Bug Summary

File:gromacs/gmxpreprocess/calc_verletbuf.c
Location:line 998, column 16
Description:Dereference of null pointer

Annotated Source Code

1/*
2 * This file is part of the GROMACS molecular simulation package.
3 *
4 * Copyright (c) 2012,2013,2014, by the GROMACS development team, led by
5 * Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl,
6 * and including many others, as listed in the AUTHORS file in the
7 * top-level source directory and at http://www.gromacs.org.
8 *
9 * GROMACS is free software; you can redistribute it and/or
10 * modify it under the terms of the GNU Lesser General Public License
11 * as published by the Free Software Foundation; either version 2.1
12 * of the License, or (at your option) any later version.
13 *
14 * GROMACS is distributed in the hope that it will be useful,
15 * but WITHOUT ANY WARRANTY; without even the implied warranty of
16 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
17 * Lesser General Public License for more details.
18 *
19 * You should have received a copy of the GNU Lesser General Public
20 * License along with GROMACS; if not, see
21 * http://www.gnu.org/licenses, or write to the Free Software Foundation,
22 * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
23 *
24 * If you want to redistribute modifications to GROMACS, please
25 * consider that scientific software is very special. Version
26 * control is crucial - bugs must be traceable. We will be happy to
27 * consider code for inclusion in the official distribution, but
28 * derived work must not be called official GROMACS. Details are found
29 * in the README & COPYING files - if they are missing, get the
30 * official version at http://www.gromacs.org.
31 *
32 * To help us fund GROMACS development, we humbly ask that you cite
33 * the research papers on the package. Check out http://www.gromacs.org.
34 */
35#ifdef HAVE_CONFIG_H1
36#include <config.h>
37#endif
38
39#include <assert.h>
40#include <math.h>
41#include <stdlib.h>
42
43#include <sys/types.h>
44
45#include "typedefs.h"
46#include "physics.h"
47#include "macros.h"
48#include "gromacs/math/vec.h"
49#include "coulomb.h"
50#include "calc_verletbuf.h"
51#include "../mdlib/nbnxn_consts.h"
52
53#include "gromacs/utility/fatalerror.h"
54#include "gromacs/utility/smalloc.h"
55
56#ifdef GMX_NBNXN_SIMD
57/* The include below sets the SIMD instruction type (precision+width)
58 * for all nbnxn SIMD search and non-bonded kernel code.
59 */
60#ifdef GMX_NBNXN_HALF_WIDTH_SIMD
61#define GMX_USE_HALF_WIDTH_SIMD_HERE
62#endif
63#include "gromacs/simd/simd.h"
64#endif
65
66
67/* The code in this file estimates a pairlist buffer length
68 * given a target energy drift per atom per picosecond.
69 * This is done by estimating the drift given a buffer length.
70 * Ideally we would like to have a tight overestimate of the drift,
71 * but that can be difficult to achieve.
72 *
73 * Significant approximations used:
74 *
75 * Uniform particle density. UNDERESTIMATES the drift by rho_global/rho_local.
76 *
77 * Interactions don't affect particle motion. OVERESTIMATES the drift on longer
78 * time scales. This approximation probably introduces the largest errors.
79 *
80 * Only take one constraint per particle into account: OVERESTIMATES the drift.
81 *
82 * For rotating constraints assume the same functional shape for time scales
83 * where the constraints rotate significantly as the exact expression for
84 * short time scales. OVERESTIMATES the drift on long time scales.
85 *
86 * For non-linear virtual sites use the mass of the lightest constructing atom
87 * to determine the displacement. OVER/UNDERESTIMATES the drift, depending on
88 * the geometry and masses of constructing atoms.
89 *
90 * Note that the formulas for normal atoms and linear virtual sites are exact,
91 * apart from the first two approximations.
92 *
93 * Note that apart from the effect of the above approximations, the actual
94 * drift of the total energy of a system can be order of magnitude smaller
95 * due to cancellation of positive and negative drift for different pairs.
96 */
97
98
99/* Struct for unique atom type for calculating the energy drift.
100 * The atom displacement depends on mass and constraints.
101 * The energy jump for given distance depend on LJ type and q.
102 */
103typedef struct
104{
105 real mass; /* mass */
106 int type; /* type (used for LJ parameters) */
107 real q; /* charge */
108 gmx_bool bConstr; /* constrained, if TRUE, use #DOF=2 iso 3 */
109 real con_mass; /* mass of heaviest atom connected by constraints */
110 real con_len; /* constraint length to the heaviest atom */
111} atom_nonbonded_kinetic_prop_t;
112
113/* Struct for unique atom type for calculating the energy drift.
114 * The atom displacement depends on mass and constraints.
115 * The energy jump for given distance depend on LJ type and q.
