?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Move main.*, splitter.*, gmx_omp_nthreads.* to mdlib
[gromacs.git] / src / gromacs / mdlib / ns.cpp
blob10ba54a3592b25f4a91a30b35f10008ae1d97fb6
1 /*
2 * This file is part of the GROMACS molecular simulation package.
3 *
4 * Copyright (c) 1991-2000, University of Groningen, The Netherlands.
5 * Copyright (c) 2001-2004, The GROMACS development team.
6 * Copyright (c) 2013,2014,2015, by the GROMACS development team, led by
7 * Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl,
8 * and including many others, as listed in the AUTHORS file in the
9 * top-level source directory and at http://www.gromacs.org.
10 *
11 * GROMACS is free software; you can redistribute it and/or
12 * modify it under the terms of the GNU Lesser General Public License
13 * as published by the Free Software Foundation; either version 2.1
14 * of the License, or (at your option) any later version.
15 *
16 * GROMACS is distributed in the hope that it will be useful,
17 * but WITHOUT ANY WARRANTY; without even the implied warranty of
18 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
19 * Lesser General Public License for more details.
20 *
21 * You should have received a copy of the GNU Lesser General Public
22 * License along with GROMACS; if not, see
23 * http://www.gnu.org/licenses, or write to the Free Software Foundation,
24 * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
25 *
26 * If you want to redistribute modifications to GROMACS, please
27 * consider that scientific software is very special. Version
28 * control is crucial - bugs must be traceable. We will be happy to
29 * consider code for inclusion in the official distribution, but
30 * derived work must not be called official GROMACS. Details are found
31 * in the README & COPYING files - if they are missing, get the
32 * official version at http://www.gromacs.org.
33 *
34 * To help us fund GROMACS development, we humbly ask that you cite
35 * the research papers on the package. Check out http://www.gromacs.org.
36 */
37 #include "gmxpre.h"
38
39 #include "ns.h"
40
41 #include <math.h>
42 #include <stdlib.h>
43 #include <string.h>
44
45 #include <cmath>
46
47 #include <algorithm>
48
49 #include "gromacs/domdec/domdec.h"
50 #include "gromacs/domdec/domdec_struct.h"
51 #include "gromacs/gmxlib/network.h"
52 #include "gromacs/gmxlib/nrnb.h"
53 #include "gromacs/gmxlib/nonbonded/nonbonded.h"
54 #include "gromacs/math/utilities.h"
55 #include "gromacs/math/vec.h"
56 #include "gromacs/math/vecdump.h"
57 #include "gromacs/mdlib/force.h"
58 #include "gromacs/mdlib/nsgrid.h"
59 #include "gromacs/mdlib/qmmm.h"
60 #include "gromacs/mdtypes/commrec.h"
61 #include "gromacs/mdtypes/group.h"
62 #include "gromacs/mdtypes/md_enums.h"
63 #include "gromacs/pbcutil/ishift.h"
64 #include "gromacs/pbcutil/pbc.h"
65 #include "gromacs/topology/mtop_util.h"
66 #include "gromacs/utility/fatalerror.h"
67 #include "gromacs/utility/smalloc.h"
68
69 /*
70 * E X C L U S I O N H A N D L I N G
71 */
72
73 #ifdef DEBUG
74 static void SETEXCL_(t_excl e[], int i, int j)
75 {
76 e[j] = e[j] | (1<<i);
77 }
78 static void RMEXCL_(t_excl e[], int i, int j)
79 {
80 e[j] = e[j] & ~(1<<i);
81 }
82 static gmx_bool ISEXCL_(t_excl e[], int i, int j)
83 {
84 return (gmx_bool)(e[j] & (1<<i));
85 }
86 static gmx_bool NOTEXCL_(t_excl e[], int i, int j)
87 {
88 return !(ISEXCL(e, i, j));
89 }
90 #else
91 #define SETEXCL(e, i, j) (e)[((int) (j))] |= (1<<((int) (i)))
92 #define RMEXCL(e, i, j) (e)[((int) (j))] &= (~(1<<((int) (i))))
93 #define ISEXCL(e, i, j) (gmx_bool) ((e)[((int) (j))] & (1<<((int) (i))))
94 #define NOTEXCL(e, i, j) !(ISEXCL(e, i, j))
95 #endif
96
97 static int
98 round_up_to_simd_width(int length, int simd_width)
99 {
100 int offset;
101
102 offset = (simd_width > 0) ? length % simd_width : 0;
103
104 return (offset == 0) ? length : length-offset+simd_width;
105 }
106 /************************************************
107 *
108 * U T I L I T I E S F O R N S
109 *
110 ************************************************/
111
112 void reallocate_nblist(t_nblist *nl)
113 {
114 if (gmx_debug_at)
115 {
116 fprintf(debug, "reallocating neigborlist (ielec=%d, ivdw=%d, igeometry=%d, type=%d), maxnri=%d\n",
117 nl->ielec, nl->ivdw, nl->igeometry, nl->type, nl->maxnri);
118 }
119 srenew(nl->iinr, nl->maxnri);
120 if (nl->igeometry == GMX_NBLIST_GEOMETRY_CG_CG)
121 {
122 srenew(nl->iinr_end, nl->maxnri);
123 }
124 srenew(nl->gid, nl->maxnri);
125 srenew(nl->shift, nl->maxnri);
126 srenew(nl->jindex, nl->maxnri+1);
127 }
128
129
130 static void init_nblist(FILE *log, t_nblist *nl_sr,
131 int maxsr,
132 int ivdw, int ivdwmod,
133 int ielec, int ielecmod,
134 int igeometry, int type,
135 gmx_bool bElecAndVdwSwitchDiffers)
136 {
137 t_nblist *nl;
138 int homenr;
139
140 {
141 nl = nl_sr;
142 homenr = maxsr;
143
144 if (nl == NULL)
145 {
146 return;
147 }
148
149
150 /* Set coul/vdw in neighborlist, and for the normal loops we determine
151 * an index of which one to call.
152 */
153 nl->ivdw = ivdw;
154 nl->ivdwmod = ivdwmod;
155 nl->ielec = ielec;
156 nl->ielecmod = ielecmod;
157 nl->type = type;
158 nl->igeometry = igeometry;
159
160 if (nl->type == GMX_NBLIST_INTERACTION_FREE_ENERGY)
161 {
162 nl->igeometry = GMX_NBLIST_GEOMETRY_PARTICLE_PARTICLE;
163 }
164
165 /* This will also set the simd_padding_width field */
166 gmx_nonbonded_set_kernel_pointers(log, nl, bElecAndVdwSwitchDiffers);
167
168 /* maxnri is influenced by the number of shifts (maximum is 8)
169 * and the number of energy groups.
170 * If it is not enough, nl memory will be reallocated during the run.
171 * 4 seems to be a reasonable factor, which only causes reallocation
172 * during runs with tiny and many energygroups.
173 */
174 nl->maxnri = homenr*4;
175 nl->maxnrj = 0;
176 nl->nri = -1;
177 nl->nrj = 0;
178 nl->iinr = NULL;
179 nl->gid = NULL;
180 nl->shift = NULL;
181 nl->jindex = NULL;
182 nl->jjnr = NULL;
183 nl->excl_fep = NULL;
184 reallocate_nblist(nl);
185 nl->jindex[0] = 0;
186
187 if (debug)
188 {
189 fprintf(debug, "Initiating neighbourlist (ielec=%d, ivdw=%d, type=%d) for %s interactions,\nwith %d SR atoms.\n",
190 nl->ielec, nl->ivdw, nl->type, gmx_nblist_geometry_names[nl->igeometry], maxsr);
191 }
192 }
193 }
194
195 void init_neighbor_list(FILE *log, t_forcerec *fr, int homenr)
196 {
197 int maxsr, maxsr_wat;
198 int ielec, ivdw, ielecmod, ivdwmod, type;
199 int igeometry_def, igeometry_w, igeometry_ww;
200 int i;
201 gmx_bool bElecAndVdwSwitchDiffers;
202 t_nblists *nbl;
203
204 /* maxsr = homenr-fr->nWatMol*3; */
205 maxsr = homenr;
206
207 if (maxsr < 0)
208 {
209 gmx_fatal(FARGS, "%s, %d: Negative number of short range atoms.\n"
210 "Call your GROMACS dealer for assistance.", __FILE__, __LINE__);
211 }
212 /* This is just for initial allocation, so we do not reallocate
213 * all the nlist arrays many times in a row.
214 * The numbers seem very accurate, but they are uncritical.
