = grid->n[ZZ];
1995     grida   = grid->a;
1996     gridind = grid->index;
1997     gridnra = grid->nra;
1998     nns     = 0;
1999
2000     gridx      = grid->cell_size[XX];
2001     gridy      = grid->cell_size[YY];
2002     gridz      = grid->cell_size[ZZ];
2003     grid_x     = 1/gridx;
2004     grid_y     = 1/gridy;
2005     grid_z     = 1/gridz;
2006     copy_rvec(grid->cell_offset,grid_offset);
2007     copy_ivec(grid->ncpddc,ncpddc);
2008     dcx2       = grid->dcx2;
2009     dcy2       = grid->dcy2;
2010     dcz2       = grid->dcz2;
2011
2012 #ifdef ALLOW_OFFDIAG_LT_HALFDIAG
2013     zsh_ty = floor(-box[ZZ][YY]/box[YY][YY]+0.5);
2014     zsh_tx = floor(-box[ZZ][XX]/box[XX][XX]+0.5);
2015     ysh_tx = floor(-box[YY][XX]/box[XX][XX]+0.5);
2016     if (zsh_tx!=0 && ysh_tx!=0)
2017     {
2018         /* This could happen due to rounding, when both ratios are 0.5 */
2019         ysh_tx = 0;
2020     }
2021 #endif
2022
2023     debug_gmx();
2024
2025     if (fr->n_tpi)
2026     {
2027         /* We only want a list for the test particle */
2028         cg0 = cgsnr - 1;
2029     }
2030     else
2031     {
2032         cg0 = grid->icg0;
2033     }
2034     cg1 = grid->icg1;
2035
2036     /* Set the shift range */
2037     for(d=0; d<DIM; d++)
2038     {
2039         sh0[d] = -1;
2040         sh1[d] = 1;
2041         /* Check if we need periodicity shifts.
2042          * Without PBC or with domain decomposition we don't need them.
2043          */
2044         if (d >= ePBC2npbcdim(fr->ePBC) || (bDomDec && dd->nc[d] > 1))
2045         {
2046             shp[d] = 0;
2047         }
2048         else
2049         {
2050             if (d == XX &&
2051                 box[XX][XX] - fabs(box[YY][XX]) - fabs(box[ZZ][XX]) < sqrt(rl2))
2052             {
2053                 shp[d] = 2;
2054             }
2055             else
2056             {
2057                 shp[d] = 1;
2058             }
2059         }
2060     }
2061
2062     /* Loop over charge groups */
2063     for(icg=cg0; (icg < cg1); icg++)
2064     {
2065         igid = GET_CGINFO_GID(cginfo[icg]);
2066         /* Skip this charge group if all energy groups are excluded! */
2067         if (bExcludeAlleg[igid])
2068         {
2069             continue;
2070         }
2071
2072         i0   = cgs->index[icg];
2073
2074         if (bMakeQMMMnblist)
2075         {
2076             /* Skip this charge group if it is not a QM atom while making a
2077              * QM/MM neighbourlist
2078              */
2079             if (md->bQM[i0]==FALSE)
2080             {
2081                 continue; /* MM particle, go to next particle */
2082             }
2083
2084             /* Compute the number of charge groups that fall within the control
2085              * of this one (icg)
2086              */
2087             naaj    = calc_naaj(icg,cgsnr);
2088             jcg0    = icg;
2089             jcg1    = icg + naaj;
2090             max_jcg = cgsnr;
2091         }
2092         else
2093         {
2094             /* make a normal neighbourlist */
2095
2096             if (bDomDec)
2097             {
2098                 /* Get the j charge-group and dd cell shift ranges */
2099                 dd_get_ns_ranges(cr->dd,icg,&jcg0,&jcg1,sh0,sh1);
2100                 max_jcg = 0;
2101             }
2102             else
2103             {
2104                 /* Compute the number of charge groups that fall within the control
2105                  * of this one (icg)
2106                  */
2107                 naaj = calc_naaj(icg,cgsnr);
2108                 jcg0 = icg;
2109                 jcg1 = icg + naaj;
2110
2111                 if (fr->n_tpi)
2112                 {
2113                     /* The i-particle is awlways the test particle,
2114                      * so we want all j-particles
2115                      */
2116                     max_jcg = cgsnr - 1;
2117                 }
2118                 else
2119                 {
2120                     max_jcg  = jcg1 - cgsnr;
2121                 }
2122             }
2123         }
2124
2125         i_egp_flags = fr->egp_flags + igid*ngid;
2126
2127         /* Set the exclusions for the atoms in charge group icg using a bitmask */