116 */
117typedef struct
118{
119 atom_nonbonded_kinetic_prop_t prop; /* non-bonded and kinetic atom prop. */
120 int n; /* #atoms of this type in the system */
121} verletbuf_atomtype_t;
122
123void verletbuf_get_list_setup(gmx_bool bGPU,
124 verletbuf_list_setup_t *list_setup)
125{
126 list_setup->cluster_size_i = NBNXN_CPU_CLUSTER_I_SIZE4;
127
128 if (bGPU)
129 {
130 list_setup->cluster_size_j = NBNXN_GPU_CLUSTER_SIZE8;
131 }
132 else
133 {
134#ifndef GMX_NBNXN_SIMD
135 list_setup->cluster_size_j = NBNXN_CPU_CLUSTER_I_SIZE4;
136#else
137 list_setup->cluster_size_j = GMX_SIMD_REAL_WIDTH;
138#ifdef GMX_NBNXN_SIMD_2XNN
139 /* We assume the smallest cluster size to be on the safe side */
140 list_setup->cluster_size_j /= 2;
141#endif
142#endif
143 }
144}
145
146static gmx_bool
147atom_nonbonded_kinetic_prop_equal(const atom_nonbonded_kinetic_prop_t *prop1,
148 const atom_nonbonded_kinetic_prop_t *prop2)
149{
150 return (prop1->mass == prop2->mass &&
151 prop1->type == prop2->type &&
152 prop1->q == prop2->q &&
153 prop1->bConstr == prop2->bConstr &&
154 prop1->con_mass == prop2->con_mass &&
155 prop1->con_len == prop2->con_len);
156}
157
158static void add_at(verletbuf_atomtype_t **att_p, int *natt_p,
159 const atom_nonbonded_kinetic_prop_t *prop,
160 int nmol)
161{
162 verletbuf_atomtype_t *att;
163 int natt, i;
164
165 if (prop->mass == 0)
166 {
167 /* Ignore massless particles */
168 return;
169 }
170
171 att = *att_p;
172 natt = *natt_p;
173
174 i = 0;
175 while (i < natt && !atom_nonbonded_kinetic_prop_equal(prop, &att[i].prop))
176 {
177 i++;
178 }
179
180 if (i < natt)
181 {
182 att[i].n += nmol;
183 }
184 else
185 {
186 (*natt_p)++;
187 srenew(*att_p, *natt_p)(*att_p) = save_realloc("*att_p", "/home/alexxy/Develop/gromacs/src/gromacs/gmxpreprocess/calc_verletbuf.c"
, 187, (*att_p), (*natt_p), sizeof(*(*att_p)))
;
188 (*att_p)[i].prop = *prop;
189 (*att_p)[i].n = nmol;
190 }
191}
192
193static void get_vsite_masses(const gmx_moltype_t *moltype,
194 const gmx_ffparams_t *ffparams,
195 real *vsite_m,
196 int *n_nonlin_vsite)
197{
198 int ft, i;
199 const t_ilist *il;
200
201 *n_nonlin_vsite = 0;
202
203 /* Check for virtual sites, determine mass from constructing atoms */
204 for (ft = 0; ft < F_NRE; ft++)
205 {
206 if (IS_VSITE(ft)(interaction_function[(ft)].flags & 1<<1))
207 {
208 il = &moltype->ilist[ft];
209
210 for (i = 0; i < il->nr; i += 1+NRAL(ft)(interaction_function[(ft)].nratoms))
211 {
212 const t_iparams *ip;
213 real cam[5], inv_mass, m_aj;
214 int a1, j, aj, coeff;
215
216 ip = &ffparams->iparams[il->iatoms[i]];
217
218 a1 = il->iatoms[i+1];
219
220 if (ft != F_VSITEN)
221 {
222 for (j = 1; j < NRAL(ft)(interaction_function[(ft)].nratoms); j++)
223 {
224 cam[j] = moltype->atoms.atom[il->iatoms[i+1+j]].m;
225 if (cam[j] == 0)
226 {
227 cam[j] = vsite_m[il->iatoms[i+1+j]];
228 }
229 if (cam[j] == 0)
230 {
231 gmx_fatal(FARGS0, "/home/alexxy/Develop/gromacs/src/gromacs/gmxpreprocess/calc_verletbuf.c"
, 231
, "In molecule type '%s' %s construction involves atom %d, which is a virtual site of equal or high complexity. This is not supported.",
232 *moltype->name,
233 interaction_function[ft].longname,
234 il->iatoms[i+1+j]+1);
235 }
236 }
237 }
238
239 switch (ft)
240 {
241 case F_VSITE2:
242 /* Exact */
243 vsite_m[a1] = (cam[1]*cam[2])/(cam[2]*sqr(1-ip->vsite.a) + cam[1]*sqr(ip->vsite.a));
244 break;
245 case F_VSITE3:
246 /* Exact */
247 vsite_m[a1] = (cam[1]*cam[2]*cam[3])/(cam[2]*cam[3]*sqr(1-ip->vsite.a-ip->vsite.b) + cam[1]*cam[3]*sqr(ip->vsite.a) + cam[1]*cam[2]*sqr(ip->vsite.b));
248 break;
249 case F_VSITEN:
250 /* Exact */
251 inv_mass = 0;
252 for (j = 0; j < 3*ip->vsiten.n; j += 3)
253 {
254 aj = il->iatoms[i+j+2];
255 coeff = ip[il->iatoms[i+j]].vsiten.a;
256 if (moltype->atoms.atom[aj].ptype == eptVSite)
257 {
258 m_aj = vsite_m[aj];
259 }
260 else
261 {
262 m_aj = moltype->atoms.atom[aj].m;
263 }
264 if (m_aj <= 0)
265 {
266 gmx_incons("The mass of a vsiten constructing atom is <= 0")_gmx_error("incons", "The mass of a vsiten constructing atom is <= 0"
, "/home/alexxy/Develop/gromacs/src/gromacs/gmxpreprocess/calc_verletbuf.c"
, 266)
;
267 }
268 inv_mass += coeff*coeff/m_aj;
269 }
270 vsite_m[a1] = 1/inv_mass;
271 break;
272 default:
273 /* Use the mass of the lightest constructing atom.
274 * This is an approximation.
275 * If the distance of the virtual site to the
276 * constructing atom is less than all distances
277 * between constructing atoms, this is a safe
278 * over-estimate of the displacement of the vsite.
279 * This condition holds for all H mass replacement
280 * vsite constructions, except for SP2/3 groups.
281 * In SP3 groups one H will have a F_VSITE3
282 * construction, so even there the total drift
283 * estimate shouldn't be far off.
