2 * This file is part of the GROMACS molecular simulation package.
4 * Copyright (c) 1991-2000, University of Groningen, The Netherlands.
5 * Copyright (c) 2001-2008, The GROMACS development team.
6 * Copyright (c) 2013, 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.
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.
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.
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.
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.
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.
50 #include "gromacs/fileio/pdbio.h"
56 #include "gmx_fatal.h"
57 #include "mtop_util.h"
62 #include "gromacs/utility/gmxmpi.h"
66 # include "genborn_sse2_double.h"
67 # include "genborn_allvsall_sse2_double.h"
69 # include "genborn_sse2_single.h"
70 # include "genborn_allvsall_sse2_single.h"
71 # endif /* GMX_DOUBLE */
72 #endif /* SSE or AVX present */
74 #include "genborn_allvsall.h"
76 /*#define DISABLE_SSE*/
85 typedef struct gbtmpnbls {
91 /* This function is exactly the same as the one in bondfree.c. The reason
92 * it is copied here is that the bonded gb-interactions are evaluated
93 * not in calc_bonds, but rather in calc_gb_forces
95 static int pbc_rvec_sub(const t_pbc *pbc, const rvec xi, const rvec xj, rvec dx)
99 return pbc_dx_aiuc(pbc, xi, xj, dx);
103 rvec_sub(xi, xj, dx);
108 int init_gb_nblist(int natoms, t_nblist *nl)
110 nl->maxnri = natoms*4;
120 /*nl->nltype = nltype;*/
122 srenew(nl->iinr, nl->maxnri);
123 srenew(nl->gid, nl->maxnri);
124 srenew(nl->shift, nl->maxnri);
125 srenew(nl->jindex, nl->maxnri+1);
132 void gb_pd_send(t_commrec gmx_unused *cr, real gmx_unused *send_data, int gmx_unused nr)
136 int *index, *sendc, *disp;
138 snew(sendc, cr->nnodes);
139 snew(disp, cr->nnodes);
141 index = pd_index(cr);
144 /* Setup count/index arrays */
145 for (i = 0; i < cr->nnodes; i++)
147 sendc[i] = index[i+1]-index[i];
151 /* Do communication */
152 MPI_Gatherv(send_data+index[cur], sendc[cur], GMX_MPI_REAL, send_data, sendc,
153 disp, GMX_MPI_REAL, 0, cr->mpi_comm_mygroup);
154 MPI_Bcast(send_data, nr, GMX_MPI_REAL, 0, cr->mpi_comm_mygroup);
160 int init_gb_plist(t_params *p_list)
163 p_list->param = NULL;
170 int init_gb_still(const t_commrec *cr,
171 const t_atomtypes *atype, t_idef *idef, t_atoms *atoms,
172 gmx_genborn_t *born, int natoms)
175 int i, j, i1, i2, k, m, nbond, nang, ia, ib, ic, id, nb, idx, idx2, at;
179 real r, ri, rj, ri2, ri3, rj2, r2, r3, r4, rk, ratio, term, h, doffset;
180 real p1, p2, p3, factor, cosine, rab, rbc;
187 snew(born->gpol_still_work, natoms+3);
193 pd_at_range(cr, &at0, &at1);
195 for (i = 0; i < natoms; i++)
212 doffset = born->gb_doffset;
214 for (i = 0; i < natoms; i++)
216 born->gpol_globalindex[i] = born->vsolv_globalindex[i] =