215 */
216 maxsr_wat = std::min(fr->nWatMol, (homenr+2)/3);
217
218 /* Determine the values for ielec/ivdw. */
219 ielec = fr->nbkernel_elec_interaction;
220 ivdw = fr->nbkernel_vdw_interaction;
221 ielecmod = fr->nbkernel_elec_modifier;
222 ivdwmod = fr->nbkernel_vdw_modifier;
223 type = GMX_NBLIST_INTERACTION_STANDARD;
224 bElecAndVdwSwitchDiffers = ( (fr->rcoulomb_switch != fr->rvdw_switch) || (fr->rcoulomb != fr->rvdw));
225
226 fr->ns->bCGlist = (getenv("GMX_NBLISTCG") != 0);
227 if (!fr->ns->bCGlist)
228 {
229 igeometry_def = GMX_NBLIST_GEOMETRY_PARTICLE_PARTICLE;
230 }
231 else
232 {
233 igeometry_def = GMX_NBLIST_GEOMETRY_CG_CG;
234 if (log != NULL)
235 {
236 fprintf(log, "\nUsing charge-group - charge-group neighbor lists and kernels\n\n");
237 }
238 }
239
240 if (fr->solvent_opt == esolTIP4P)
241 {
242 igeometry_w = GMX_NBLIST_GEOMETRY_WATER4_PARTICLE;
243 igeometry_ww = GMX_NBLIST_GEOMETRY_WATER4_WATER4;
244 }
245 else
246 {
247 igeometry_w = GMX_NBLIST_GEOMETRY_WATER3_PARTICLE;
248 igeometry_ww = GMX_NBLIST_GEOMETRY_WATER3_WATER3;
249 }
250
251 for (i = 0; i < fr->nnblists; i++)
252 {
253 nbl = &(fr->nblists[i]);
254
255 init_nblist(log, &nbl->nlist_sr[eNL_VDWQQ],
256 maxsr, ivdw, ivdwmod, ielec, ielecmod, igeometry_def, type, bElecAndVdwSwitchDiffers);
257 init_nblist(log, &nbl->nlist_sr[eNL_VDW],
258 maxsr, ivdw, ivdwmod, GMX_NBKERNEL_ELEC_NONE, eintmodNONE, igeometry_def, type, bElecAndVdwSwitchDiffers);
259 init_nblist(log, &nbl->nlist_sr[eNL_QQ],
260 maxsr, GMX_NBKERNEL_VDW_NONE, eintmodNONE, ielec, ielecmod, igeometry_def, type, bElecAndVdwSwitchDiffers);
261 init_nblist(log, &nbl->nlist_sr[eNL_VDWQQ_WATER],
262 maxsr_wat, ivdw, ivdwmod, ielec, ielecmod, igeometry_w, type, bElecAndVdwSwitchDiffers);
263 init_nblist(log, &nbl->nlist_sr[eNL_QQ_WATER],
264 maxsr_wat, GMX_NBKERNEL_VDW_NONE, eintmodNONE, ielec, ielecmod, igeometry_w, type, bElecAndVdwSwitchDiffers);
265 init_nblist(log, &nbl->nlist_sr[eNL_VDWQQ_WATERWATER],
266 maxsr_wat, ivdw, ivdwmod, ielec, ielecmod, igeometry_ww, type, bElecAndVdwSwitchDiffers);
267 init_nblist(log, &nbl->nlist_sr[eNL_QQ_WATERWATER],
268 maxsr_wat, GMX_NBKERNEL_VDW_NONE, eintmodNONE, ielec, ielecmod, igeometry_ww, type, bElecAndVdwSwitchDiffers);
269
270 /* Did we get the solvent loops so we can use optimized water kernels? */
271 if (nbl->nlist_sr[eNL_VDWQQ_WATER].kernelptr_vf == NULL
272 || nbl->nlist_sr[eNL_QQ_WATER].kernelptr_vf == NULL
273 || nbl->nlist_sr[eNL_VDWQQ_WATERWATER].kernelptr_vf == NULL
274 || nbl->nlist_sr[eNL_QQ_WATERWATER].kernelptr_vf == NULL)
275 {
276 fr->solvent_opt = esolNO;
277 if (log != NULL)
278 {
279 fprintf(log, "Note: The available nonbonded kernels do not support water optimization - disabling.\n");
280 }
281 }
282
283 if (fr->efep != efepNO)
284 {
285 init_nblist(log, &nbl->nlist_sr[eNL_VDWQQ_FREE],
286 maxsr, ivdw, ivdwmod, ielec, ielecmod, GMX_NBLIST_GEOMETRY_PARTICLE_PARTICLE, GMX_NBLIST_INTERACTION_FREE_ENERGY, bElecAndVdwSwitchDiffers);
287 init_nblist(log, &nbl->nlist_sr[eNL_VDW_FREE],
288 maxsr, ivdw, ivdwmod, GMX_NBKERNEL_ELEC_NONE, eintmodNONE, GMX_NBLIST_GEOMETRY_PARTICLE_PARTICLE, GMX_NBLIST_INTERACTION_FREE_ENERGY, bElecAndVdwSwitchDiffers);
289 init_nblist(log, &nbl->nlist_sr[eNL_QQ_FREE],
290 maxsr, GMX_NBKERNEL_VDW_NONE, eintmodNONE, ielec, ielecmod, GMX_NBLIST_GEOMETRY_PARTICLE_PARTICLE, GMX_NBLIST_INTERACTION_FREE_ENERGY, bElecAndVdwSwitchDiffers);
291 }
292 }
293 /* QMMM MM list */
294 if (fr->bQMMM && fr->qr->QMMMscheme != eQMMMschemeoniom)
295 {
296 if (NULL == fr->QMMMlist)
297 {
298 snew(fr->QMMMlist, 1);
299 }
300 init_nblist(log, fr->QMMMlist,
301 maxsr, 0, 0, ielec, ielecmod, GMX_NBLIST_GEOMETRY_PARTICLE_PARTICLE, GMX_NBLIST_INTERACTION_STANDARD, bElecAndVdwSwitchDiffers);
302 }
303
304 if (log != NULL)
305 {
306 fprintf(log, "\n");
307 }
308
309 fr->ns->nblist_initialized = TRUE;
310 }
311
312 static void reset_nblist(t_nblist *nl)
313 {
314 nl->nri = -1;
315 nl->nrj = 0;
316 if (nl->jindex)
317 {
318 nl->jindex[0] = 0;
319 }
320 }
321
322 static void reset_neighbor_lists(t_forcerec *fr)
323 {
324 int n, i;
325
326 if (fr->bQMMM)
327 {
328 /* only reset the short-range nblist */
329 reset_nblist(fr->QMMMlist);
330 }
331
332 for (n = 0; n < fr->nnblists; n++)
333 {
334 for (i = 0; i < eNL_NR; i++)
335 {
336 reset_nblist( &(fr->nblists[n].nlist_sr[i]) );
337 }
338 }
339 }
340
341
342
343
344 static gmx_inline void new_i_nblist(t_nblist *nlist, int i_atom, int shift, int gid)
345 {
346 int nri = nlist->nri;
347
348 /* Check whether we have to increase the i counter */
349 if ((nri == -1) ||
350 (nlist->iinr[nri] != i_atom) ||
351 (nlist->shift[nri] != shift) ||
352 (nlist->gid[nri] != gid))
353 {
354 /* This is something else. Now see if any entries have
355 * been added in the list of the previous atom.
356 */
357 if ((nri == -1) ||
358 ((nlist->jindex[nri+1] > nlist->jindex[nri]) &&
359 (nlist->gid[nri] != -1)))
360 {
361 /* If so increase the counter */
362 nlist->nri++;
363 nri++;
364 if (nlist->nri >= nlist->maxnri)
365 {
366 nlist->maxnri += over_alloc_large(nlist->nri);
367 reallocate_nblist(nlist);
368 }
369 }
370 /* Set the number of neighbours and the atom number */
371 nlist->jindex[nri+1] = nlist->jindex[nri];
372 nlist->iinr[nri] = i_atom;
373 nlist->gid[nri] = gid;
374 nlist->shift[nri] = shift;
375 }
376 else
377 {
378 /* Adding to previous list. First remove possible previous padding */
379 if (nlist->simd_padding_width > 1)
380 {
381 while (nlist->nrj > 0 && nlist->jjnr[nlist->nrj-1] < 0)
382 {
383 nlist->nrj--;
384 }
385 }
386 }
387 }
388
389 static gmx_inline void close_i_nblist(t_nblist *nlist)
390 {
391 int nri = nlist->nri;
392 int len;
393
394 if (nri >= 0)
395 {
396 /* Add elements up to padding. Since we allocate memory in units
397 * of the simd_padding width, we do not have to check for possible
398 * list reallocation here.
399 */
400 while ((nlist->nrj % nlist->simd_padding_width) != 0)
401 {
402 /* Use -4 here, so we can write forces for 4 atoms before real data */
403 nlist->jjnr[nlist->nrj++] = -4;
404 }
405 nlist->jindex[nri+1] = nlist->nrj;
406
407 len = nlist->nrj - nlist->jindex[nri];
408 /* If there are no j-particles we have to reduce the
409 * number of i-particles again, to prevent errors in the
410 * kernel functions.
411 */
412 if ((len == 0) && (nlist->nri > 0))
413 {
414 nlist->nri--;
415 }
416 }
417 }
418
419 static gmx_inline void close_nblist(t_nblist *nlist)
420 {
421 /* Only close this nblist when it has been initialized.
422 * Avoid the creation of i-lists with no j-particles.
423 */
424 if (nlist->nrj == 0)
425 {
426 /* Some assembly kernels do not support empty lists,
427 * make sure here that we don't generate any empty lists.
428 * With the current ns code this branch is taken in two cases:
429 * No i-particles at all: nri=-1 here
430 * There are i-particles, but no j-particles; nri=0 here
431 */
432 nlist->nri = 0;
433 }
434 else
435 {
436 /* Close list number nri by incrementing the count */
437 nlist->nri++;
438 }
439 }
440
441 static gmx_inline void close_neighbor_lists(t_forcerec *fr, gmx_bool bMakeQMMMnblist)
442 {
443 int n, i;
444
445 if (bMakeQMMMnblist)
446 {
447 close_nblist(fr->QMMMlist);
448 }
449
450 for (n = 0; n < fr->nnblists; n++)
451 {
452 for (i = 0; (i < eNL_NR); i++)
453 {
454 close_nblist(&(fr->nblists[n].nlist_sr[i]));
455 }
456 }
457 }
458
459
460 static gmx_inline void add_j_to_nblist(t_nblist *nlist, int j_atom)
461 {
462 int nrj = nlist->nrj;
463
464 if (nlist->nrj >= nlist->maxnrj)
465 {
466 nlist->maxnrj = round_up_to_simd_width(over_alloc_small(nlist->nrj + 1), nlist->simd_padding_width);
467
468 if (gmx_debug_at)
469 {
470 fprintf(debug, "Increasing SR nblist (ielec=%d,ivdw=%d,type=%d,igeometry=%d) j size to %d\n",
471 nlist->ielec, nlist->ivdw, nlist->type, nlist->igeometry, nlist->maxnrj);
472 }
473
474 srenew(nlist->jjnr, nlist->maxnrj);
475 }
476
477 nlist->jjnr[nrj] = j_atom;
478 nlist->nrj++;
479 }
480
481 static gmx_inline void add_j_to_nblist_cg(t_nblist *nlist,
482 int j_start, int j_end,
483 t_excl *bexcl, gmx_bool i_is_j)
484 {
485 int nrj = nlist->nrj;
486 int j;
487
488 if (nlist->nrj >= nlist->maxnrj)
489 {
490 nlist->maxnrj = over_alloc_small(nlist->nrj + 1);
491 if (gmx_debug_at)
492 {
493 fprintf(debug, "Increasing SR nblist (ielec=%d,ivdw=%d,type=%d,igeometry=%d) j size to %d\n",
494 nlist->ielec, nlist->ivdw, nlist->type, nlist->igeometry, nlist->maxnrj);
495 }
496
497 srenew(nlist->jjnr, nlist->maxnrj);
498 srenew(nlist->jjnr_end, nlist->maxnrj);
499 srenew(nlist->excl, nlist->maxnrj*MAX_CGCGSIZE);
500 }
501
502 nlist->jjnr[nrj] = j_start;
503 nlist->jjnr_end[nrj] = j_end;
504
505 if (j_end - j_start > MAX_CGCGSIZE)
506 {
507 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);
508 }
509
510 /* Set the exclusions */
511 for (j = j_start; j < j_end; j++)
512 {
513 nlist->excl[nrj*MAX_CGCGSIZE + j - j_start] = bexcl[j];
514 }
515 if (i_is_j)
516 {
517 /* Avoid double counting of intra-cg interactions */
518 for (j = 1; j < j_end-j_start; j++)
519 {
520 nlist->excl[nrj*MAX_CGCGSIZE + j] |= (1<<j) - 1;
521 }
522 }
523
524 nlist->nrj++;
525 }
526
527 typedef void
528 put_in_list_t (gmx_bool bHaveVdW[],
529 int ngid,
530 t_mdatoms * md,
531 int icg,
532 int jgid,
533 int nj,
534 int jjcg[],
535 int index[],
536 t_excl bExcl[],
537 int shift,
538 t_forcerec * fr,
539 gmx_bool bDoVdW,
540 gmx_bool bDoCoul,
541 int solvent_opt);
542
543 static void
544 put_in_list_at(gmx_bool bHaveVdW[],
545 int ngid,
546 t_mdatoms * md,
547 int icg,
548 int jgid,
549 int nj,
550 int jjcg[],
551 int index[],
552 t_excl bExcl[],
553 int shift,
554 t_forcerec * fr,
555 gmx_bool bDoVdW,
556 gmx_bool bDoCoul,
557 int solvent_opt)
558 {
559 /* The a[] index has been removed,
560 * to put it back in i_atom should be a[i0] and jj should be a[jj].