284 */
285 assert(j >= 1)((void) (0));
286 vsite_m[a1] = cam[1];
287 for (j = 2; j < NRAL(ft)(interaction_function[(ft)].nratoms); j++)
288 {
289 vsite_m[a1] = min(vsite_m[a1], cam[j])(((vsite_m[a1]) < (cam[j])) ? (vsite_m[a1]) : (cam[j]) );
290 }
291 (*n_nonlin_vsite)++;
292 break;
293 }
294 if (gmx_debug_at)
295 {
296 fprintf(debug, "atom %4d %-20s mass %6.3f\n",
297 a1, interaction_function[ft].longname, vsite_m[a1]);
298 }
299 }
300 }
301 }
302}
303
304static void get_verlet_buffer_atomtypes(const gmx_mtop_t *mtop,
305 verletbuf_atomtype_t **att_p,
306 int *natt_p,
307 int *n_nonlin_vsite)
308{
309 verletbuf_atomtype_t *att;
310 int natt;
311 int mb, nmol, ft, i, a1, a2, a3, a;
312 const t_atoms *atoms;
313 const t_ilist *il;
314 const t_iparams *ip;
315 atom_nonbonded_kinetic_prop_t *prop;
316 real *vsite_m;
317 int n_nonlin_vsite_mol;
318
319 att = NULL((void*)0);
5
Null pointer value stored to 'att'
320 natt = 0;
321
322 if (n_nonlin_vsite != NULL((void*)0))
6
Assuming 'n_nonlin_vsite' is equal to null
7
Taking false branch
323 {
324 *n_nonlin_vsite = 0;
325 }
326
327 for (mb = 0; mb < mtop->nmolblock; mb++)
8
Loop condition is false. Execution continues on line 421
328 {
329 nmol = mtop->molblock[mb].nmol;
330
331 atoms = &mtop->moltype[mtop->molblock[mb].type].atoms;
332
333 /* Check for constraints, as they affect the kinetic energy.
334 * For virtual sites we need the masses and geometry of
335 * the constructing atoms to determine their velocity distribution.
336 */
337 snew(prop, atoms->nr)(prop) = save_calloc("prop", "/home/alexxy/Develop/gromacs/src/gromacs/gmxpreprocess/calc_verletbuf.c"
, 337, (atoms->nr), sizeof(*(prop)))
;
338 snew(vsite_m, atoms->nr)(vsite_m) = save_calloc("vsite_m", "/home/alexxy/Develop/gromacs/src/gromacs/gmxpreprocess/calc_verletbuf.c"
, 338, (atoms->nr), sizeof(*(vsite_m)))
;
339
340 for (ft = F_CONSTR; ft <= F_CONSTRNC; ft++)
341 {
342 il = &mtop->moltype[mtop->molblock[mb].type].ilist[ft];
343
344 for (i = 0; i < il->nr; i += 1+NRAL(ft)(interaction_function[(ft)].nratoms))
345 {
346 ip = &mtop->ffparams.iparams[il->iatoms[i]];
347 a1 = il->iatoms[i+1];
348 a2 = il->iatoms[i+2];
349 if (atoms->atom[a2].m > prop[a1].con_mass)
350 {
351 prop[a1].con_mass = atoms->atom[a2].m;
352 prop[a1].con_len = ip->constr.dA;
353 }
354 if (atoms->atom[a1].m > prop[a2].con_mass)
355 {
356 prop[a2].con_mass = atoms->atom[a1].m;
357 prop[a2].con_len = ip->constr.dA;
358 }
359 }
360 }
361
362 il = &mtop->moltype[mtop->molblock[mb].type].ilist[F_SETTLE];
363
364 for (i = 0; i < il->nr; i += 1+NRAL(F_SETTLE)(interaction_function[(F_SETTLE)].nratoms))
365 {
366 ip = &mtop->ffparams.iparams[il->iatoms[i]];
367 a1 = il->iatoms[i+1];
368 a2 = il->iatoms[i+2];
369 a3 = il->iatoms[i+3];
370 /* Usually the mass of a1 (usually oxygen) is larger than a2/a3.
371 * If this is not the case, we overestimate the displacement,
372 * which leads to a larger buffer (ok since this is an exotic case).
373 */
374 prop[a1].con_mass = atoms->atom[a2].m;
375 prop[a1].con_len = ip->settle.doh;
376
377 prop[a2].con_mass = atoms->atom[a1].m;
378 prop[a2].con_len = ip->settle.doh;
379
380 prop[a3].con_mass = atoms->atom[a1].m;
381 prop[a3].con_len = ip->settle.doh;
382 }
383
384 get_vsite_masses(&mtop->moltype[mtop->molblock[mb].type],
385 &mtop->ffparams,
386 vsite_m,
387 &n_nonlin_vsite_mol);
388 if (n_nonlin_vsite != NULL((void*)0))
389 {
390 *n_nonlin_vsite += nmol*n_nonlin_vsite_mol;
391 }
392
393 for (a = 0; a < atoms->nr; a++)
394 {
395 if (atoms->atom[a].ptype == eptVSite)
396 {
397 prop[a].mass = vsite_m[a];
398 }
399 else
400 {
401 prop[a].mass = atoms->atom[a].m;
402 }
403 prop[a].type = atoms->atom[a].type;
404 prop[a].q = atoms->atom[a].q;
405 /* We consider an atom constrained, #DOF=2, when it is
406 * connected with constraints to (at least one) atom with
407 * a mass of more than 0.4x its own mass. This is not a critical
408 * parameter, since with roughly equal masses the unconstrained
409 * and constrained displacement will not differ much (and both
410 * overestimate the displacement).
411 */
412 prop[a].bConstr = (prop[a].con_mass > 0.4*prop[a].mass);
413
414 add_at(&att, &natt, &prop[a], nmol);
415 }
416
417 sfree(vsite_m)save_free("vsite_m", "/home/alexxy/Develop/gromacs/src/gromacs/gmxpreprocess/calc_verletbuf.c"
, 417, (vsite_m))
;
418 sfree(prop)save_free("prop", "/home/alexxy/Develop/gromacs/src/gromacs/gmxpreprocess/calc_verletbuf.c"
, 418, (prop))
;
419 }
420
421 if (gmx_debug_at)
9
Assuming 'gmx_debug_at' is 0
10
Taking false branch
422 {
423 for (a = 0; a < natt; a++)
424 {
425 fprintf(debug, "type %d: m %5.2f t %d q %6.3f con %d con_m %5.3f con_l %5.3f n %d\n",
426 a, att[a].prop.mass, att[a].prop.type, att[a].prop.q,
427 att[a].prop.bConstr, att[a].prop.con_mass, att[a].prop.con_len,
428 att[a].n);
429 }
430 }
431
432 *att_p = att;
11
Null pointer value stored to 'att'
433 *natt_p = natt;
434}
435
436/* This function computes two components of the estimate of the variance
437 * in the displacement of one atom in a system of two constrained atoms.