217 born->gb_radius_globalindex[i] = 0;
220 /* Compute atomic solvation volumes for Still method */
221 for (i = 0; i < natoms; i++)
223 ri = atype->gb_radius[atoms->atom[i].type];
224 born->gb_radius_globalindex[i] = ri;
226 born->vsolv_globalindex[i] = (4*M_PI/3)*r3;
229 for (j = 0; j < idef->il[F_GB12].nr; j += 3)
231 m = idef->il[F_GB12].iatoms[j];
232 ia = idef->il[F_GB12].iatoms[j+1];
233 ib = idef->il[F_GB12].iatoms[j+2];
235 r = 1.01*idef->iparams[m].gb.st;
237 ri = atype->gb_radius[atoms->atom[ia].type];
238 rj = atype->gb_radius[atoms->atom[ib].type];
244 ratio = (rj2-ri2-r*r)/(2*ri*r);
246 term = (M_PI/3.0)*h*h*(3.0*ri-h);
254 born->vsolv_globalindex[ia] -= term;
257 ratio = (ri2-rj2-r*r)/(2*rj*r);
259 term = (M_PI/3.0)*h*h*(3.0*rj-h);
267 born->vsolv_globalindex[ib] -= term;
273 gmx_sum(natoms, vsol, cr);
275 for (i = 0; i < natoms; i++)
277 born->vsolv_globalindex[i] = born->vsolv_globalindex[i]-vsol[i];
281 /* Get the self-, 1-2 and 1-3 polarization energies for analytical Still
284 for (j = 0; j < natoms; j++)
286 if (born->use_globalindex[j] == 1)
288 born->gpol_globalindex[j] = -0.5*ONE_4PI_EPS0/
289 (atype->gb_radius[atoms->atom[j].type]-doffset+STILL_P1);
294 for (j = 0; j < idef->il[F_GB12].nr; j += 3)
296 m = idef->il[F_GB12].iatoms[j];
297 ia = idef->il[F_GB12].iatoms[j+1];
298 ib = idef->il[F_GB12].iatoms[j+2];
300 r = idef->iparams[m].gb.st;
306 gp[ia] += STILL_P2*born->vsolv_globalindex[ib]/r4;
307 gp[ib] += STILL_P2*born->vsolv_globalindex[ia]/r4;
311 born->gpol_globalindex[ia] = born->gpol_globalindex[ia]+
312 STILL_P2*born->vsolv_globalindex[ib]/r4;
313 born->gpol_globalindex[ib] = born->gpol_globalindex[ib]+
314 STILL_P2*born->vsolv_globalindex[ia]/r4;
319 for (j = 0; j < idef->il[F_GB13].nr; j += 3)
321 m = idef->il[F_GB13].iatoms[j];
322 ia = idef->il[F_GB13].iatoms[j+1];
323 ib = idef->il[F_GB13].iatoms[j+2];
325 r = idef->iparams[m].gb.st;
330 gp[ia] += STILL_P3*born->vsolv[ib]/r4;
331 gp[ib] += STILL_P3*born->vsolv[ia]/r4;
335 born->gpol_globalindex[ia] = born->gpol_globalindex[ia]+
336 STILL_P3*born->vsolv_globalindex[ib]/r4;
337 born->gpol_globalindex[ib] = born->gpol_globalindex[ib]+
338 STILL_P3*born->vsolv_globalindex[ia]/r4;
344 gmx_sum(natoms, gp, cr);
346 for (i = 0; i < natoms; i++)
348 born->gpol_globalindex[i] = born->gpol_globalindex[i]+gp[i];
358 /* Initialize all GB datastructs and compute polarization energies */
359 int init_gb(gmx_genborn_t **p_born,
360 const t_commrec *cr, t_forcerec *fr, const t_inputrec *ir,
361 const gmx_mtop_t *mtop, int gb_algorithm)
363 int i, j, m, ai, aj, jj, natoms, nalloc;
364 real rai, sk, p, doffset;
368 gmx_localtop_t *localtop;
370 natoms = mtop->natoms;
372 atoms = gmx_mtop_global_atoms(mtop);
373 localtop = gmx_mtop_generate_local_top(mtop, ir);