561 */
562 t_nblist * vdwc;
563 t_nblist * vdw;
564 t_nblist * coul;
565 t_nblist * vdwc_free = NULL;
566 t_nblist * vdw_free = NULL;
567 t_nblist * coul_free = NULL;
568 t_nblist * vdwc_ww = NULL;
569 t_nblist * coul_ww = NULL;
570
571 int i, j, jcg, igid, gid, nbl_ind;
572 int jj, jj0, jj1, i_atom;
573 int i0, nicg;
574
575 int *cginfo;
576 int *type, *typeB;
577 real *charge, *chargeB;
578 real qi, qiB;
579 gmx_bool bFreeEnergy, bFree, bFreeJ, bNotEx, *bPert;
580 gmx_bool bDoVdW_i, bDoCoul_i, bDoCoul_i_sol;
581 int iwater, jwater;
582 t_nblist *nlist;
583
584 /* Copy some pointers */
585 cginfo = fr->cginfo;
586 charge = md->chargeA;
587 chargeB = md->chargeB;
588 type = md->typeA;
589 typeB = md->typeB;
590 bPert = md->bPerturbed;
591
592 /* Get atom range */
593 i0 = index[icg];
594 nicg = index[icg+1]-i0;
595
596 /* Get the i charge group info */
597 igid = GET_CGINFO_GID(cginfo[icg]);
598
599 iwater = (solvent_opt != esolNO) ? GET_CGINFO_SOLOPT(cginfo[icg]) : esolNO;
600
601 bFreeEnergy = FALSE;
602 if (md->nPerturbed)
603 {
604 /* Check if any of the particles involved are perturbed.
605 * If not we can do the cheaper normal put_in_list
606 * and use more solvent optimization.
607 */
608 for (i = 0; i < nicg; i++)
609 {
610 bFreeEnergy |= bPert[i0+i];
611 }
612 /* Loop over the j charge groups */
613 for (j = 0; (j < nj && !bFreeEnergy); j++)
614 {
615 jcg = jjcg[j];
616 jj0 = index[jcg];
617 jj1 = index[jcg+1];
618 /* Finally loop over the atoms in the j-charge group */
619 for (jj = jj0; jj < jj1; jj++)
620 {
621 bFreeEnergy |= bPert[jj];
622 }
623 }
624 }
625
626 /* Unpack pointers to neighbourlist structs */
627 if (fr->nnblists == 1)
628 {
629 nbl_ind = 0;
630 }
631 else
632 {
633 nbl_ind = fr->gid2nblists[GID(igid, jgid, ngid)];
634 }
635 nlist = fr->nblists[nbl_ind].nlist_sr;
636
637 if (iwater != esolNO)
638 {
639 vdwc = &nlist[eNL_VDWQQ_WATER];
640 vdw = &nlist[eNL_VDW];
641 coul = &nlist[eNL_QQ_WATER];
642 vdwc_ww = &nlist[eNL_VDWQQ_WATERWATER];
643 coul_ww = &nlist[eNL_QQ_WATERWATER];
644 }
645 else
646 {
647 vdwc = &nlist[eNL_VDWQQ];
648 vdw = &nlist[eNL_VDW];
649 coul = &nlist[eNL_QQ];
650 }
651
652 if (!bFreeEnergy)
653 {
654 if (iwater != esolNO)
655 {
656 /* Loop over the atoms in the i charge group */
657 i_atom = i0;
658 gid = GID(igid, jgid, ngid);
659 /* Create new i_atom for each energy group */
660 if (bDoCoul && bDoVdW)
661 {
662 new_i_nblist(vdwc, i_atom, shift, gid);
663 new_i_nblist(vdwc_ww, i_atom, shift, gid);
664 }
665 if (bDoVdW)
666 {
667 new_i_nblist(vdw, i_atom, shift, gid);
668 }
669 if (bDoCoul)
670 {
671 new_i_nblist(coul, i_atom, shift, gid);
672 new_i_nblist(coul_ww, i_atom, shift, gid);
673 }
674 /* Loop over the j charge groups */
675 for (j = 0; (j < nj); j++)
676 {
677 jcg = jjcg[j];
678
679 if (jcg == icg)
680 {
681 continue;
682 }
683
684 jj0 = index[jcg];
685 jwater = GET_CGINFO_SOLOPT(cginfo[jcg]);
686
687 if (iwater == esolSPC && jwater == esolSPC)
688 {
689 /* Interaction between two SPC molecules */
690 if (!bDoCoul)
691 {
692 /* VdW only - only first atoms in each water interact */
693 add_j_to_nblist(vdw, jj0);
694 }
695 else
696 {
697 /* One entry for the entire water-water interaction */
698 if (!bDoVdW)
699 {
700 add_j_to_nblist(coul_ww, jj0);
701 }
702 else
703 {
704 add_j_to_nblist(vdwc_ww, jj0);
705 }
706 }
707 }
708 else if (iwater == esolTIP4P && jwater == esolTIP4P)
709 {
710 /* Interaction between two TIP4p molecules */
711 if (!bDoCoul)
712 {
713 /* VdW only - only first atoms in each water interact */
714 add_j_to_nblist(vdw, jj0);
715 }
716 else
717 {
718 /* One entry for the entire water-water interaction */
719 if (!bDoVdW)
720 {
721 add_j_to_nblist(coul_ww, jj0);
722 }
723 else
724 {
725 add_j_to_nblist(vdwc_ww, jj0);
726 }
727 }
728 }
729 else
730 {
731 /* j charge group is not water, but i is.
732 * Add entries to the water-other_atom lists; the geometry of the water
733 * molecule doesn't matter - that is taken care of in the nonbonded kernel,
734 * so we don't care if it is SPC or TIP4P...
735 */
736
737 jj1 = index[jcg+1];
738
739 if (!bDoVdW)
740 {
741 for (jj = jj0; (jj < jj1); jj++)
742 {
743 if (charge[jj] != 0)
744 {
745 add_j_to_nblist(coul, jj);
746 }
747 }
748 }
749 else if (!bDoCoul)
750 {
751 for (jj = jj0; (jj < jj1); jj++)
752 {
753 if (bHaveVdW[type[jj]])
754 {
755 add_j_to_nblist(vdw, jj);
756 }
757 }
758 }
759 else
760 {
761 /* _charge_ _groups_ interact with both coulomb and LJ */
762 /* Check which atoms we should add to the lists! */
763 for (jj = jj0; (jj < jj1); jj++)
764 {
765 if (bHaveVdW[type[jj]])
766 {
767 if (charge[jj] != 0)
768 {
769 add_j_to_nblist(vdwc, jj);
770 }
771 else
772 {
773 add_j_to_nblist(vdw, jj);
774 }
775 }
776 else if (charge[jj] != 0)
777 {
778 add_j_to_nblist(coul, jj);
779 }
780 }
781 }
782 }
783 }
784 close_i_nblist(vdw);
785 close_i_nblist(coul);
786 close_i_nblist(vdwc);
787 close_i_nblist(coul_ww);
788 close_i_nblist(vdwc_ww);
789 }
790 else
791 {
792 /* no solvent as i charge group */
793 /* Loop over the atoms in the i charge group */
794 for (i = 0; i < nicg; i++)
795 {
796 i_atom = i0+i;
797 gid = GID(igid, jgid, ngid);
798 qi = charge[i_atom];
799
800 /* Create new i_atom for each energy group */
801 if (bDoVdW && bDoCoul)
802 {
803 new_i_nblist(vdwc, i_atom, shift, gid);
804 }
805 if (bDoVdW)
806 {
807 new_i_nblist(vdw, i_atom, shift, gid);
808 }
809 if (bDoCoul)
810 {
811 new_i_nblist(coul, i_atom, shift, gid);
812 }
813 bDoVdW_i = (bDoVdW && bHaveVdW[type[i_atom]]);
814 bDoCoul_i = (bDoCoul && qi != 0);
815
816 if (bDoVdW_i || bDoCoul_i)
817 {
818 /* Loop over the j charge groups */
819 for (j = 0; (j < nj); j++)
820 {
821 jcg = jjcg[j];
822
823 /* Check for large charge groups */
824 if (jcg == icg)
825 {
826 jj0 = i0 + i + 1;
827 }
828 else
829 {
830 jj0 = index[jcg];
831 }
832
833 jj1 = index[jcg+1];
834 /* Finally loop over the atoms in the j-charge group */
835 for (jj = jj0; jj < jj1; jj++)
836 {
837 bNotEx = NOTEXCL(bExcl, i, jj);
838
839 if (bNotEx)
840 {
841 if (!bDoVdW_i)
842 {
843 if (charge[jj] != 0)
844 {
845 add_j_to_nblist(coul, jj);
846 }
847 }
848 else if (!bDoCoul_i)
849 {
850 if (bHaveVdW[type[jj]])
851 {
852 add_j_to_nblist(vdw, jj);
853 }
854 }
855 else
856 {
857 if (bHaveVdW[type[jj]])
858 {
859 if (charge[jj] != 0)
860 {
861 add_j_to_nblist(vdwc, jj);
862 }
863 else
864 {
865 add_j_to_nblist(vdw, jj);
866 }
867 }
868 else if (charge[jj] != 0)
869 {
870 add_j_to_nblist(coul, jj);
871 }
872 }
873 }
874 }
875 }
876 }
877 close_i_nblist(vdw);
878 close_i_nblist(coul);
879 close_i_nblist(vdwc);
880 }
881 }
882 }
883 else
884 {
885 /* we are doing free energy */
886 vdwc_free = &nlist[eNL_VDWQQ_FREE];
887 vdw_free = &nlist[eNL_VDW_FREE];
888 coul_free = &nlist[eNL_QQ_FREE];
889 /* Loop over the atoms in the i charge group */
890 for (i = 0; i < nicg; i++)
891 {
892 i_atom = i0+i;
893 gid = GID(igid, jgid, ngid);
894 qi = charge[i_atom];
895 qiB = chargeB[i_atom];
896
897 /* Create new i_atom for each energy group */
898 if (bDoVdW && bDoCoul)
899 {
900 new_i_nblist(vdwc, i_atom, shift, gid);
901 }
902 if (bDoVdW)
903 {
904 new_i_nblist(vdw, i_atom, shift, gid);
905 }
906 if (bDoCoul)
907 {
908 new_i_nblist(coul, i_atom, shift, gid);
909 }
910
911 new_i_nblist(vdw_free, i_atom, shift, gid);
912 new_i_nblist(coul_free, i_atom, shift, gid);
913 new_i_nblist(vdwc_free, i_atom, shift, gid);
914
915 bDoVdW_i = (bDoVdW &&
916 (bHaveVdW[type[i_atom]] || bHaveVdW[typeB[i_atom]]));
917 bDoCoul_i = (bDoCoul && (qi != 0 || qiB != 0));
918 /* For TIP4P the first atom does not have a charge,
919 * but the last three do. So we should still put an atom
920 * without LJ but with charge in the water-atom neighborlist
921 * for a TIP4p i charge group.