438 * Returns in sigma2_2d the variance due to rotation of the constrained
439 * atom around the atom to which it constrained.
440 * Returns in sigma2_3d the variance due to displacement of the COM
441 * of the whole system of the two constrained atoms.
442 *
443 * Note that we only take a single constraint (the one to the heaviest atom)
444 * into account. If an atom has multiple constraints, this will result in
445 * an overestimate of the displacement, which gives a larger drift and buffer.
446 */
447static void constrained_atom_sigma2(real kT_fac,
448 const atom_nonbonded_kinetic_prop_t *prop,
449 real *sigma2_2d,
450 real *sigma2_3d)
451{
452 real sigma2_rot;
453 real com_dist;
454 real sigma2_rel;
455 real scale;
456
457 /* Here we decompose the motion of a constrained atom into two
458 * components: rotation around the COM and translation of the COM.
459 */
460
461 /* Determine the variance for the displacement of the rotational mode */
462 sigma2_rot = kT_fac/(prop->mass*(prop->mass + prop->con_mass)/prop->con_mass);
463
464 /* The distance from the atom to the COM, i.e. the rotational arm */
465 com_dist = prop->con_len*prop->con_mass/(prop->mass + prop->con_mass);
466
467 /* The variance relative to the arm */
468 sigma2_rel = sigma2_rot/(com_dist*com_dist);
469 /* At 6 the scaling formula has slope 0,
470 * so we keep sigma2_2d constant after that.
471 */
472 if (sigma2_rel < 6)
473 {
474 /* A constrained atom rotates around the atom it is constrained to.
475 * This results in a smaller linear displacement than for a free atom.
476 * For a perfectly circular displacement, this lowers the displacement
477 * by: 1/arcsin(arc_length)
478 * and arcsin(x) = 1 + x^2/6 + ...
479 * For sigma2_rel<<1 the displacement distribution is erfc
480 * (exact formula is provided below). For larger sigma, it is clear
481 * that the displacement can't be larger than 2*com_dist.
482 * It turns out that the distribution becomes nearly uniform.
483 * For intermediate sigma2_rel, scaling down sigma with the third
484 * order expansion of arcsin with argument sigma_rel turns out
485 * to give a very good approximation of the distribution and variance.
486 * Even for larger values, the variance is only slightly overestimated.
487 * Note that the most relevant displacements are in the long tail.
488 * This rotation approximation always overestimates the tail (which
489 * runs to infinity, whereas it should be <= 2*com_dist).
490 * Thus we always overestimate the drift and the buffer size.
491 */
492 scale = 1/(1 + sigma2_rel/6);
493 *sigma2_2d = sigma2_rot*scale*scale;
494 }
495 else
496 {
497 /* sigma_2d is set to the maximum given by the scaling above.
498 * For large sigma2 the real displacement distribution is close
499 * to uniform over -2*con_len to 2*com_dist.
500 * Our erfc with sigma_2d=sqrt(1.5)*com_dist (which means the sigma
501 * of the erfc output distribution is con_dist) overestimates
502 * the variance and additionally has a long tail. This means
503 * we have a (safe) overestimation of the drift.
504 */
505 *sigma2_2d = 1.5*com_dist*com_dist;
506 }
507
508 /* The constrained atom also moves (in 3D) with the COM of both atoms */
509 *sigma2_3d = kT_fac/(prop->mass + prop->con_mass);
510}
511
512static void get_atom_sigma2(real kT_fac,
513 const atom_nonbonded_kinetic_prop_t *prop,
514 real *sigma2_2d,
515 real *sigma2_3d)
516{
517 if (prop->bConstr)
518 {
519 /* Complicated constraint calculation in a separate function */
520 constrained_atom_sigma2(kT_fac, prop, sigma2_2d, sigma2_3d);
521 }
522 else
523 {
524 /* Unconstrained atom: trivial */
525 *sigma2_2d = 0;
526 *sigma2_3d = kT_fac/prop->mass;
527 }
528}
529
530static void approx_2dof(real s2, real x, real *shift, real *scale)
531{
532 /* A particle with 1 DOF constrained has 2 DOFs instead of 3.
533 * This code is also used for particles with multiple constraints,
534 * in which case we overestimate the displacement.
535 * The 2DOF distribution is sqrt(pi/2)*erfc(r/(sqrt(2)*s))/(2*s).
536 * We approximate this with scale*Gaussian(s,r+shift),
537 * by matching the distribution value and derivative at x.
538 * This is a tight overestimate for all r>=0 at any s and x.
539 */
540 real ex, er;
541
542 ex = exp(-x*x/(2*s2));
543 er = gmx_erfc(x/sqrt(2*s2))gmx_erfcf(x/sqrt(2*s2));
544
545 *shift = -x + sqrt(2*s2/M_PI3.14159265358979323846)*ex/er;
546 *scale = 0.5*M_PI3.14159265358979323846*exp(ex*ex/(M_PI3.14159265358979323846*er*er))*er;
547}
548
549static real ener_drift(const verletbuf_atomtype_t *att, int natt,
550 const gmx_ffparams_t *ffp,
551 real kT_fac,
552 real md1_ljd, real d2_ljd, real md3_ljd,
553 real md1_ljr, real d2_ljr, real md3_ljr,
554 real md1_el, real d2_el,
555 real r_buffer,
556 real rlist, real boxvol)
557{
558 double drift_tot, pot1, pot2, pot3, pot;
559 int i, j;
560 real s2i_2d, s2i_3d, s2j_2d, s2j_3d, s2, s;
561 int ti, tj;
562 real md1, d2, md3;
563 real sc_fac, rsh, rsh2;
564 double c_exp, c_erfc;
565
566 drift_tot = 0;
567
568 /* Loop over the different atom type pairs */
569 for (i = 0; i < natt; i++)
570 {
571 get_atom_sigma2(kT_fac, &att[i].prop, &s2i_2d, &s2i_3d);
572 ti = att[i].prop.type;
573
574 for (j = i; j < natt; j++)
575 {
576 get_atom_sigma2(kT_fac, &att[j].prop, &s2j_2d, &s2j_3d);
577 tj = att[j].prop.type;
578
579 /* Add up the up to four independent variances */
580 s2 = s2i_2d + s2i_3d + s2j_2d + s2j_3d;
581
582 /* Note that attractive and repulsive potentials for individual
583 * pairs will partially cancel.