2128         setexcl(i0,cgs->index[icg+1],&top->excls,TRUE,bexcl);
2129
2130         ci2xyz(grid,icg,&cell_x,&cell_y,&cell_z);
2131
2132         /* Changed iicg to icg, DvdS 990115
2133          * (but see consistency check above, DvdS 990330)
2134          */
2135 #ifdef NS5DB
2136         fprintf(log,"icg=%5d, naaj=%5d, cell %d %d %d\n",
2137                 icg,naaj,cell_x,cell_y,cell_z);
2138 #endif
2139         /* Loop over shift vectors in three dimensions */
2140         for (tz=-shp[ZZ]; tz<=shp[ZZ]; tz++)
2141         {
2142             ZI = cgcm[icg][ZZ]+tz*box[ZZ][ZZ];
2143             /* Calculate range of cells in Z direction that have the shift tz */
2144             zgi = cell_z + tz*Nz;
2145 #define FAST_DD_NS
2146 #ifndef FAST_DD_NS
2147             get_dx(Nz,gridz,rl2,zgi,ZI,&dz0,&dz1,dcz2);
2148 #else
2149             get_dx_dd(Nz,gridz,rl2,zgi,ZI-grid_offset[ZZ],
2150                       ncpddc[ZZ],sh0[ZZ],sh1[ZZ],&dz0,&dz1,dcz2);
2151 #endif
2152             if (dz0 > dz1)
2153             {
2154                 continue;
2155             }
2156             for (ty=-shp[YY]; ty<=shp[YY]; ty++)
2157             {
2158                 YI = cgcm[icg][YY]+ty*box[YY][YY]+tz*box[ZZ][YY];
2159                 /* Calculate range of cells in Y direction that have the shift ty */
2160                 if (bTriclinicY)
2161                 {
2162                     ygi = (int)(Ny + (YI - grid_offset[YY])*grid_y) - Ny;
2163                 }
2164                 else
2165                 {
2166                     ygi = cell_y + ty*Ny;
2167                 }
2168 #ifndef FAST_DD_NS
2169                 get_dx(Ny,gridy,rl2,ygi,YI,&dy0,&dy1,dcy2);
2170 #else
2171                 get_dx_dd(Ny,gridy,rl2,ygi,YI-grid_offset[YY],
2172                           ncpddc[YY],sh0[YY],sh1[YY],&dy0,&dy1,dcy2);
2173 #endif
2174                 if (dy0 > dy1)
2175                 {
2176                     continue;
2177                 }
2178                 for (tx=-shp[XX]; tx<=shp[XX]; tx++)
2179                 {
2180                     XI = cgcm[icg][XX]+tx*box[XX][XX]+ty*box[YY][XX]+tz*box[ZZ][XX];
2181                     /* Calculate range of cells in X direction that have the shift tx */
2182                     if (bTriclinicX)
2183                     {
2184                         xgi = (int)(Nx + (XI - grid_offset[XX])*grid_x) - Nx;
2185                     }
2186                     else
2187                     {
2188                         xgi = cell_x + tx*Nx;
2189                     }
2190 #ifndef FAST_DD_NS
2191                     get_dx(Nx,gridx,rl2,xgi*Nx,XI,&dx0,&dx1,dcx2);
2192 #else
2193                     get_dx_dd(Nx,gridx,rl2,xgi,XI-grid_offset[XX],
2194                               ncpddc[XX],sh0[XX],sh1[XX],&dx0,&dx1,dcx2);
2195 #endif
2196                     if (dx0 > dx1)
2197                     {
2198                         continue;
2199                     }
2200                     /* Get shift vector */
2201                     shift=XYZ2IS(tx,ty,tz);
2202 #ifdef NS5DB
2203                     range_check(shift,0,SHIFTS);
2204 #endif
2205                     for(nn=0; (nn<ngid); nn++)
2206                     {
2207                         nsr[nn]      = 0;
2208                         nlr_ljc[nn]  = 0;
2209                         nlr_one[nn] = 0;
2210                     }
2211 #ifdef NS5DB
2212                     fprintf(log,"shift: %2d, dx0,1: %2d,%2d, dy0,1: %2d,%2d, dz0,1: %2d,%2d\n",
2213                             shift,dx0,dx1,dy0,dy1,dz0,dz1);
2214                     fprintf(log,"cgcm: %8.3f  %8.3f  %8.3f\n",cgcm[icg][XX],
2215                             cgcm[icg][YY],cgcm[icg][ZZ]);
2216                     fprintf(log,"xi:   %8.3f  %8.3f  %8.3f\n",XI,YI,ZI);
2217 #endif
2218                     for (dx=dx0; (dx<=dx1); dx++)
2219                     {
2220                         tmp1 = rl2 - dcx2[dx];
2221                         for (dy=dy0; (dy<=dy1); dy++)
2222                         {
2223                             tmp2 = tmp1 - dcy2[dy];
2224                             if (tmp2 > 0)
2225                             {
2226                                 for (dz=dz0; (dz<=dz1); dz++) {