380 snew(born->drobc, natoms);
381 snew(born->bRad, natoms);
383 /* Allocate memory for the global data arrays */
384 snew(born->param_globalindex, natoms+3);
385 snew(born->gpol_globalindex, natoms+3);
386 snew(born->vsolv_globalindex, natoms+3);
387 snew(born->gb_radius_globalindex, natoms+3);
388 snew(born->use_globalindex, natoms+3);
390 snew(fr->invsqrta, natoms);
391 snew(fr->dvda, natoms);
394 fr->dadx_rawptr = NULL;
396 born->gpol_still_work = NULL;
397 born->gpol_hct_work = NULL;
399 /* snew(born->asurf,natoms); */
400 /* snew(born->dasurf,natoms); */
402 /* Initialize the gb neighbourlist */
403 init_gb_nblist(natoms, &(fr->gblist));
405 /* Do the Vsites exclusions (if any) */
406 for (i = 0; i < natoms; i++)
408 jj = atoms.atom[i].type;
409 if (mtop->atomtypes.gb_radius[atoms.atom[i].type] > 0)
411 born->use_globalindex[i] = 1;
415 born->use_globalindex[i] = 0;
418 /* If we have a Vsite, put vs_globalindex[i]=0 */
419 if (C6 (fr->nbfp, fr->ntype, jj, jj) == 0 &&
420 C12(fr->nbfp, fr->ntype, jj, jj) == 0 &&
421 atoms.atom[i].q == 0)
423 born->use_globalindex[i] = 0;
427 /* Copy algorithm parameters from inputrecord to local structure */
428 born->obc_alpha = ir->gb_obc_alpha;
429 born->obc_beta = ir->gb_obc_beta;
430 born->obc_gamma = ir->gb_obc_gamma;
431 born->gb_doffset = ir->gb_dielectric_offset;
432 born->gb_epsilon_solvent = ir->gb_epsilon_solvent;
433 born->epsilon_r = ir->epsilon_r;
435 doffset = born->gb_doffset;
437 /* Set the surface tension */
438 born->sa_surface_tension = ir->sa_surface_tension;
440 /* If Still model, initialise the polarisation energies */
441 if (gb_algorithm == egbSTILL)
443 init_gb_still(cr, &(mtop->atomtypes), &(localtop->idef), &atoms,
448 /* If HCT/OBC, precalculate the sk*atype->S_hct factors */
449 else if (gb_algorithm == egbHCT || gb_algorithm == egbOBC)
452 snew(born->gpol_hct_work, natoms+3);
454 for (i = 0; i < natoms; i++)
456 if (born->use_globalindex[i] == 1)
458 rai = mtop->atomtypes.gb_radius[atoms.atom[i].type]-doffset;
459 sk = rai * mtop->atomtypes.S_hct[atoms.atom[i].type];
460 born->param_globalindex[i] = sk;
461 born->gb_radius_globalindex[i] = rai;
465 born->param_globalindex[i] = 0;
466 born->gb_radius_globalindex[i] = 0;
471 /* Allocate memory for work arrays for temporary use */
472 snew(born->work, natoms+4);
473 snew(born->count, natoms);
474 snew(born->nblist_work, natoms);
476 /* Domain decomposition specific stuff */
485 calc_gb_rad_still(t_commrec *cr, t_forcerec *fr, int natoms, gmx_localtop_t *top,
486 rvec x[], t_nblist *nl,
487 gmx_genborn_t *born, t_mdatoms *md)
489 int i, k, n, nj0, nj1, ai, aj, type;
492 real gpi, dr, dr2, dr4, idr4, rvdw, ratio, ccf, theta, term, rai, raj;
493 real ix1, iy1, iz1, jx1, jy1, jz1, dx11, dy11, dz11;