922 * For SPC type water the first atom has LJ and charge,
923 * so there is no such problem.
924 */
925 if (iwater == esolNO)
926 {
927 bDoCoul_i_sol = bDoCoul_i;
928 }
929 else
930 {
931 bDoCoul_i_sol = bDoCoul;
932 }
933
934 if (bDoVdW_i || bDoCoul_i_sol)
935 {
936 /* Loop over the j charge groups */
937 for (j = 0; (j < nj); j++)
938 {
939 jcg = jjcg[j];
940
941 /* Check for large charge groups */
942 if (jcg == icg)
943 {
944 jj0 = i0 + i + 1;
945 }
946 else
947 {
948 jj0 = index[jcg];
949 }
950
951 jj1 = index[jcg+1];
952 /* Finally loop over the atoms in the j-charge group */
953 bFree = bPert[i_atom];
954 for (jj = jj0; (jj < jj1); jj++)
955 {
956 bFreeJ = bFree || bPert[jj];
957 /* Complicated if, because the water H's should also
958 * see perturbed j-particles
959 */
960 if (iwater == esolNO || i == 0 || bFreeJ)
961 {
962 bNotEx = NOTEXCL(bExcl, i, jj);
963
964 if (bNotEx)
965 {
966 if (bFreeJ)
967 {
968 if (!bDoVdW_i)
969 {
970 if (charge[jj] != 0 || chargeB[jj] != 0)
971 {
972 add_j_to_nblist(coul_free, jj);
973 }
974 }
975 else if (!bDoCoul_i)
976 {
977 if (bHaveVdW[type[jj]] || bHaveVdW[typeB[jj]])
978 {
979 add_j_to_nblist(vdw_free, jj);
980 }
981 }
982 else
983 {
984 if (bHaveVdW[type[jj]] || bHaveVdW[typeB[jj]])
985 {
986 if (charge[jj] != 0 || chargeB[jj] != 0)
987 {
988 add_j_to_nblist(vdwc_free, jj);
989 }
990 else
991 {
992 add_j_to_nblist(vdw_free, jj);
993 }
994 }
995 else if (charge[jj] != 0 || chargeB[jj] != 0)
996 {
997 add_j_to_nblist(coul_free, jj);
998 }
999 }
1000 }
1001 else if (!bDoVdW_i)
1002 {
1003 /* This is done whether or not bWater is set */
1004 if (charge[jj] != 0)
1005 {
1006 add_j_to_nblist(coul, jj);
1007 }
1008 }
1009 else if (!bDoCoul_i_sol)
1010 {
1011 if (bHaveVdW[type[jj]])
1012 {
1013 add_j_to_nblist(vdw, jj);
1014 }
1015 }
1016 else
1017 {
1018 if (bHaveVdW[type[jj]])
1019 {
1020 if (charge[jj] != 0)
1021 {
1022 add_j_to_nblist(vdwc, jj);
1023 }
1024 else
1025 {
1026 add_j_to_nblist(vdw, jj);
1027 }
1028 }
1029 else if (charge[jj] != 0)
1030 {
1031 add_j_to_nblist(coul, jj);
1032 }
1033 }
1034 }
1035 }
1036 }
1037 }
1038 }
1039 close_i_nblist(vdw);
1040 close_i_nblist(coul);
1041 close_i_nblist(vdwc);
1042 close_i_nblist(vdw_free);
1043 close_i_nblist(coul_free);
1044 close_i_nblist(vdwc_free);
1045 }
1046 }
1047 }
1048
1049 static void
1050 put_in_list_qmmm(gmx_bool gmx_unused bHaveVdW[],
1051 int ngid,
1052 t_mdatoms gmx_unused * md,
1053 int icg,
1054 int jgid,
1055 int nj,
1056 int jjcg[],
1057 int index[],
1058 t_excl bExcl[],
1059 int shift,
1060 t_forcerec * fr,
1061 gmx_bool gmx_unused bDoVdW,
1062 gmx_bool gmx_unused bDoCoul,
1063 int gmx_unused solvent_opt)
1064 {
1065 t_nblist * coul;
1066 int i, j, jcg, igid, gid;
1067 int jj, jj0, jj1, i_atom;
1068 int i0, nicg;
1069 gmx_bool bNotEx;
1070
1071 /* Get atom range */
1072 i0 = index[icg];
1073 nicg = index[icg+1]-i0;
1074
1075 /* Get the i charge group info */
1076 igid = GET_CGINFO_GID(fr->cginfo[icg]);
1077
1078 coul = fr->QMMMlist;
1079
1080 /* Loop over atoms in the ith charge group */
1081 for (i = 0; i < nicg; i++)
1082 {
1083 i_atom = i0+i;
1084 gid = GID(igid, jgid, ngid);
1085 /* Create new i_atom for each energy group */
1086 new_i_nblist(coul, i_atom, shift, gid);
1087
1088 /* Loop over the j charge groups */
1089 for (j = 0; j < nj; j++)
1090 {
1091 jcg = jjcg[j];
1092
1093 /* Charge groups cannot have QM and MM atoms simultaneously */
1094 if (jcg != icg)
1095 {
1096 jj0 = index[jcg];
1097 jj1 = index[jcg+1];
1098 /* Finally loop over the atoms in the j-charge group */
1099 for (jj = jj0; jj < jj1; jj++)
1100 {
1101 bNotEx = NOTEXCL(bExcl, i, jj);
1102 if (bNotEx)
1103 {
1104 add_j_to_nblist(coul, jj);
1105 }
1106 }
1107 }
1108 }
1109 close_i_nblist(coul);
1110 }
1111 }
1112
1113 static void
1114 put_in_list_cg(gmx_bool gmx_unused bHaveVdW[],
1115 int ngid,
1116 t_mdatoms gmx_unused * md,
1117 int icg,
1118 int jgid,
1119 int nj,
1120 int jjcg[],
1121 int index[],
1122 t_excl bExcl[],
1123 int shift,
1124 t_forcerec * fr,
1125 gmx_bool gmx_unused bDoVdW,
1126 gmx_bool gmx_unused bDoCoul,
1127 int gmx_unused solvent_opt)
1128 {
1129 int cginfo;
1130 int igid, gid, nbl_ind;
1131 t_nblist * vdwc;
1132 int j, jcg;
1133
1134 cginfo = fr->cginfo[icg];
1135
1136 igid = GET_CGINFO_GID(cginfo);
1137 gid = GID(igid, jgid, ngid);
1138
1139 /* Unpack pointers to neighbourlist structs */
1140 if (fr->nnblists == 1)
1141 {
1142 nbl_ind = 0;
1143 }
1144 else
1145 {
1146 nbl_ind = fr->gid2nblists[gid];
1147 }
1148 vdwc = &fr->nblists[nbl_ind].nlist_sr[eNL_VDWQQ];
1149
1150 /* Make a new neighbor list for charge group icg.
1151 * Currently simply one neighbor list is made with LJ and Coulomb.
1152 * If required, zero interactions could be removed here
1153 * or in the force loop.
1154 */
1155 new_i_nblist(vdwc, index[icg], shift, gid);
1156 vdwc->iinr_end[vdwc->nri] = index[icg+1];
1157
1158 for (j = 0; (j < nj); j++)
1159 {
1160 jcg = jjcg[j];
1161 /* Skip the icg-icg pairs if all self interactions are excluded */
1162 if (!(jcg == icg && GET_CGINFO_EXCL_INTRA(cginfo)))
1163 {
1164 /* Here we add the j charge group jcg to the list,
1165 * exclusions are also added to the list.