2227                                     if (tmp2 > dcz2[dz]) {
2228                                         /* Find grid-cell cj in which possible neighbours are */
2229                                         cj   = xyz2ci(Ny,Nz,dx,dy,dz);
2230
2231                                         /* Check out how many cgs (nrj) there in this cell */
2232                                         nrj  = gridnra[cj];
2233
2234                                         /* Find the offset in the cg list */
2235                                         cgj0 = gridind[cj];
2236
2237                                         /* Check if all j's are out of range so we
2238                                          * can skip the whole cell.
2239                                          * Should save some time, especially with DD.
2240                                          */
2241                                         if (nrj == 0 ||
2242                                             (grida[cgj0] >= max_jcg &&
2243                                              (grida[cgj0] >= jcg1 || grida[cgj0+nrj-1] < jcg0)))
2244                                         {
2245                                             continue;
2246                                         }
2247
2248                                         /* Loop over cgs */
2249                                         for (j=0; (j<nrj); j++)
2250                                         {
2251                                             jjcg = grida[cgj0+j];
2252
2253                                             /* check whether this guy is in range! */
2254                                             if ((jjcg >= jcg0 && jjcg < jcg1) ||
2255                                                 (jjcg < max_jcg))
2256                                             {
2257                                                 r2=calc_dx2(XI,YI,ZI,cgcm[jjcg]);
2258                                                 if (r2 < rl2) {
2259                                                     /* jgid = gid[cgsatoms[cgsindex[jjcg]]]; */
2260                                                     jgid = GET_CGINFO_GID(cginfo[jjcg]);
2261                                                     /* check energy group exclusions */
2262                                                     if (!(i_egp_flags[jgid] & EGP_EXCL))
2263                                                     {
2264                                                         if (r2 < rs2)
2265                                                         {
2266                                                             if (nsr[jgid] >= MAX_CG)
2267                                                             {
2268                                                                 put_in_list(bHaveVdW,ngid,md,icg,jgid,
2269                                                                             nsr[jgid],nl_sr[jgid],
2270                                                                             cgs->index,/* cgsatoms, */ bexcl,
2271                                                                             shift,fr,FALSE,TRUE,TRUE);
2272                                                                 nsr[jgid]=0;
2273                                                             }
2274                                                             nl_sr[jgid][nsr[jgid]++]=jjcg;
2275                                                         }
2276                                                         else if (r2 < rm2)
2277                                                         {
2278                                                             if (nlr_ljc[jgid] >= MAX_CG)
2279                                                             {
2280                                                                 do_longrange(cr,top,fr,ngid,md,icg,jgid,
2281                                                                              nlr_ljc[jgid],
2282                                                                              nl_lr_ljc[jgid],bexcl,shift,x,
2283                                                                              box_size,nrnb,