584 */
585 /* -dV/dr at the cut-off for LJ + Coulomb */
586 md1 =
587 md1_ljd*ffp->iparams[ti*ffp->atnr+tj].lj.c6 +
588 md1_ljr*ffp->iparams[ti*ffp->atnr+tj].lj.c12 +
589 md1_el*att[i].prop.q*att[j].prop.q;
590
591 /* d2V/dr2 at the cut-off for LJ + Coulomb */
592 d2 =
593 d2_ljd*ffp->iparams[ti*ffp->atnr+tj].lj.c6 +
594 d2_ljr*ffp->iparams[ti*ffp->atnr+tj].lj.c12 +
595 d2_el*att[i].prop.q*att[j].prop.q;
596
597 /* -d3V/dr3 at the cut-off for LJ, we neglect Coulomb */
598 md3 =
599 md3_ljd*ffp->iparams[ti*ffp->atnr+tj].lj.c6 +
600 md3_ljr*ffp->iparams[ti*ffp->atnr+tj].lj.c12;
601
602 rsh = r_buffer;
603 sc_fac = 1.0;
604 /* For constraints: adapt r and scaling for the Gaussian */
605 if (att[i].prop.bConstr)
606 {
607 real sh, sc;
608
609 approx_2dof(s2i_2d, r_buffer*s2i_2d/s2, &sh, &sc);
610 rsh += sh;
611 sc_fac *= sc;
612 }
613 if (att[j].prop.bConstr)
614 {
615 real sh, sc;
616
617 approx_2dof(s2j_2d, r_buffer*s2j_2d/s2, &sh, &sc);
618 rsh += sh;
619 sc_fac *= sc;
620 }
621
622 /* Exact contribution of an atom pair with Gaussian displacement
623 * with sigma s to the energy drift for a potential with
624 * derivative -md and second derivative dd at the cut-off.
625 * The only catch is that for potentials that change sign
626 * near the cut-off there could be an unlucky compensation
627 * of positive and negative energy drift.
628 * Such potentials are extremely rare though.
629 *
630 * Note that pot has unit energy*length, as the linear
631 * atom density still needs to be put in.
632 */
633 c_exp = exp(-rsh*rsh/(2*s2))/sqrt(2*M_PI3.14159265358979323846);
634 c_erfc = 0.5*gmx_erfc(rsh/(sqrt(2*s2)))gmx_erfcf(rsh/(sqrt(2*s2)));
635 s = sqrt(s2);
636 rsh2 = rsh*rsh;
637
638 pot1 = sc_fac*
639 md1/2*((rsh2 + s2)*c_erfc - rsh*s*c_exp);
640 pot2 = sc_fac*
641 d2/6*(s*(rsh2 + 2*s2)*c_exp - rsh*(rsh2 + 3*s2)*c_erfc);
642 pot3 =
643 md3/24*((rsh2*rsh2 + 6*rsh2*s2 + 3*s2*s2)*c_erfc - rsh*s*(rsh2 + 5*s2)*c_exp);
644 pot = pot1 + pot2 + pot3;
645
646 if (gmx_debug_at)
647 {
648 fprintf(debug, "n %d %d d s %.3f %.3f %.3f %.3f con %d -d1 %8.1e d2 %8.1e -d3 %8.1e pot1 %8.1e pot2 %8.1e pot3 %8.1e pot %8.1e\n",
649 att[i].n, att[j].n,
650 sqrt(s2i_2d), sqrt(s2i_3d),
651 sqrt(s2j_2d), sqrt(s2j_3d),
652 att[i].prop.bConstr+att[j].prop.bConstr,
653 md1, d2, md3,
654 pot1, pot2, pot3, pot);
655 }
656
657 /* Multiply by the number of atom pairs */
658 if (j == i)
659 {
660 pot *= (double)att[i].n*(att[i].n - 1)/2;
661 }
662 else
663 {
664 pot *= (double)att[i].n*att[j].n;
665 }
666 /* We need the line density to get the energy drift of the system.
667 * The effective average r^2 is close to (rlist+sigma)^2.
668 */
669 pot *= 4*M_PI3.14159265358979323846*sqr(rlist + s)/boxvol;
670
671 /* Add the unsigned drift to avoid cancellation of errors */
672 drift_tot += fabs(pot);
673 }
674 }
675
676 return drift_tot;
677}
678
679static real surface_frac(int cluster_size, real particle_distance, real rlist)
680{
681 real d, area_rel;
682
683 if (rlist < 0.5*particle_distance)
684 {
685 /* We have non overlapping spheres */
686 return 1.0;
687 }
688
689 /* Half the inter-particle distance relative to rlist */
690 d = 0.5*particle_distance/rlist;
691
692 /* Determine the area of the surface at distance rlist to the closest
693 * particle, relative to surface of a sphere of radius rlist.
694 * The formulas below assume close to cubic cells for the pair search grid,
695 * which the pair search code tries to achieve.
696 * Note that in practice particle distances will not be delta distributed,
697 * but have some spread, often involving shorter distances,
698 * as e.g. O-H bonds in a water molecule. Thus the estimates below will
699 * usually be slightly too high and thus conservative.
700 */
701 switch (cluster_size)
702 {
703 case 1:
704 /* One particle: trivial */
705 area_rel = 1.0;
706 break;
707 case 2:
708 /* Two particles: two spheres at fractional distance 2*a */
709 area_rel = 1.0 + d;
710 break;
711 case 4:
712 /* We assume a perfect, symmetric tetrahedron geometry.