1166 */
1167 add_j_to_nblist_cg(vdwc, index[jcg], index[jcg+1], bExcl, icg == jcg);
1168 }
1169 }
1170
1171 close_i_nblist(vdwc);
1172 }
1173
1174 static void setexcl(int start, int end, t_blocka *excl, gmx_bool b,
1175 t_excl bexcl[])
1176 {
1177 int i, k;
1178
1179 if (b)
1180 {
1181 for (i = start; i < end; i++)
1182 {
1183 for (k = excl->index[i]; k < excl->index[i+1]; k++)
1184 {
1185 SETEXCL(bexcl, i-start, excl->a[k]);
1186 }
1187 }
1188 }
1189 else
1190 {
1191 for (i = start; i < end; i++)
1192 {
1193 for (k = excl->index[i]; k < excl->index[i+1]; k++)
1194 {
1195 RMEXCL(bexcl, i-start, excl->a[k]);
1196 }
1197 }
1198 }
1199 }
1200
1201 int calc_naaj(int icg, int cgtot)
1202 {
1203 int naaj;
1204
1205 if ((cgtot % 2) == 1)
1206 {
1207 /* Odd number of charge groups, easy */
1208 naaj = 1 + (cgtot/2);
1209 }
1210 else if ((cgtot % 4) == 0)
1211 {
1212 /* Multiple of four is hard */
1213 if (icg < cgtot/2)
1214 {
1215 if ((icg % 2) == 0)
1216 {
1217 naaj = 1+(cgtot/2);
1218 }
1219 else
1220 {
1221 naaj = cgtot/2;
1222 }
1223 }
1224 else
1225 {
1226 if ((icg % 2) == 1)
1227 {
1228 naaj = 1+(cgtot/2);
1229 }
1230 else
1231 {
1232 naaj = cgtot/2;
1233 }
1234 }
1235 }
1236 else
1237 {
1238 /* cgtot/2 = odd */
1239 if ((icg % 2) == 0)
1240 {
1241 naaj = 1+(cgtot/2);
1242 }
1243 else
1244 {
1245 naaj = cgtot/2;
1246 }
1247 }
1248 #ifdef DEBUG
1249 fprintf(log, "naaj=%d\n", naaj);
1250 #endif
1251
1252 return naaj;
1253 }
1254
1255 /************************************************
1256 *
1257 * S I M P L E C O R E S T U F F
1258 *
1259 ************************************************/
1260
1261 static real calc_image_tric(rvec xi, rvec xj, matrix box,
1262 rvec b_inv, int *shift)
1263 {
1264 /* This code assumes that the cut-off is smaller than
1265 * a half times the smallest diagonal element of the box.
1266 */
1267 const real h25 = 2.5;
1268 real dx, dy, dz;
1269 real r2;
1270 int tx, ty, tz;
1271
1272 /* Compute diff vector */
1273 dz = xj[ZZ] - xi[ZZ];
1274 dy = xj[YY] - xi[YY];
1275 dx = xj[XX] - xi[XX];
1276
1277 /* Perform NINT operation, using trunc operation, therefore
1278 * we first add 2.5 then subtract 2 again
1279 */
1280 tz = static_cast<int>(dz*b_inv[ZZ] + h25);
1281 tz -= 2;
1282 dz -= tz*box[ZZ][ZZ];
1283 dy -= tz*box[ZZ][YY];
1284 dx -= tz*box[ZZ][XX];
1285
1286 ty = static_cast<int>(dy*b_inv[YY] + h25);
1287 ty -= 2;
1288 dy -= ty*box[YY][YY];
1289 dx -= ty*box[YY][XX];
1290
1291 tx = static_cast<int>(dx*b_inv[XX]+h25);
1292 tx -= 2;
1293 dx -= tx*box[XX][XX];
1294
1295 /* Distance squared */
1296 r2 = (dx*dx) + (dy*dy) + (dz*dz);
1297
1298 *shift = XYZ2IS(tx, ty, tz);
1299
1300 return r2;
1301 }
1302
1303 static real calc_image_rect(rvec xi, rvec xj, rvec box_size,
1304 rvec b_inv, int *shift)
1305 {
1306 const real h15 = 1.5;
1307 real ddx, ddy, ddz;
1308 real dx, dy, dz;
1309 real r2;
1310 int tx, ty, tz;
1311
1312 /* Compute diff vector */
1313 dx = xj[XX] - xi[XX];
1314 dy = xj[YY] - xi[YY];
1315 dz = xj[ZZ] - xi[ZZ];
1316
1317 /* Perform NINT operation, using trunc operation, therefore
1318 * we first add 1.5 then subtract 1 again
1319 */
1320 tx = static_cast<int>(dx*b_inv[XX] + h15);
1321 ty = static_cast<int>(dy*b_inv[YY] + h15);
1322 tz = static_cast<int>(dz*b_inv[ZZ] + h15);
1323 tx--;
1324 ty--;
1325 tz--;
1326
1327 /* Correct diff vector for translation */
1328 ddx = tx*box_size[XX] - dx;
1329 ddy = ty*box_size[YY] - dy;
1330 ddz = tz*box_size[ZZ] - dz;
1331
1332 /* Distance squared */
1333 r2 = (ddx*ddx) + (ddy*ddy) + (ddz*ddz);
1334
1335 *shift = XYZ2IS(tx, ty, tz);
1336
1337 return r2;
1338 }
1339
1340 static void add_simple(t_ns_buf * nsbuf, int nrj, int cg_j,
1341 gmx_bool bHaveVdW[], int ngid, t_mdatoms *md,
1342 int icg, int jgid, t_block *cgs, t_excl bexcl[],
1343 int shift, t_forcerec *fr, put_in_list_t *put_in_list)
1344 {
1345 if (nsbuf->nj + nrj > MAX_CG)
1346 {
1347 put_in_list(bHaveVdW, ngid, md, icg, jgid, nsbuf->ncg, nsbuf->jcg,
1348 cgs->index, bexcl, shift, fr, TRUE, TRUE, fr->solvent_opt);
1349 /* Reset buffer contents */
1350 nsbuf->ncg = nsbuf->nj = 0;
1351 }
1352 nsbuf->jcg[nsbuf->ncg++] = cg_j;
1353 nsbuf->nj += nrj;
1354 }
1355
1356 static void ns_inner_tric(rvec x[], int icg, int *i_egp_flags,
1357 int njcg, int jcg[],
1358 matrix box, rvec b_inv, real rcut2,
1359 t_block *cgs, t_ns_buf **ns_buf,
1360 gmx_bool bHaveVdW[], int ngid, t_mdatoms *md,
1361 t_excl bexcl[], t_forcerec *fr,
1362 put_in_list_t *put_in_list)
1363 {
1364 int shift;
1365 int j, nrj, jgid;
1366 int *cginfo = fr->cginfo;
1367 int cg_j, *cgindex;
1368
1369 cgindex = cgs->index;
1370 shift = CENTRAL;
1371 for (j = 0; (j < njcg); j++)
1372 {
1373 cg_j = jcg[j];
1374 nrj = cgindex[cg_j+1]-cgindex[cg_j];
1375 if (calc_image_tric(x[icg], x[cg_j], box, b_inv, &shift) < rcut2)
1376 {
1377 jgid = GET_CGINFO_GID(cginfo[cg_j]);
1378 if (!(i_egp_flags[jgid] & EGP_EXCL))
1379 {
1380 add_simple(&ns_buf[jgid][shift], nrj, cg_j,
1381 bHaveVdW, ngid, md, icg, jgid, cgs, bexcl, shift, fr,
1382 put_in_list);
1383 }
1384 }
1385 }
1386 }
1387
1388 static void ns_inner_rect(rvec x[], int icg, int *i_egp_flags,
1389 int njcg, int jcg[],
1390 gmx_bool bBox, rvec box_size, rvec b_inv, real rcut2,
1391 t_block *cgs, t_ns_buf **ns_buf,
1392 gmx_bool bHaveVdW[], int ngid, t_mdatoms *md,
1393 t_excl bexcl[], t_forcerec *fr,
1394 put_in_list_t *put_in_list)
1395 {
1396 int shift;
1397 int j, nrj, jgid;
1398 int *cginfo = fr->cginfo;
1399 int cg_j, *cgindex;
1400
1401 cgindex = cgs->index;
1402 if (bBox)
1403 {
1404 shift = CENTRAL;
1405 for (j = 0; (j < njcg); j++)
1406 {
1407 cg_j = jcg[j];
1408 nrj = cgindex[cg_j+1]-cgindex[cg_j];
1409 if (calc_image_rect(x[icg], x[cg_j], box_size, b_inv, &shift) < rcut2)
1410 {
1411 jgid = GET_CGINFO_GID(cginfo[cg_j]);
1412 if (!(i_egp_flags[jgid] & EGP_EXCL))
1413 {
1414 add_simple(&ns_buf[jgid][shift], nrj, cg_j,
1415 bHaveVdW, ngid, md, icg, jgid, cgs, bexcl, shift, fr,
1416 put_in_list);
1417 }
1418 }
1419 }
1420 }
1421 else
1422 {
1423 for (j = 0; (j < njcg); j++)
1424 {
1425 cg_j = jcg[j];
1426 nrj = cgindex[cg_j+1]-cgindex[cg_j];
1427 if ((rcut2 == 0) || (distance2(x[icg], x[cg_j]) < rcut2))
1428 {
1429 jgid = GET_CGINFO_GID(cginfo[cg_j]);
1430 if (!(i_egp_flags[jgid] & EGP_EXCL))
1431 {
1432 add_simple(&ns_buf[jgid][CENTRAL], nrj, cg_j,
1433 bHaveVdW, ngid, md, icg, jgid, cgs, bexcl, CENTRAL, fr,
1434 put_in_list);
1435 }
1436 }
1437 }
1438 }
1439 }
1440
1441 /* ns_simple_core needs to be adapted for QMMM still 2005 */
1442
1443 static int ns_simple_core(t_forcerec *fr,
1444 gmx_localtop_t *top,
1445 t_mdatoms *md,
1446 matrix box, rvec box_size,
1447 t_excl bexcl[], int *aaj,
1448 int ngid, t_ns_buf **ns_buf,
1449 put_in_list_t *put_in_list, gmx_bool bHaveVdW[])
1450 {
1451 int naaj, k;
1452 real rlist2;
1453 int nsearch, icg, igid, nn;
1454 int *cginfo;
1455 t_ns_buf *nsbuf;
1456 /* int *i_atoms; */
1457 t_block *cgs = &(top->cgs);
1458 t_blocka *excl = &(top->excls);
1459 rvec b_inv;
1460 int m;
1461 gmx_bool bBox, bTriclinic;
1462 int *i_egp_flags;
1463
1464 rlist2 = sqr(fr->rlist);
1465
1466 bBox = (fr->ePBC != epbcNONE);
1467 if (bBox)
1468 {
1469 for (m = 0; (m < DIM); m++)
1470 {