2284                                                                              lambda,dvdlambda,
2285                                                                              grppener,
2286                                                                              TRUE,TRUE,FALSE,
2287                                                                              put_in_list,
2288                                                                              bHaveVdW,
2289                                                                              bDoForces,f);
2290                                                                 nlr_ljc[jgid]=0;
2291                                                             }
2292                                                             nl_lr_ljc[jgid][nlr_ljc[jgid]++]=jjcg;
2293                                                         }
2294                                                         else
2295                                                         {
2296                                                             if (nlr_one[jgid] >= MAX_CG) {
2297                                                                 do_longrange(cr,top,fr,ngid,md,icg,jgid,
2298                                                                              nlr_one[jgid],
2299                                                                              nl_lr_one[jgid],bexcl,shift,x,
2300                                                                              box_size,nrnb,
2301                                                                              lambda,dvdlambda,
2302                                                                              grppener,
2303                                                                              rvdw_lt_rcoul,rcoul_lt_rvdw,FALSE,
2304                                                                              put_in_list,
2305                                                                              bHaveVdW,
2306                                                                              bDoForces,f);
2307                                                                 nlr_one[jgid]=0;
2308                                                             }
2309                                                             nl_lr_one[jgid][nlr_one[jgid]++]=jjcg;
2310                                                         }
2311                                                     }
2312                                                 }
2313                                                 nns++;
2314                                             }
2315                                         }
2316                                     }
2317                                 }
2318                             }
2319                         }
2320                     }
2321                     /* CHECK whether there is anything left in the buffers */
2322                     for(nn=0; (nn<ngid); nn++)
2323                     {
2324                         if (nsr[nn] > 0)
2325                         {
2326                             put_in_list(bHaveVdW,ngid,md,icg,nn,nsr[nn],nl_sr[nn],
2327                                         cgs->index, /* cgsatoms, */ bexcl,
2328                                         shift,fr,FALSE,TRUE,TRUE);
2329                         }
2330
2331                         if (nlr_ljc[nn] > 0)
2332                         {
2333                             do_longrange(cr,top,fr,ngid,md,icg,nn,nlr_ljc[nn],
2334                                          nl_lr_ljc[nn],bexcl,shift,x,box_size,nrnb,
2335                                          lambda,dvdlambda,grppener,TRUE,TRUE,FALSE,
2336                                          put_in_list,bHaveVdW,bDoForces,f);
2337                         }
2338
2339                         if (nlr_one[nn] > 0)
2340                         {
2341                             do_longrange(cr,top,fr,ngid,md,icg,nn,nlr_one[nn],
2342                                          nl_lr_one[nn],bexcl,shift,x,box_size,nrnb,
2343                                          lambda,dvdlambda,grppener,
2344                                          rvdw_lt_rcoul,rcoul_lt_rvdw,FALSE,