713 * The surface around a tetrahedron is too complex for a full
714 * analytical solution, so we use a Taylor expansion.
715 */
716 area_rel = (1.0 + 1/M_PI3.14159265358979323846*(6*acos(1/sqrt(3))*d +
717 sqrt(3)*d*d*(1.0 +
718 5.0/18.0*d*d +
719 7.0/45.0*d*d*d*d +
720 83.0/756.0*d*d*d*d*d*d)));
721 break;
722 default:
723 gmx_incons("surface_frac called with unsupported cluster_size")_gmx_error("incons", "surface_frac called with unsupported cluster_size"
, "/home/alexxy/Develop/gromacs/src/gromacs/gmxpreprocess/calc_verletbuf.c"
, 723)
;
724 area_rel = 1.0;
725 }
726
727 return area_rel/cluster_size;
728}
729
730/* Returns the negative of the third derivative of a potential r^-p
731 * with a force-switch function, evaluated at the cut-off rc.
732 */
733static real md3_force_switch(real p, real rswitch, real rc)
734{
735 /* The switched force function is:
736 * p*r^-(p+1) + a*(r - rswitch)^2 + b*(r - rswitch)^3
737 */
738 real a, b;
739 real md3_pot, md3_sw;
740
741 a = -((p + 4)*rc - (p + 1)*rswitch)/(pow(rc, p+2)*pow(rc-rswitch, 2));
742 b = ((p + 3)*rc - (p + 1)*rswitch)/(pow(rc, p+2)*pow(rc-rswitch, 3));
743
744 md3_pot = (p + 2)*(p + 1)*p*pow(rc, p+3);
745 md3_sw = 2*a + 6*b*(rc - rswitch);
746
747 return md3_pot + md3_sw;
748}
749
750void calc_verlet_buffer_size(const gmx_mtop_t *mtop, real boxvol,
751 const t_inputrec *ir,
752 real reference_temperature,
753 const verletbuf_list_setup_t *list_setup,
754 int *n_nonlin_vsite,
755 real *rlist)
756{
757 double resolution;
758 char *env;
759
760 real particle_distance;
761 real nb_clust_frac_pairs_not_in_list_at_cutoff;
762
763 verletbuf_atomtype_t *att = NULL((void*)0);
764 int natt = -1, i;
765 double reppow;
766 real md1_ljd, d2_ljd, md3_ljd;
767 real md1_ljr, d2_ljr, md3_ljr;
768 real md1_el, d2_el;
769 real elfac;
770 real kT_fac, mass_min;
771 int ib0, ib1, ib;
772 real rb, rl;
773 real drift;
774
775 if (reference_temperature < 0)
1
Taking false branch
776 {
777 if (EI_MD(ir->eI)((ir->eI) == eiMD || ((ir->eI) == eiVV || (ir->eI) ==
eiVVAK))
&& ir->etc == etcNO)
778 {
779 /* This case should be handled outside calc_verlet_buffer_size */
780 gmx_incons("calc_verlet_buffer_size called with an NVE ensemble and reference_temperature < 0")_gmx_error("incons", "calc_verlet_buffer_size called with an NVE ensemble and reference_temperature < 0"
, "/home/alexxy/Develop/gromacs/src/gromacs/gmxpreprocess/calc_verletbuf.c"
, 780)
;
781 }
782
783 /* We use the maximum temperature with multiple T-coupl groups.
784 * We could use a per particle temperature, but since particles
785 * interact, this might underestimate the buffer size.
786 */
787 reference_temperature = 0;
788 for (i = 0; i < ir->opts.ngtc; i++)
789 {
790 if (ir->opts.tau_t[i] >= 0)
791 {
792 reference_temperature = max(reference_temperature,(((reference_temperature) > (ir->opts.ref_t[i])) ? (reference_temperature
) : (ir->opts.ref_t[i]) )
793 ir->opts.ref_t[i])(((reference_temperature) > (ir->opts.ref_t[i])) ? (reference_temperature
) : (ir->opts.ref_t[i]) )
;
794 }
795 }
796 }
797
798 /* Resolution of the buffer size */
799 resolution = 0.001;
800
801 env = getenv("GMX_VERLET_BUFFER_RES");
802 if (env != NULL((void*)0))
2
Assuming 'env' is equal to null
3
Taking false branch
803 {
804 sscanf(env, "%lf", &resolution);
805 }
806
807 /* In an atom wise pair-list there would be no pairs in the list
808 * beyond the pair-list cut-off.
809 * However, we use a pair-list of groups vs groups of atoms.
810 * For groups of 4 atoms, the parallelism of SSE instructions, only
811 * 10% of the atoms pairs are not in the list just beyond the cut-off.
812 * As this percentage increases slowly compared to the decrease of the
813 * Gaussian displacement distribution over this range, we can simply
814 * reduce the drift by this fraction.
815 * For larger groups, e.g. of 8 atoms, this fraction will be lower,
816 * so then buffer size will be on the conservative (large) side.
817 *
818 * Note that the formulas used here do not take into account
819 * cancellation of errors which could occur by missing both
820 * attractive and repulsive interactions.
821 *
822 * The only major assumption is homogeneous particle distribution.
823 * For an inhomogeneous system, such as a liquid-vapor system,
824 * the buffer will be underestimated. The actual energy drift
825 * will be higher by the factor: local/homogeneous particle density.
826 *
827 * The results of this estimate have been checked againt simulations.
828 * In most cases the real drift differs by less than a factor 2.