1471 if (gmx_numzero(box_size[m]))
1472 {
1473 gmx_fatal(FARGS, "Dividing by zero box size!");
1474 }
1475 b_inv[m] = 1.0/box_size[m];
1476 }
1477 bTriclinic = TRICLINIC(box);
1478 }
1479 else
1480 {
1481 bTriclinic = FALSE;
1482 }
1483
1484 cginfo = fr->cginfo;
1485
1486 nsearch = 0;
1487 for (icg = fr->cg0; (icg < fr->hcg); icg++)
1488 {
1489 /*
1490 i0 = cgs->index[icg];
1491 nri = cgs->index[icg+1]-i0;
1492 i_atoms = &(cgs->a[i0]);
1493 i_eg_excl = fr->eg_excl + ngid*md->cENER[*i_atoms];
1494 setexcl(nri,i_atoms,excl,TRUE,bexcl);
1495 */
1496 igid = GET_CGINFO_GID(cginfo[icg]);
1497 i_egp_flags = fr->egp_flags + ngid*igid;
1498 setexcl(cgs->index[icg], cgs->index[icg+1], excl, TRUE, bexcl);
1499
1500 naaj = calc_naaj(icg, cgs->nr);
1501 if (bTriclinic)
1502 {
1503 ns_inner_tric(fr->cg_cm, icg, i_egp_flags, naaj, &(aaj[icg]),
1504 box, b_inv, rlist2, cgs, ns_buf,
1505 bHaveVdW, ngid, md, bexcl, fr, put_in_list);
1506 }
1507 else
1508 {
1509 ns_inner_rect(fr->cg_cm, icg, i_egp_flags, naaj, &(aaj[icg]),
1510 bBox, box_size, b_inv, rlist2, cgs, ns_buf,
1511 bHaveVdW, ngid, md, bexcl, fr, put_in_list);
1512 }
1513 nsearch += naaj;
1514
1515 for (nn = 0; (nn < ngid); nn++)
1516 {
1517 for (k = 0; (k < SHIFTS); k++)
1518 {
1519 nsbuf = &(ns_buf[nn][k]);
1520 if (nsbuf->ncg > 0)
1521 {
1522 put_in_list(bHaveVdW, ngid, md, icg, nn, nsbuf->ncg, nsbuf->jcg,
1523 cgs->index, bexcl, k, fr, TRUE, TRUE, fr->solvent_opt);
1524 nsbuf->ncg = nsbuf->nj = 0;
1525 }
1526 }
1527 }
1528 /* setexcl(nri,i_atoms,excl,FALSE,bexcl); */
1529 setexcl(cgs->index[icg], cgs->index[icg+1], excl, FALSE, bexcl);
1530 }
1531 close_neighbor_lists(fr, FALSE);
1532
1533 return nsearch;
1534 }
1535
1536 /************************************************
1537 *
1538 * N S 5 G R I D S T U F F
1539 *
1540 ************************************************/
1541
1542 static gmx_inline void get_dx_dd(int Nx, real gridx, real rc2, int xgi, real x,
1543 int ncpddc, int shift_min, int shift_max,
1544 int *g0, int *g1, real *dcx2)
1545 {
1546 real dcx, tmp;
1547 int g_min, g_max, shift_home;
1548
1549 if (xgi < 0)
1550 {
1551 g_min = 0;
1552 g_max = Nx - 1;
1553 *g0 = 0;
1554 *g1 = -1;
1555 }
1556 else if (xgi >= Nx)
1557 {
1558 g_min = 0;
1559 g_max = Nx - 1;
1560 *g0 = Nx;
1561 *g1 = Nx - 1;
1562 }
1563 else
1564 {
1565 if (ncpddc == 0)
1566 {
1567 g_min = 0;
1568 g_max = Nx - 1;
1569 }
1570 else
1571 {
1572 if (xgi < ncpddc)
1573 {
1574 shift_home = 0;
1575 }
1576 else
1577 {
1578 shift_home = -1;
1579 }
1580 g_min = (shift_min == shift_home ? 0 : ncpddc);
1581 g_max = (shift_max == shift_home ? ncpddc - 1 : Nx - 1);
1582 }
1583 if (shift_min > 0)
1584 {
1585 *g0 = g_min;
1586 *g1 = g_min - 1;
1587 }
1588 else if (shift_max < 0)
1589 {
1590 *g0 = g_max + 1;
1591 *g1 = g_max;
1592 }
1593 else
1594 {
1595 *g0 = xgi;
1596 *g1 = xgi;
1597 dcx2[xgi] = 0;
1598 }
1599 }
1600
1601 while (*g0 > g_min)
1602 {
1603 /* Check one grid cell down */
1604 dcx = ((*g0 - 1) + 1)*gridx - x;
1605 tmp = dcx*dcx;
1606 if (tmp >= rc2)
1607 {
1608 break;
1609 }
1610 (*g0)--;
1611 dcx2[*g0] = tmp;
1612 }
1613
1614 while (*g1 < g_max)
1615 {
1616 /* Check one grid cell up */
1617 dcx = (*g1 + 1)*gridx - x;
1618 tmp = dcx*dcx;
1619 if (tmp >= rc2)
1620 {
1621 break;
1622 }
1623 (*g1)++;
1624 dcx2[*g1] = tmp;
1625 }
1626 }
1627
1628
1629 #define sqr(x) ((x)*(x))
1630 #define calc_dx2(XI, YI, ZI, y) (sqr(XI-y[XX]) + sqr(YI-y[YY]) + sqr(ZI-y[ZZ]))
1631 #define calc_cyl_dx2(XI, YI, y) (sqr(XI-y[XX]) + sqr(YI-y[YY]))
1632 /****************************************************
1633 *
1634 * F A S T N E I G H B O R S E A R C H I N G
1635 *
1636 * Optimized neighboursearching routine using grid
1637 * at least 1x1x1, see GROMACS manual
1638 *
1639 ****************************************************/
1640
1641
1642 static void get_cutoff2(t_forcerec *fr, real *rs2)
1643 {
1644 *rs2 = sqr(fr->rlist);
1645 }
1646
1647 static void init_nsgrid_lists(t_forcerec *fr, int ngid, gmx_ns_t *ns)
1648 {
1649 real rs2;
1650 int j;
1651
1652 get_cutoff2(fr, &rs2);
1653
1654 /* Short range buffers */
1655 snew(ns->nl_sr, ngid);
1656 /* Counters */
1657 snew(ns->nsr, ngid);
1658
1659 for (j = 0; (j < ngid); j++)
1660 {
1661 snew(ns->nl_sr[j], MAX_CG);
1662 }
1663 if (debug)
1664 {
1665 fprintf(debug,
1666 "ns5_core: rs2 = %g (nm^2)\n",
1667 rs2);
1668 }
1669 }
1670
1671 static int nsgrid_core(t_commrec *cr, t_forcerec *fr,
1672 matrix box, int ngid,
1673 gmx_localtop_t *top,
1674 t_grid *grid,
1675 t_excl bexcl[], gmx_bool *bExcludeAlleg,
1676 t_mdatoms *md,
1677 put_in_list_t *put_in_list,
1678 gmx_bool bHaveVdW[],
1679 gmx_bool bMakeQMMMnblist)
1680 {
1681 gmx_ns_t *ns;
1682 int **nl_sr;
1683 int *nsr;
1684 gmx_domdec_t *dd;
1685 t_block *cgs = &(top->cgs);
1686 int *cginfo = fr->cginfo;
1687 /* int *i_atoms,*cgsindex=cgs->index; */
1688 ivec sh0, sh1, shp;
1689 int cell_x, cell_y, cell_z;
1690 int d, tx, ty, tz, dx, dy, dz, cj;
1691 #ifdef ALLOW_OFFDIAG_LT_HALFDIAG
1692 int zsh_ty, zsh_tx, ysh_tx;
1693 #endif
1694 int dx0, dx1, dy0, dy1, dz0, dz1;
1695 int Nx, Ny, Nz, shift = -1, j, nrj, nns, nn = -1;
1696 real gridx, gridy, gridz, grid_x, grid_y;
1697 real *dcx2, *dcy2, *dcz2;
1698 int zgi, ygi, xgi;
1699 int cg0, cg1, icg = -1, cgsnr, i0, igid, naaj, max_jcg;
1700 int jcg0, jcg1, jjcg, cgj0, jgid;
1701 int *grida, *gridnra, *gridind;
1702 rvec *cgcm, grid_offset;
1703 real r2, rs2, XI, YI, ZI, tmp1, tmp2;
1704 int *i_egp_flags;
1705 gmx_bool bDomDec, bTriclinicX, bTriclinicY;
1706 ivec ncpddc;
1707
1708 ns = fr->ns;
1709
1710 bDomDec = DOMAINDECOMP(cr);
1711 dd = cr->dd;
1712
1713 bTriclinicX = ((YY < grid->npbcdim &&
1714 (!bDomDec || dd->nc[YY] == 1) && box[YY][XX] != 0) ||
1715 (ZZ < grid->npbcdim &&
1716 (!bDomDec || dd->nc[ZZ] == 1) && box[ZZ][XX] != 0));
1717 bTriclinicY = (ZZ < grid->npbcdim &&
1718 (!bDomDec || dd->nc[ZZ] == 1) && box[ZZ][YY] != 0);
1719
1720 cgsnr = cgs->nr;
1721
1722 get_cutoff2(fr, &rs2);
1723
1724 nl_sr = ns->nl_sr;
1725 nsr = ns->nsr;
1726
1727 /* Unpack arrays */
1728 cgcm = fr->cg_cm;
1729 Nx = grid->n[XX];
1730 Ny = grid->n[YY];
1731 Nz = grid->n[ZZ];
1732 grida = grid->a;
1733 gridind = grid->index;
1734 gridnra = grid->nra;
1735 nns = 0;
1736
1737 gridx = grid->cell_size[XX];
1738 gridy = grid->cell_size[YY];
1739 gridz = grid->cell_size[ZZ];
1740 grid_x = 1/gridx;
1741 grid_y = 1/gridy;
1742 copy_rvec(grid->cell_offset, grid_offset);
1743 copy_ivec(grid->ncpddc, ncpddc);
1744 dcx2 = grid->dcx2;
1745 dcy2 = grid->dcy2;
1746 dcz2 = grid->dcz2;
1747
1748 #ifdef ALLOW_OFFDIAG_LT_HALFDIAG
1749 zsh_ty = floor(-box[ZZ][YY]/box[YY][YY]+0.5);
1750 zsh_tx = floor(-box[ZZ][XX]/box[XX][XX]+0.5);
1751 ysh_tx = floor(-box[YY][XX]/box[XX][XX]+0.5);
1752 if (zsh_tx != 0 && ysh_tx != 0)
1753 {
1754 /* This could happen due to rounding, when both ratios are 0.5 */
1755 ysh_tx = 0;
1756 }
1757 #endif
1758
1759 if (fr->n_tpi)
1760 {
1761 /* We only want a list for the test particle */
1762 cg0 = cgsnr - 1;
1763 }
1764 else
1765 {
1766 cg0 = grid->icg0;
1767 }
1768 cg1 = grid->icg1;
1769
1770 /* Set the shift range */
1771 for (d = 0; d < DIM; d++)
1772 {
1773 sh0[d] = -1;
1774 sh1[d] = 1;
1775 /* Check if we need periodicity shifts.