2345                                          put_in_list,bHaveVdW,bDoForces,f);
2346                         }
2347                     }
2348                 }
2349             }
2350         }
2351         /* setexcl(nri,i_atoms,&top->atoms.excl,FALSE,bexcl); */
2352         setexcl(cgs->index[icg],cgs->index[icg+1],&top->excls,FALSE,bexcl);
2353     }
2354     /* Perform any left over force calculations */
2355     for (nn=0; (nn<ngid); nn++)
2356     {
2357         if (rm2 > rs2)
2358         {
2359             do_longrange(cr,top,fr,0,md,icg,nn,nlr_ljc[nn],
2360                          nl_lr_ljc[nn],bexcl,shift,x,box_size,nrnb,
2361                          lambda,dvdlambda,grppener,
2362                          TRUE,TRUE,TRUE,put_in_list,bHaveVdW,bDoForces,f);
2363         }
2364         if (rl2 > rm2) {
2365             do_longrange(cr,top,fr,0,md,icg,nn,nlr_one[nn],
2366                          nl_lr_one[nn],bexcl,shift,x,box_size,nrnb,
2367                          lambda,dvdlambda,grppener,
2368                          rvdw_lt_rcoul,rcoul_lt_rvdw,
2369                          TRUE,put_in_list,bHaveVdW,bDoForces,f);
2370         }
2371     }
2372     debug_gmx();
2373
2374     /* Close off short range neighbourlists */
2375     close_neighbor_list(fr,FALSE,-1,-1,bMakeQMMMnblist);
2376
2377     return nns;
2378 }
2379
2380 void ns_realloc_natoms(gmx_ns_t *ns,int natoms)
2381 {
2382     int i;
2383
2384     if (natoms > ns->nra_alloc)
2385     {
2386         ns->nra_alloc = over_alloc_dd(natoms);
2387         srenew(ns->bexcl,ns->nra_alloc);
2388         for(i=0; i<ns->nra_alloc; i++)
2389         {
2390             ns->bexcl[i] = 0;
2391         }
2392     }
2393 }
2394
2395 void init_ns(FILE *fplog,const t_commrec *cr,
2396              gmx_ns_t *ns,t_forcerec *fr,
2397              const gmx_mtop_t *mtop,
2398              matrix box)
2399 {
2400     int  mt,icg,nr_in_cg,maxcg,i,j,jcg,ngid,ncg;
2401     t_block *cgs;
2402     char *ptr;
2403
2404     /* Compute largest charge groups size (# atoms) */
2405     nr_in_cg=1;
2406     for(mt=0; mt<mtop->nmoltype; mt++) {
2407         cgs = &mtop->moltype[mt].cgs;
2408         for (icg=0; (icg < cgs->nr); icg++)
2409         {
2410             nr_in_cg=max(nr_in_cg,(int)(cgs->index[icg+1]-cgs->index[icg]));
2411         }
2412     }
2413
2414     /* Verify whether largest charge group is <= max cg.
2415      * This is determined by the type of the local exclusion type
2416      * Exclusions are stored in bits. (If the type is not large
2417      * enough, enlarge it, unsigned char -> unsigned short -> unsigned long)
2418      */
2419     maxcg = sizeof(t_excl)*8;
2420     if (nr_in_cg > maxcg)
2421     {
2422         gmx_fatal(FARGS,"Max #atoms in a charge group: %d > %d\n",
2423                   nr_in_cg,maxcg);
2424     }
2425
2426     ngid = mtop->groups.grps[egcENER].nr;
2427     snew(ns->bExcludeAlleg,ngid);
2428     for(i=0; i<ngid; i++) {
2429         ns->bExcludeAlleg[i] = TRUE;
2430         for(j=0; j<ngid; j++)
2431         {
2432             if (!(fr->egp_flags[i*ngid+j] & EGP_EXCL))
2433             {
2434                 ns->bExcludeAlleg[i] = FALSE;
2435             }
2436         }
2437     }
2438
2439     if (fr->bGrid) {
2440         /* Grid search */
2441         ns->grid = init_grid(fplog,fr);
2442         init_nsgrid_lists(fr,ngid,ns);
2443     }
2444     else
2445     {
2446         /* Simple search */
2447         snew(ns->ns_buf,ngid);
2448         for(i=0; (i<ngid); i++)
2449         {
2450             snew(ns->ns_buf[i],SHIFTS);
2451         }
2452         ncg = ncg_mtop(mtop);
2453         snew(ns->simple_aaj,2*ncg);
2454         for(jcg=0; (jcg<ncg); jcg++)
2455         {
2456             ns->simple_aaj[jcg]     = jcg;
2457             ns->simple_aaj[jcg+ncg] = jcg;
2458         }
2459     }
2460
2461     /* Create array that determines whether or not atoms have VdW */
2462     snew(ns->bHaveVdW,fr->ntype);
2463     for(i=0; (i<fr->ntype); i++)
2464     {
2465         for(j=0; (j<fr->ntype); j++)
2466         {
2467             ns->bHaveVdW[i] = (ns->bHaveVdW[i] ||
2468                                (fr->bBHAM ?