829 */
830
831 /* Worst case assumption: HCP packing of particles gives largest distance */
832 particle_distance = pow(boxvol*sqrt(2)/mtop->natoms, 1.0/3.0);
833
834 get_verlet_buffer_atomtypes(mtop, &att, &natt, n_nonlin_vsite);
4
Calling 'get_verlet_buffer_atomtypes'
12
Returning from 'get_verlet_buffer_atomtypes'
835 assert(att != NULL && natt >= 0)((void) (0));
836
837 if (debug)
13
Assuming 'debug' is null
14
Taking false branch
838 {
839 fprintf(debug, "particle distance assuming HCP packing: %f nm\n",
840 particle_distance);
841 fprintf(debug, "energy drift atom types: %d\n", natt);
842 }
843
844 reppow = mtop->ffparams.reppow;
845 md1_ljd = 0;
846 d2_ljd = 0;
847 md3_ljd = 0;
848 md1_ljr = 0;
849 d2_ljr = 0;
850 md3_ljr = 0;
851 if (ir->vdwtype == evdwCUT)
15
Taking true branch
852 {
853 real sw_range, md3_pswf;
854
855 switch (ir->vdw_modifier)
16
Control jumps to 'case eintmodPOTSHIFT:' at line 858
856 {
857 case eintmodNONE:
858 case eintmodPOTSHIFT:
859 /* -dV/dr of -r^-6 and r^-reppow */
860 md1_ljd = -6*pow(ir->rvdw, -7.0);
861 md1_ljr = reppow*pow(ir->rvdw, -(reppow+1));
862 /* The contribution of the higher derivatives is negligible */
863 break;
17
Execution continues on line 903
864 case eintmodFORCESWITCH:
865 /* At the cut-off: V=V'=V''=0, so we use only V''' */
866 md3_ljd = -md3_force_switch(6.0, ir->rvdw_switch, ir->rvdw);
867 md3_ljr = md3_force_switch(reppow, ir->rvdw_switch, ir->rvdw);
868 break;
869 case eintmodPOTSWITCH:
870 /* At the cut-off: V=V'=V''=0.
871 * V''' is given by the original potential times
872 * the third derivative of the switch function.
873 */
874 sw_range = ir->rvdw - ir->rvdw_switch;
875 md3_pswf = 60.0*pow(sw_range, -3.0);
876
877 md3_ljd = -pow(ir->rvdw, -6.0 )*md3_pswf;
878 md3_ljr = pow(ir->rvdw, -reppow)*md3_pswf;
879 break;
880 default:
881 gmx_incons("Unimplemented VdW modifier")_gmx_error("incons", "Unimplemented VdW modifier", "/home/alexxy/Develop/gromacs/src/gromacs/gmxpreprocess/calc_verletbuf.c"
, 881)
;
882 }
883 }
884 else if (EVDW_PME(ir->vdwtype)((ir->vdwtype) == evdwPME))
885 {
886 real b, r, br, br2, br4, br6;
887 b = calc_ewaldcoeff_lj(ir->rvdw, ir->ewald_rtol_lj);
888 r = ir->rvdw;
889 br = b*r;
890 br2 = br*br;
891 br4 = br2*br2;
892 br6 = br4*br2;
893 /* -dV/dr of g(br)*r^-6 [where g(x) = exp(-x^2)(1+x^2+x^4/2), see LJ-PME equations in manual] and r^-reppow */
894 md1_ljd = -exp(-br2)*(br6 + 3.0*br4 + 6.0*br2 + 6.0)*pow(r, -7.0);
895 md1_ljr = reppow*pow(r, -(reppow+1));
896 /* The contribution of the higher derivatives is negligible */
897 }
898 else
899 {
900 gmx_fatal(FARGS0, "/home/alexxy/Develop/gromacs/src/gromacs/gmxpreprocess/calc_verletbuf.c"
, 900
, "Energy drift calculation is only implemented for plain cut-off Lennard-Jones interactions");
901 }
902
903 elfac = ONE_4PI_EPS0((332.0636930*(4.184))*0.1)/ir->epsilon_r;
904
905 /* Determine md=-dV/dr and dd=d^2V/dr^2 */
906 md1_el = 0;
907 d2_el = 0;
908 if (ir->coulombtype == eelCUT || EEL_RF(ir->coulombtype)((ir->coulombtype) == eelRF || (ir->coulombtype) == eelGRF
|| (ir->coulombtype) == eelRF_NEC || (ir->coulombtype)
== eelRF_ZERO )
)
909 {
910 real eps_rf, k_rf;
911
912 if (ir->coulombtype == eelCUT)
18
Taking true branch
913 {
914 eps_rf = 1;
915 k_rf = 0;
916 }
917 else
918 {
919 eps_rf = ir->epsilon_rf/ir->epsilon_r;
920 if (eps_rf != 0)
921 {
922 k_rf = pow(ir->rcoulomb, -3.0)*(eps_rf - ir->epsilon_r)/(2*eps_rf + ir->epsilon_r);
923 }
924 else
925 {
926 /* epsilon_rf = infinity */
927 k_rf = 0.5*pow(ir->rcoulomb, -3.0);
928 }
929 }
930
931 if (eps_rf > 0)
19
Taking false branch
932 {
933 md1_el = elfac*(pow(ir->rcoulomb, -2.0) - 2*k_rf*ir->rcoulomb);
934 }
935 d2_el = elfac*(2*pow(ir->rcoulomb, -3.0) + 2*k_rf);
936 }
937 else if (EEL_PME(ir->coulombtype)((ir->coulombtype) == eelPME || (ir->coulombtype) == eelPMESWITCH
|| (ir->coulombtype) == eelPMEUSER || (ir->coulombtype
) == eelPMEUSERSWITCH || (ir->coulombtype) == eelP3M_AD)
|| ir->coulombtype == eelEWALD)
938 {
939 real b, rc, br;
940
941 b = calc_ewaldcoeff_q(ir->rcoulomb, ir->ewald_rtol);
942 rc = ir->rcoulomb;
943 br = b*rc;
944 md1_el = elfac*(b*exp(-br*br)*M_2_SQRTPI1.12837916709551257390/rc + gmx_erfc(br)gmx_erfcf(br)/(rc*rc));
945 d2_el = elfac/(rc*rc)*(2*b*(1 + br*br)*exp(-br*br)*M_2_SQRTPI1.12837916709551257390 + 2*gmx_erfc(br)gmx_erfcf(br)/rc);
946 }
947 else
948 {
949 gmx_fatal(FARGS0, "/home/alexxy/Develop/gromacs/src/gromacs/gmxpreprocess/calc_verletbuf.c"
, 949
, "Energy drift calculation is only implemented for Reaction-Field and Ewald electrostatics");
950 }
951
952 /* Determine the variance of the atomic displacement
953 * over nstlist-1 steps: kT_fac
954 * For inertial dynamics (not Brownian dynamics) the mass factor
955 * is not included in kT_fac, it is added later.