1776 * Without PBC or with domain decomposition we don't need them.
1777 */
1778 if (d >= ePBC2npbcdim(fr->ePBC) || (bDomDec && dd->nc[d] > 1))
1779 {
1780 shp[d] = 0;
1781 }
1782 else
1783 {
1784 if (d == XX &&
1785 box[XX][XX] - fabs(box[YY][XX]) - fabs(box[ZZ][XX]) < std::sqrt(rs2))
1786 {
1787 shp[d] = 2;
1788 }
1789 else
1790 {
1791 shp[d] = 1;
1792 }
1793 }
1794 }
1795
1796 /* Loop over charge groups */
1797 for (icg = cg0; (icg < cg1); icg++)
1798 {
1799 igid = GET_CGINFO_GID(cginfo[icg]);
1800 /* Skip this charge group if all energy groups are excluded! */
1801 if (bExcludeAlleg[igid])
1802 {
1803 continue;
1804 }
1805
1806 i0 = cgs->index[icg];
1807
1808 if (bMakeQMMMnblist)
1809 {
1810 /* Skip this charge group if it is not a QM atom while making a
1811 * QM/MM neighbourlist
1812 */
1813 if (md->bQM[i0] == FALSE)
1814 {
1815 continue; /* MM particle, go to next particle */
1816 }
1817
1818 /* Compute the number of charge groups that fall within the control
1819 * of this one (icg)
1820 */
1821 naaj = calc_naaj(icg, cgsnr);
1822 jcg0 = icg;
1823 jcg1 = icg + naaj;
1824 max_jcg = cgsnr;
1825 }
1826 else
1827 {
1828 /* make a normal neighbourlist */
1829
1830 if (bDomDec)
1831 {
1832 /* Get the j charge-group and dd cell shift ranges */
1833 dd_get_ns_ranges(cr->dd, icg, &jcg0, &jcg1, sh0, sh1);
1834 max_jcg = 0;
1835 }
1836 else
1837 {
1838 /* Compute the number of charge groups that fall within the control
1839 * of this one (icg)
1840 */
1841 naaj = calc_naaj(icg, cgsnr);
1842 jcg0 = icg;
1843 jcg1 = icg + naaj;
1844
1845 if (fr->n_tpi)
1846 {
1847 /* The i-particle is awlways the test particle,
1848 * so we want all j-particles
1849 */
1850 max_jcg = cgsnr - 1;
1851 }
1852 else
1853 {
1854 max_jcg = jcg1 - cgsnr;
1855 }
1856 }
1857 }
1858
1859 i_egp_flags = fr->egp_flags + igid*ngid;
1860
1861 /* Set the exclusions for the atoms in charge group icg using a bitmask */
1862 setexcl(i0, cgs->index[icg+1], &top->excls, TRUE, bexcl);
1863
1864 ci2xyz(grid, icg, &cell_x, &cell_y, &cell_z);
1865
1866 /* Changed iicg to icg, DvdS 990115
1867 * (but see consistency check above, DvdS 990330)
1868 */
1869 #ifdef NS5DB
1870 fprintf(log, "icg=%5d, naaj=%5d, cell %d %d %d\n",
1871 icg, naaj, cell_x, cell_y, cell_z);
1872 #endif
1873 /* Loop over shift vectors in three dimensions */
1874 for (tz = -shp[ZZ]; tz <= shp[ZZ]; tz++)
1875 {
1876 ZI = cgcm[icg][ZZ]+tz*box[ZZ][ZZ];
1877 /* Calculate range of cells in Z direction that have the shift tz */
1878 zgi = cell_z + tz*Nz;
1879 get_dx_dd(Nz, gridz, rs2, zgi, ZI-grid_offset[ZZ],
1880 ncpddc[ZZ], sh0[ZZ], sh1[ZZ], &dz0, &dz1, dcz2);
1881 if (dz0 > dz1)
1882 {
1883 continue;
1884 }
1885 for (ty = -shp[YY]; ty <= shp[YY]; ty++)
1886 {
1887 YI = cgcm[icg][YY]+ty*box[YY][YY]+tz*box[ZZ][YY];
1888 /* Calculate range of cells in Y direction that have the shift ty */
1889 if (bTriclinicY)
1890 {
1891 ygi = (int)(Ny + (YI - grid_offset[YY])*grid_y) - Ny;
1892 }
1893 else
1894 {
1895 ygi = cell_y + ty*Ny;
1896 }
1897 get_dx_dd(Ny, gridy, rs2, ygi, YI-grid_offset[YY],
1898 ncpddc[YY], sh0[YY], sh1[YY], &dy0, &dy1, dcy2);
1899 if (dy0 > dy1)
1900 {
1901 continue;
1902 }
1903 for (tx = -shp[XX]; tx <= shp[XX]; tx++)
1904 {
1905 XI = cgcm[icg][XX]+tx*box[XX][XX]+ty*box[YY][XX]+tz*box[ZZ][XX];
1906 /* Calculate range of cells in X direction that have the shift tx */
1907 if (bTriclinicX)
1908 {
1909 xgi = (int)(Nx + (XI - grid_offset[XX])*grid_x) - Nx;
1910 }
1911 else
1912 {
1913 xgi = cell_x + tx*Nx;
1914 }
1915 get_dx_dd(Nx, gridx, rs2, xgi, XI-grid_offset[XX],
1916 ncpddc[XX], sh0[XX], sh1[XX], &dx0, &dx1, dcx2);
1917 if (dx0 > dx1)
1918 {
1919 continue;
1920 }
1921 /* Get shift vector */
1922 shift = XYZ2IS(tx, ty, tz);
1923 #ifdef NS5DB
1924 range_check(shift, 0, SHIFTS);
1925 #endif
1926 for (nn = 0; (nn < ngid); nn++)
1927 {
1928 nsr[nn] = 0;
1929 }
1930 #ifdef NS5DB
1931 fprintf(log, "shift: %2d, dx0,1: %2d,%2d, dy0,1: %2d,%2d, dz0,1: %2d,%2d\n",
1932 shift, dx0, dx1, dy0, dy1, dz0, dz1);
1933 fprintf(log, "cgcm: %8.3f %8.3f %8.3f\n", cgcm[icg][XX],
1934 cgcm[icg][YY], cgcm[icg][ZZ]);
1935 fprintf(log, "xi: %8.3f %8.3f %8.3f\n", XI, YI, ZI);
1936 #endif
1937 for (dx = dx0; (dx <= dx1); dx++)
1938 {
1939 tmp1 = rs2 - dcx2[dx];
1940 for (dy = dy0; (dy <= dy1); dy++)
1941 {
1942 tmp2 = tmp1 - dcy2[dy];
1943 if (tmp2 > 0)
1944 {
1945 for (dz = dz0; (dz <= dz1); dz++)
1946 {
1947 if (tmp2 > dcz2[dz])
1948 {
1949 /* Find grid-cell cj in which possible neighbours are */
1950 cj = xyz2ci(Ny, Nz, dx, dy, dz);
1951
1952 /* Check out how many cgs (nrj) there in this cell */
1953 nrj = gridnra[cj];
1954
1955 /* Find the offset in the cg list */
1956 cgj0 = gridind[cj];
1957
1958 /* Check if all j's are out of range so we
1959 * can skip the whole cell.
1960 * Should save some time, especially with DD.
1961 */
1962 if (nrj == 0 ||
1963 (grida[cgj0] >= max_jcg &&
1964 (grida[cgj0] >= jcg1 || grida[cgj0+nrj-1] < jcg0)))
1965 {
1966 continue;
1967 }
1968
1969 /* Loop over cgs */
1970 for (j = 0; (j < nrj); j++)
1971 {
1972 jjcg = grida[cgj0+j];
1973
1974 /* check whether this guy is in range! */
1975 if ((jjcg >= jcg0 && jjcg < jcg1) ||
1976 (jjcg < max_jcg))
1977 {
1978 r2 = calc_dx2(XI, YI, ZI, cgcm[jjcg]);
1979 if (r2 < rs2)
1980 {
1981 /* jgid = gid[cgsatoms[cgsindex[jjcg]]]; */
1982 jgid = GET_CGINFO_GID(cginfo[jjcg]);
1983 /* check energy group exclusions */
1984 if (!(i_egp_flags[jgid] & EGP_EXCL))
1985 {
1986 if (nsr[jgid] >= MAX_CG)
1987 {
1988 /* Add to short-range list */
1989 put_in_list(bHaveVdW, ngid, md, icg, jgid,
1990 nsr[jgid], nl_sr[jgid],
1991 cgs->index, /* cgsatoms, */ bexcl,
1992 shift, fr, TRUE, TRUE, fr->solvent_opt);
1993 nsr[jgid] = 0;
1994 }
1995 nl_sr[jgid][nsr[jgid]++] = jjcg;
1996 }
1997 }
1998 nns++;
1999 }
2000 }
2001 }
2002 }
2003 }
2004 }
2005 }
2006 /* CHECK whether there is anything left in the buffers */
2007 for (nn = 0; (nn < ngid); nn++)
2008 {
2009 if (nsr[nn] > 0)
2010 {
2011 put_in_list(bHaveVdW, ngid, md, icg, nn, nsr[nn], nl_sr[nn],
2012 cgs->index, /* cgsatoms, */ bexcl,
2013 shift, fr, TRUE, TRUE, fr->solvent_opt);
2014 }
2015 }
2016 }
2017 }
2018 }
2019 setexcl(cgs->index[icg], cgs->index[icg+1], &top->excls, FALSE, bexcl);
2020 }
2021 /* No need to perform any left-over force calculations anymore (as we used to do here)
2022 * since we now save the proper long-range lists for later evaluation.