494 real rinv, idr2, idr6, vaj, dccf, cosq, sinq, prod, gpi2;
496 real vai, prod_ai, icf4, icf6;
498 factor = 0.5*ONE_4PI_EPS0;
501 for (i = 0; i < born->nr; i++)
503 born->gpol_still_work[i] = 0;
506 for (i = 0; i < nl->nri; i++)
511 nj1 = nl->jindex[i+1];
513 /* Load shifts for this list */
514 shift = nl->shift[i];
515 shX = fr->shift_vec[shift][0];
516 shY = fr->shift_vec[shift][1];
517 shZ = fr->shift_vec[shift][2];
521 rai = top->atomtypes.gb_radius[md->typeA[ai]];
522 vai = born->vsolv[ai];
523 prod_ai = STILL_P4*vai;
525 /* Load atom i coordinates, add shift vectors */
526 ix1 = shX + x[ai][0];
527 iy1 = shY + x[ai][1];
528 iz1 = shZ + x[ai][2];
530 for (k = nj0; k < nj1 && nl->jjnr[k] >= 0; k++)
541 dr2 = dx11*dx11+dy11*dy11+dz11*dz11;
542 rinv = gmx_invsqrt(dr2);
547 raj = top->atomtypes.gb_radius[md->typeA[aj]];
551 ratio = dr2 / (rvdw * rvdw);
552 vaj = born->vsolv[aj];
554 if (ratio > STILL_P5INV)
561 theta = ratio*STILL_PIP5;
563 term = 0.5*(1.0-cosq);
565 sinq = 1.0 - cosq*cosq;
566 dccf = 2.0*term*sinq*gmx_invsqrt(sinq)*theta;
571 icf6 = (4*ccf-dccf)*idr6;
572 born->gpol_still_work[aj] += prod_ai*icf4;
575 /* Save ai->aj and aj->ai chain rule terms */
576 fr->dadx[n++] = prod*icf6;
577 fr->dadx[n++] = prod_ai*icf6;
579 born->gpol_still_work[ai] += gpi;
582 /* Parallel summations */
585 gmx_sum(natoms, born->gpol_still_work, cr);
587 else if (DOMAINDECOMP(cr))
589 dd_atom_sum_real(cr->dd, born->gpol_still_work);
592 /* Calculate the radii */
593 for (i = 0; i < fr->natoms_force; i++) /* PELA born->nr */
595 if (born->use[i] != 0)
597 gpi = born->gpol[i]+born->gpol_still_work[i];
599 born->bRad[i] = factor*gmx_invsqrt(gpi2);
600 fr->invsqrta[i] = gmx_invsqrt(born->bRad[i]);
604 /* Extra communication required for DD */
605 if (DOMAINDECOMP(cr))
607 dd_atom_spread_real(cr->dd, born->bRad);
608 dd_atom_spread_real(cr->dd, fr->invsqrta);
617 calc_gb_rad_hct(t_commrec *cr, t_forcerec *fr, int natoms, gmx_localtop_t *top,
618 rvec x[], t_nblist *nl,
619 gmx_genborn_t *born, t_mdatoms *md)
621 int i, k, n, ai, aj, nj0, nj1, at0, at1;
624 real rai, raj, gpi, dr2, dr, sk, sk_ai, sk2, sk2_ai, lij, uij, diff2, tmp, sum_ai;
625 real rad, min_rad, rinv, rai_inv;
626 real ix1, iy1, iz1, jx1, jy1, jz1, dx11, dy11, dz11;
627 real lij2, uij2, lij3, uij3, t1, t2, t3;
628 real lij_inv, dlij, duij, sk2_rinv, prod, log_term;
629 real doffset, raj_inv, dadx_val;
632 doffset = born->gb_doffset;
633 gb_radius = born->gb_radius;
635 for (i = 0; i < born->nr; i++)
637 born->gpol_hct_work[i] = 0;
640 /* Keep the compiler happy */
644 for (i = 0; i < nl->nri; i++)
649 nj1 = nl->jindex[i+1];
651 /* Load shifts for this list */
652 shift = nl->shift[i];
653 shX = fr->shift_vec[shift][0];
654 shY = fr->shift_vec[shift][1];
655 shZ = fr->shift_vec[shift][2];