956 */
957 if (ir->eI == eiBD)
20
Taking false branch
958 {
959 /* Get the displacement distribution from the random component only.
960 * With accurate integration the systematic (force) displacement
961 * should be negligible (unless nstlist is extremely large, which
962 * you wouldn't do anyhow).
963 */
964 kT_fac = 2*BOLTZ(((1.380658e-23)*(6.0221367e23))/(1e3))*reference_temperature*(ir->nstlist-1)*ir->delta_t;
965 if (ir->bd_fric > 0)
966 {
967 /* This is directly sigma^2 of the displacement */
968 kT_fac /= ir->bd_fric;
969
970 /* Set the masses to 1 as kT_fac is the full sigma^2,
971 * but we divide by m in ener_drift().
972 */
973 for (i = 0; i < natt; i++)
974 {
975 att[i].prop.mass = 1;
976 }
977 }
978 else
979 {
980 real tau_t;
981
982 /* Per group tau_t is not implemented yet, use the maximum */
983 tau_t = ir->opts.tau_t[0];
984 for (i = 1; i < ir->opts.ngtc; i++)
985 {
986 tau_t = max(tau_t, ir->opts.tau_t[i])(((tau_t) > (ir->opts.tau_t[i])) ? (tau_t) : (ir->opts
.tau_t[i]) )
;
987 }
988
989 kT_fac *= tau_t;
990 /* This kT_fac needs to be divided by the mass to get sigma^2 */
991 }
992 }
993 else
994 {
995 kT_fac = BOLTZ(((1.380658e-23)*(6.0221367e23))/(1e3))*reference_temperature*sqr((ir->nstlist-1)*ir->delta_t);
996 }
997
998 mass_min = att[0].prop.mass;
21
Dereference of null pointer
999 for (i = 1; i < natt; i++)
1000 {
1001 mass_min = min(mass_min, att[i].prop.mass)(((mass_min) < (att[i].prop.mass)) ? (mass_min) : (att[i].
prop.mass) )
;
1002 }
1003
1004 if (debug)
1005 {
1006 fprintf(debug, "md1_ljd %9.2e d2_ljd %9.2e md3_ljd %9.2e\n", md1_ljd, d2_ljd, md3_ljd);
1007 fprintf(debug, "md1_ljr %9.2e d2_ljr %9.2e md3_ljr %9.2e\n", md1_ljr, d2_ljr, md3_ljr);
1008 fprintf(debug, "md1_el %9.2e d2_el %9.2e\n", md1_el, d2_el);
1009 fprintf(debug, "sqrt(kT_fac) %f\n", sqrt(kT_fac));
1010 fprintf(debug, "mass_min %f\n", mass_min);
1011 }
1012
1013 /* Search using bisection */
1014 ib0 = -1;
1015 /* The drift will be neglible at 5 times the max sigma */
1016 ib1 = (int)(5*2*sqrt(kT_fac/mass_min)/resolution) + 1;
1017 while (ib1 - ib0 > 1)
1018 {
1019 ib = (ib0 + ib1)/2;
1020 rb = ib*resolution;
1021 rl = max(ir->rvdw, ir->rcoulomb)(((ir->rvdw) > (ir->rcoulomb)) ? (ir->rvdw) : (ir
->rcoulomb) )
+ rb;
1022
1023 /* Calculate the average energy drift at the last step
1024 * of the nstlist steps at which the pair-list is used.
1025 */
1026 drift = ener_drift(att, natt, &mtop->ffparams,
1027 kT_fac,
1028 md1_ljd, d2_ljd, md3_ljd,
1029 md1_ljr, d2_ljr, md3_ljr,
1030 md1_el, d2_el,
1031 rb,
1032 rl, boxvol);
1033
1034 /* Correct for the fact that we are using a Ni x Nj particle pair list
1035 * and not a 1 x 1 particle pair list. This reduces the drift.
1036 */
1037 /* We don't have a formula for 8 (yet), use 4 which is conservative */
1038 nb_clust_frac_pairs_not_in_list_at_cutoff =
1039 surface_frac(min(list_setup->cluster_size_i, 4)(((list_setup->cluster_size_i) < (4)) ? (list_setup->
cluster_size_i) : (4) )
,
1040 particle_distance, rl)*
1041 surface_frac(min(list_setup->cluster_size_j, 4)(((list_setup->cluster_size_j) < (4)) ? (list_setup->
cluster_size_j) : (4) )
,
1042 particle_distance, rl);
1043 drift *= nb_clust_frac_pairs_not_in_list_at_cutoff;
1044
1045 /* Convert the drift to drift per unit time per atom */
1046 drift /= ir->nstlist*ir->delta_t*mtop->natoms;
1047
1048 if (debug)
1049 {
1050 fprintf(debug, "ib %3d %3d %3d rb %.3f %dx%d fac %.3f drift %f\n",
1051 ib0, ib, ib1, rb,
1052 list_setup->cluster_size_i, list_setup->cluster_size_j,
1053 nb_clust_frac_pairs_not_in_list_at_cutoff,
1054 drift);
1055 }
1056
1057 if (fabs(drift) > ir->verletbuf_tol)
1058 {
1059 ib0 = ib;
1060 }
1061 else
1062 {
1063 ib1 = ib;
1064 }
1065 }
1066
1067 sfree(att)save_free("att", "/home/alexxy/Develop/gromacs/src/gromacs/gmxpreprocess/calc_verletbuf.c"
, 1067, (att))
;
1068
1069 *rlist = max(ir->rvdw, ir->rcoulomb)(((ir->rvdw) > (ir->rcoulomb)) ? (ir->rvdw) : (ir
->rcoulomb) )
+ ib1*resolution;
1070}