2469                                 ((BHAMA(fr->nbfp,fr->ntype,i,j) != 0) ||
2470                                  (BHAMB(fr->nbfp,fr->ntype,i,j) != 0) ||
2471                                  (BHAMC(fr->nbfp,fr->ntype,i,j) != 0)) :
2472                                 ((C6(fr->nbfp,fr->ntype,i,j) != 0) ||
2473                                  (C12(fr->nbfp,fr->ntype,i,j) != 0))));
2474         }
2475     }
2476     if (debug)
2477         pr_bvec(debug,0,"bHaveVdW",ns->bHaveVdW,fr->ntype,TRUE);
2478
2479     ns->nra_alloc = 0;
2480     ns->bexcl = NULL;
2481     if (!DOMAINDECOMP(cr))
2482     {
2483         /* This could be reduced with particle decomposition */
2484         ns_realloc_natoms(ns,mtop->natoms);
2485     }
2486
2487     ns->nblist_initialized=FALSE;
2488
2489     /* nbr list debug dump */
2490     {
2491         char *ptr=getenv("GMX_DUMP_NL");
2492         if (ptr)
2493         {
2494             ns->dump_nl=strtol(ptr,NULL,10);
2495             if (fplog)
2496             {
2497                 fprintf(fplog, "GMX_DUMP_NL = %d", ns->dump_nl);
2498             }
2499         }
2500         else
2501         {
2502             ns->dump_nl=0;
2503         }
2504     }
2505 }
2506
2507
2508 int search_neighbours(FILE *log,t_forcerec *fr,
2509                       rvec x[],matrix box,
2510                       gmx_localtop_t *top,
2511                       gmx_groups_t *groups,
2512                       t_commrec *cr,
2513                       t_nrnb *nrnb,t_mdatoms *md,
2514                       real lambda,real *dvdlambda,
2515                       gmx_grppairener_t *grppener,
2516                       gmx_bool bFillGrid,
2517                       gmx_bool bDoLongRange,
2518                       gmx_bool bDoForces,rvec *f)
2519 {
2520     t_block  *cgs=&(top->cgs);
2521     rvec     box_size,grid_x0,grid_x1;
2522     int      i,j,m,ngid;
2523     real     min_size,grid_dens;
2524     int      nsearch;
2525     gmx_bool     bGrid;
2526     char     *ptr;
2527     gmx_bool     *i_egp_flags;
2528     int      cg_start,cg_end,start,end;
2529     gmx_ns_t *ns;
2530     t_grid   *grid;
2531     gmx_domdec_zones_t *dd_zones;
2532     put_in_list_t *put_in_list;
2533
2534     ns = &fr->ns;
2535
2536     /* Set some local variables */
2537     bGrid = fr->bGrid;
2538     ngid = groups->grps[egcENER].nr;
2539
2540     for(m=0; (m<DIM); m++)
2541     {
2542         box_size[m] = box[m][m];
2543     }
2544
2545     if (fr->ePBC != epbcNONE)
2546     {
2547         if (sqr(fr->rlistlong) >= max_cutoff2(fr->ePBC,box))
2548         {
2549             gmx_fatal(FARGS,"One of the box vectors has become shorter than twice the cut-off length or box_yy-|box_zy| or box_zz has become smaller than the cut-off.");
2550         }
2551         if (!bGrid)
2552         {
2553             min_size = min(box_size[XX],min(box_size[YY],box_size[ZZ]));
2554             if (2*fr->rlistlong >= min_size)
2555                 gmx_fatal(FARGS,"One of the box diagonal elements has become smaller than twice the cut-off length.");
2556         }
2557     }
2558
2559     if (DOMAINDECOMP(cr))
2560     {
2561         ns_realloc_natoms(ns,cgs->index[cgs->nr]);
2562     }
2563     debug_gmx();
2564
2565     /* Reset the neighbourlists */
2566     reset_neighbor_list(fr,FALSE,-1,-1);
2567
2568     if (bGrid && bFillGrid)
2569     {
2570
2571         grid = ns->grid;
2572         if (DOMAINDECOMP(cr))
2573         {
2574             dd_zones = domdec_zones(cr->dd);
2575         }
2576         else
2577         {
2578             dd_zones = NULL;
2579
2580             get_nsgrid_boundaries(grid,NULL,box,NULL,NULL,NULL,