2023 */
2024
2025 /* Close neighbourlists */
2026 close_neighbor_lists(fr, bMakeQMMMnblist);
2027
2028 return nns;
2029 }
2030
2031 void ns_realloc_natoms(gmx_ns_t *ns, int natoms)
2032 {
2033 int i;
2034
2035 if (natoms > ns->nra_alloc)
2036 {
2037 ns->nra_alloc = over_alloc_dd(natoms);
2038 srenew(ns->bexcl, ns->nra_alloc);
2039 for (i = 0; i < ns->nra_alloc; i++)
2040 {
2041 ns->bexcl[i] = 0;
2042 }
2043 }
2044 }
2045
2046 void init_ns(FILE *fplog, const t_commrec *cr,
2047 gmx_ns_t *ns, t_forcerec *fr,
2048 const gmx_mtop_t *mtop)
2049 {
2050 int mt, icg, nr_in_cg, maxcg, i, j, jcg, ngid, ncg;
2051 t_block *cgs;
2052
2053 /* Compute largest charge groups size (# atoms) */
2054 nr_in_cg = 1;
2055 for (mt = 0; mt < mtop->nmoltype; mt++)
2056 {
2057 cgs = &mtop->moltype[mt].cgs;
2058 for (icg = 0; (icg < cgs->nr); icg++)
2059 {
2060 nr_in_cg = std::max(nr_in_cg, (int)(cgs->index[icg+1]-cgs->index[icg]));
2061 }
2062 }
2063
2064 /* Verify whether largest charge group is <= max cg.
2065 * This is determined by the type of the local exclusion type
2066 * Exclusions are stored in bits. (If the type is not large
2067 * enough, enlarge it, unsigned char -> unsigned short -> unsigned long)
2068 */
2069 maxcg = sizeof(t_excl)*8;
2070 if (nr_in_cg > maxcg)
2071 {
2072 gmx_fatal(FARGS, "Max #atoms in a charge group: %d > %d\n",
2073 nr_in_cg, maxcg);
2074 }
2075
2076 ngid = mtop->groups.grps[egcENER].nr;
2077 snew(ns->bExcludeAlleg, ngid);
2078 for (i = 0; i < ngid; i++)
2079 {
2080 ns->bExcludeAlleg[i] = TRUE;
2081 for (j = 0; j < ngid; j++)
2082 {
2083 if (!(fr->egp_flags[i*ngid+j] & EGP_EXCL))
2084 {
2085 ns->bExcludeAlleg[i] = FALSE;
2086 }
2087 }
2088 }
2089
2090 if (fr->bGrid)
2091 {
2092 /* Grid search */
2093 ns->grid = init_grid(fplog, fr);
2094 init_nsgrid_lists(fr, ngid, ns);
2095 }
2096 else
2097 {
2098 /* Simple search */
2099 snew(ns->ns_buf, ngid);
2100 for (i = 0; (i < ngid); i++)
2101 {
2102 snew(ns->ns_buf[i], SHIFTS);
2103 }
2104 ncg = ncg_mtop(mtop);
2105 snew(ns->simple_aaj, 2*ncg);
2106 for (jcg = 0; (jcg < ncg); jcg++)
2107 {
2108 ns->simple_aaj[jcg] = jcg;
2109 ns->simple_aaj[jcg+ncg] = jcg;
2110 }
2111 }
2112
2113 /* Create array that determines whether or not atoms have VdW */
2114 snew(ns->bHaveVdW, fr->ntype);
2115 for (i = 0; (i < fr->ntype); i++)
2116 {
2117 for (j = 0; (j < fr->ntype); j++)
2118 {
2119 ns->bHaveVdW[i] = (ns->bHaveVdW[i] ||
2120 (fr->bBHAM ?
2121 ((BHAMA(fr->nbfp, fr->ntype, i, j) != 0) ||
2122 (BHAMB(fr->nbfp, fr->ntype, i, j) != 0) ||
2123 (BHAMC(fr->nbfp, fr->ntype, i, j) != 0)) :
2124 ((C6(fr->nbfp, fr->ntype, i, j) != 0) ||
2125 (C12(fr->nbfp, fr->ntype, i, j) != 0))));
2126 }
2127 }
2128 if (debug)
2129 {
2130 pr_bvec(debug, 0, "bHaveVdW", ns->bHaveVdW, fr->ntype, TRUE);
2131 }
2132
2133 ns->nra_alloc = 0;
2134 ns->bexcl = NULL;
2135 if (!DOMAINDECOMP(cr))
2136 {
2137 ns_realloc_natoms(ns, mtop->natoms);
2138 }
2139
2140 ns->nblist_initialized = FALSE;
2141
2142 /* nbr list debug dump */
2143 {
2144 char *ptr = getenv("GMX_DUMP_NL");
2145 if (ptr)
2146 {
2147 ns->dump_nl = strtol(ptr, NULL, 10);
2148 if (fplog)
2149 {
2150 fprintf(fplog, "GMX_DUMP_NL = %d", ns->dump_nl);
2151 }
2152 }
2153 else
2154 {
2155 ns->dump_nl = 0;
2156 }
2157 }
2158 }
2159
2160
2161 int search_neighbours(FILE *log, t_forcerec *fr,
2162 matrix box,
2163 gmx_localtop_t *top,
2164 gmx_groups_t *groups,
2165 t_commrec *cr,
2166 t_nrnb *nrnb, t_mdatoms *md,
2167 gmx_bool bFillGrid)
2168 {
2169 t_block *cgs = &(top->cgs);
2170 rvec box_size, grid_x0, grid_x1;
2171 int m, ngid;
2172 real min_size, grid_dens;
2173 int nsearch;
2174 gmx_bool bGrid;
2175 int start, end;
2176 gmx_ns_t *ns;
2177 t_grid *grid;
2178 gmx_domdec_zones_t *dd_zones;
2179 put_in_list_t *put_in_list;
2180
2181 ns = fr->ns;
2182
2183 /* Set some local variables */
2184 bGrid = fr->bGrid;
2185 ngid = groups->grps[egcENER].nr;
2186
2187 for (m = 0; (m < DIM); m++)
2188 {
2189 box_size[m] = box[m][m];
2190 }
2191
2192 if (fr->ePBC != epbcNONE)
2193 {
2194 if (sqr(fr->rlist) >= max_cutoff2(fr->ePBC, box))
2195 {
2196 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.");
2197 }
2198 if (!bGrid)
2199 {
2200 min_size = std::min(box_size[XX], std::min(box_size[YY], box_size[ZZ]));
2201 if (2*fr->rlist >= min_size)
2202 {
2203 gmx_fatal(FARGS, "One of the box diagonal elements has become smaller than twice the cut-off length.");
2204 }
2205 }
2206 }
2207
2208 if (DOMAINDECOMP(cr))
2209 {
2210 ns_realloc_natoms(ns, cgs->index[cgs->nr]);
2211 }
2212
2213 /* Reset the neighbourlists */
2214 reset_neighbor_lists(fr);
2215
2216 if (bGrid && bFillGrid)
2217 {
2218
2219 grid = ns->grid;
2220 if (DOMAINDECOMP(cr))
2221 {
2222 dd_zones = domdec_zones(cr->dd);
2223 }
2224 else
2225 {
2226 dd_zones = NULL;
2227
2228 get_nsgrid_boundaries(grid->nboundeddim, box, NULL, NULL, NULL, NULL,
2229 cgs->nr, fr->cg_cm, grid_x0, grid_x1, &grid_dens);
2230
2231 grid_first(log, grid, NULL, NULL, box, grid_x0, grid_x1,
2232 fr->rlist, grid_dens);
2233 }
2234
2235 start = 0;
2236 end = cgs->nr;
2237
2238 if (DOMAINDECOMP(cr))
2239 {
2240 end = cgs->nr;
2241 fill_grid(dd_zones, grid, end, -1, end, fr->cg_cm);
2242 grid->icg0 = 0;
2243 grid->icg1 = dd_zones->izone[dd_zones->nizone-1].cg1;
2244 }
2245 else
2246 {
2247 fill_grid(NULL, grid, cgs->nr, fr->cg0, fr->hcg, fr->cg_cm);
2248 grid->icg0 = fr->cg0;
2249 grid->icg1 = fr->hcg;
2250 }
2251
2252 calc_elemnr(grid, start, end, cgs->nr);
2253 calc_ptrs(grid);
2254 grid_last(grid, start, end, cgs->nr);
2255
2256 if (gmx_debug_at)
2257 {
2258 check_grid(grid);
2259 print_grid(debug, grid);
2260 }
2261 }
2262 else if (fr->n_tpi)
2263 {
2264 /* Set the grid cell index for the test particle only.
2265 * The cell to cg index is not corrected, but that does not matter.
2266 */
2267 fill_grid(NULL, ns->grid, fr->hcg, fr->hcg-1, fr->hcg, fr->cg_cm);
2268 }
2269
2270 if (!fr->ns->bCGlist)
2271 {
2272 put_in_list = put_in_list_at;
2273 }
2274 else
2275 {
2276 put_in_list = put_in_list_cg;
2277 }
2278
2279 /* Do the core! */
2280 if (bGrid)
2281 {
2282 grid = ns->grid;
2283 nsearch = nsgrid_core(cr, fr, box, ngid, top,
2284 grid, ns->bexcl, ns->bExcludeAlleg,
2285 md, put_in_list, ns->bHaveVdW,
2286 FALSE);
2287
2288 /* neighbour searching withouth QMMM! QM atoms have zero charge in
2289 * the classical calculation. The charge-charge interaction
2290 * between QM and MM atoms is handled in the QMMM core calculation
2291 * (see QMMM.c). The VDW however, we'd like to compute classically
2292 * and the QM MM atom pairs have just been put in the
2293 * corresponding neighbourlists. in case of QMMM we still need to
2294 * fill a special QMMM neighbourlist that contains all neighbours
2295 * of the QM atoms. If bQMMM is true, this list will now be made:
2296 */
2297 if (fr->bQMMM && fr->qr->QMMMscheme != eQMMMschemeoniom)
2298 {
2299 nsearch += nsgrid_core(cr, fr, box, ngid, top,
2300 grid, ns->bexcl, ns->bExcludeAlleg,
2301 md, put_in_list_qmmm, ns->bHaveVdW,
2302 TRUE);
2303 }
2304 }
2305 else
2306 {
2307 nsearch = ns_simple_core(fr, top, md, box, box_size,
2308 ns->bexcl, ns->simple_aaj,
2309 ngid, ns->ns_buf, put_in_list, ns->bHaveVdW);
2310 }
2311
2312 inc_nrnb(nrnb, eNR_NS, nsearch);
2313
2314 return nsearch;
2315 }
The ultimate molecular dynamics simulation package