2581                                   cgs->nr,fr->cg_cm,grid_x0,grid_x1,&grid_dens);
2582
2583             grid_first(log,grid,NULL,NULL,fr->ePBC,box,grid_x0,grid_x1,
2584                        fr->rlistlong,grid_dens);
2585         }
2586         debug_gmx();
2587
2588         /* Don't know why this all is... (DvdS 3/99) */
2589 #ifndef SEGV
2590         start = 0;
2591         end   = cgs->nr;
2592 #else
2593         start = fr->cg0;
2594         end   = (cgs->nr+1)/2;
2595 #endif
2596
2597         if (DOMAINDECOMP(cr))
2598         {
2599             end = cgs->nr;
2600             fill_grid(log,dd_zones,grid,end,-1,end,fr->cg_cm);
2601             grid->icg0 = 0;
2602             grid->icg1 = dd_zones->izone[dd_zones->nizone-1].cg1;
2603         }
2604         else
2605         {
2606             fill_grid(log,NULL,grid,cgs->nr,fr->cg0,fr->hcg,fr->cg_cm);
2607             grid->icg0 = fr->cg0;
2608             grid->icg1 = fr->hcg;
2609             debug_gmx();
2610
2611             if (PARTDECOMP(cr))
2612                 mv_grid(cr,grid);
2613             debug_gmx();
2614         }
2615
2616         calc_elemnr(log,grid,start,end,cgs->nr);
2617         calc_ptrs(grid);
2618         grid_last(log,grid,start,end,cgs->nr);
2619
2620         if (gmx_debug_at)
2621         {
2622             check_grid(debug,grid);
2623             print_grid(debug,grid);
2624         }
2625     }
2626     else if (fr->n_tpi)
2627     {
2628         /* Set the grid cell index for the test particle only.
2629          * The cell to cg index is not corrected, but that does not matter.
2630          */
2631         fill_grid(log,NULL,ns->grid,fr->hcg,fr->hcg-1,fr->hcg,fr->cg_cm);
2632     }
2633     debug_gmx();
2634
2635     if (!fr->ns.bCGlist)
2636     {
2637         put_in_list = put_in_list_at;
2638     }
2639     else
2640     {
2641         put_in_list = put_in_list_cg;
2642     }
2643
2644     /* Do the core! */
2645     if (bGrid)
2646     {
2647         grid = ns->grid;
2648         nsearch = nsgrid_core(log,cr,fr,box,box_size,ngid,top,
2649                               grid,x,ns->bexcl,ns->bExcludeAlleg,
2650                               nrnb,md,lambda,dvdlambda,grppener,
2651                               put_in_list,ns->bHaveVdW,
2652                               bDoLongRange,bDoForces,f,
2653                               FALSE);
2654
2655         /* neighbour searching withouth QMMM! QM atoms have zero charge in
2656          * the classical calculation. The charge-charge interaction
2657          * between QM and MM atoms is handled in the QMMM core calculation
2658          * (see QMMM.c). The VDW however, we'd like to compute classically
2659          * and the QM MM atom pairs have just been put in the
2660          * corresponding neighbourlists. in case of QMMM we still need to
2661          * fill a special QMMM neighbourlist that contains all neighbours
2662          * of the QM atoms. If bQMMM is true, this list will now be made:
2663          */
2664         if (fr->bQMMM && fr->qr->QMMMscheme!=eQMMMschemeoniom)
2665         {
2666             nsearch += nsgrid_core(log,cr,fr,box,box_size,ngid,top,
2667                                    grid,x,ns->bexcl,ns->bExcludeAlleg,
2668                                    nrnb,md,lambda,dvdlambda,grppener,
2669                                    put_in_list_qmmm,ns->bHaveVdW,
2670                                    bDoLongRange,bDoForces,f,
2671                                    TRUE);
2672         }
2673     }
2674     else
2675     {
2676         nsearch = ns_simple_core(fr,top,md,box,box_size,
2677                                  ns->bexcl,ns->simple_aaj,
2678                                  ngid,ns->ns_buf,put_in_list,ns->bHaveVdW);
2679     }
2680     debug_gmx();
2681
2682 #ifdef DEBUG
2683     pr_nsblock(log);
2684 #endif
2685
2686     inc_nrnb(nrnb,eNR_NS,nsearch);
2687     /* inc_nrnb(nrnb,eNR_LR,fr->nlr); */
2688
2689     return nsearch;
2690 }
2691
2692 int natoms_beyond_ns_buffer(t_inputrec *ir,t_forcerec *fr,t_block *cgs,
2693                             matrix scale_tot,rvec *x)
2694 {
2695     int  cg0,cg1,cg,a0,a1,a,i,j;
2696     real rint,hbuf2,scale;
2697     rvec *cg_cm,cgsc;
2698     gmx_bool bIsotropic;
2699     int  nBeyond;
2700
2701     nBeyond = 0;
2702
2703     rint = max(ir->rcoulomb,ir->rvdw);
2704     if (ir->rlist < rint)
2705     {
2706         gmx_fatal(FARGS,"The neighbor search buffer has negative size: %f nm",
2707                   ir->rlist - rint);
2708     }
2709     cg_cm = fr->cg_cm;
2710
2711     cg0 = fr->cg0;
2712     cg1 = fr->hcg;
2713
2714     if (!EI_DYNAMICS(ir->eI) || !DYNAMIC_BOX(*ir))
2715     {
2716         hbuf2 = sqr(0.5*(ir->rlist - rint));
2717         for(cg=cg0; cg<cg1; cg++)
2718         {
2719             a0 = cgs->index[cg];
2720             a1 = cgs->index[cg+1];
2721             for(a=a0; a<a1; a++)
2722             {
2723                 if (distance2(cg_cm[cg],x[a]) > hbuf2)
2724                 {
2725                     nBeyond++;
2726                 }
2727             }
2728         }
2729     }
2730     else
2731     {
2732         bIsotropic = TRUE;
2733         scale = scale_tot[0][0];
2734         for(i=1; i<DIM; i++)
2735         {
2736             /* With anisotropic scaling, the original spherical ns volumes become
2737              * ellipsoids. To avoid costly transformations we use the minimum
2738              * eigenvalue of the scaling matrix for determining the buffer size.
2739              * Since the lower half is 0, the eigenvalues are the diagonal elements.
2740              */
2741             scale = min(scale,scale_tot[i][i]);
2742             if (scale_tot[i][i] != scale_tot[i-1][i-1])
2743             {
2744                 bIsotropic = FALSE;
2745             }
2746             for(j=0; j<i; j++)
2747             {
2748                 if (scale_tot[i][j] != 0)
2749                 {
2750                     bIsotropic = FALSE;
2751                 }
2752             }
2753         }
2754         hbuf2 = sqr(0.5*(scale*ir->rlist - rint));
2755         if (bIsotropic)
2756         {
2757             for(cg=cg0; cg<cg1; cg++)
2758             {
2759                 svmul(scale,cg_cm[cg],cgsc);
2760                 a0 = cgs->index[cg];
2761                 a1 = cgs->index[cg+1];
2762                 for(a=a0; a<a1; a++)
2763                 {
2764                     if (distance2(cgsc,x[a]) > hbuf2)
2765                     {
2766                         nBeyond++;
2767                     }
2768                 }
2769             }
2770         }
2771         else
2772         {
2773             /* Anistropic scaling */
2774             for(cg=cg0; cg<cg1; cg++)
2775             {
2776                 /* Since scale_tot contains the transpose of the scaling matrix,
2777                  * we need to multiply with the transpose.
2778                  */
2779                 tmvmul_ur0(scale_tot,cg_cm[cg],cgsc);
2780                 a0 = cgs->index[cg];
2781                 a1 = cgs->index[cg+1];
2782                 for(a=a0; a<a1; a++)
2783                 {
2784                     if (distance2(cgsc,x[a]) > hbuf2)
2785                     {
2786                         nBeyond++;
2787                     }
2788                 }
2789             }
2790         }
2791     }
2792
2793     return nBeyond;
2794 }