660 sk_ai = born->param[ai];
661 sk2_ai = sk_ai*sk_ai;
663 /* Load atom i coordinates, add shift vectors */
664 ix1 = shX + x[ai][0];
665 iy1 = shY + x[ai][1];
666 iz1 = shZ + x[ai][2];
670 for (k = nj0; k < nj1 && nl->jjnr[k] >= 0; k++)
682 dr2 = dx11*dx11+dy11*dy11+dz11*dz11;
683 rinv = gmx_invsqrt(dr2);
686 sk = born->param[aj];
689 /* aj -> ai interaction */
710 lij_inv = gmx_invsqrt(lij2);
713 prod = 0.25*sk2_rinv;
715 log_term = log(uij*lij_inv);
717 tmp = lij-uij + 0.25*dr*diff2 + (0.5*rinv)*log_term +
722 tmp = tmp + 2.0 * (rai_inv-lij);
725 t1 = 0.5*lij2 + prod*lij3 - 0.25*(lij*rinv+lij3*dr);
726 t2 = -0.5*uij2 - 0.25*sk2_rinv*uij3 + 0.25*(uij*rinv+uij3*dr);
727 t3 = 0.125*(1.0+sk2_rinv*rinv)*(-diff2)+0.25*log_term*rinv*rinv;
729 dadx_val = (dlij*t1+t2+t3)*rinv; /* rb2 is moved to chainrule */
730 /* fr->dadx[n++] = (dlij*t1+duij*t2+t3)*rinv; */
731 /* rb2 is moved to chainrule */
739 fr->dadx[n++] = dadx_val;
742 /* ai -> aj interaction */
743 if (raj < dr + sk_ai)
745 lij = 1.0/(dr-sk_ai);
758 uij = 1.0/(dr+sk_ai);
764 lij_inv = gmx_invsqrt(lij2);
765 sk2 = sk2_ai; /* sk2_ai = sk_ai * sk_ai in i loop above */
767 prod = 0.25 * sk2_rinv;
769 /* log_term = table_log(uij*lij_inv,born->log_table,
770 LOG_TABLE_ACCURACY); */
771 log_term = log(uij*lij_inv);
773 tmp = lij-uij + 0.25*dr*diff2 + (0.5*rinv)*log_term +
778 tmp = tmp + 2.0 * (raj_inv-lij);
782 t1 = 0.5*lij2 + prod*lij3 - 0.25*(lij*rinv+lij3*dr);
783 t2 = -0.5*uij2 - 0.25*sk2_rinv*uij3 + 0.25*(uij*rinv+uij3*dr);
784 t3 = 0.125*(1.0+sk2_rinv*rinv)*(-diff2)+0.25*log_term*rinv*rinv;
786 dadx_val = (dlij*t1+t2+t3)*rinv; /* rb2 is moved to chainrule */
787 /* fr->dadx[n++] = (dlij*t1+duij*t2+t3)*rinv; */ /* rb2 is moved to chainrule */
789 born->gpol_hct_work[aj] += 0.5*tmp;
795 fr->dadx[n++] = dadx_val;
798 born->gpol_hct_work[ai] += sum_ai;
801 /* Parallel summations */
804 gmx_sum(natoms, born->gpol_hct_work, cr);
806 else if (DOMAINDECOMP(cr))
808 dd_atom_sum_real(cr->dd, born->gpol_hct_work);
811 for (i = 0; i < fr->natoms_force; i++) /* PELA born->nr */
813 if (born->use[i] != 0)
815 rai = top->atomtypes.gb_radius[md->typeA[i]]-doffset;
816 sum_ai = 1.0/rai - born->gpol_hct_work[i];
817 min_rad = rai + doffset;
820 born->bRad[i] = rad > min_rad ? rad : min_rad;
821 fr->invsqrta[i] = gmx_invsqrt(born->bRad[i]);
825 /* Extra communication required for DD */
826 if (DOMAINDECOMP(cr))
828 dd_atom_spread_real(cr->dd, born->bRad);
829 dd_atom_spread_real(cr->dd, fr->invsqrta);
837 calc_gb_rad_obc(t_commrec *cr, t_forcerec *fr, int natoms, gmx_localtop_t *top,
838 rvec x[], t_nblist *nl, gmx_genborn_t *born, t_mdatoms *md)
840 int i, k, ai, aj, nj0, nj1, n, at0, at1;
843 real rai, raj, gpi, dr2, dr, sk, sk2, lij, uij, diff2, tmp, sum_ai;
844 real rad, min_rad, sum_ai2, sum_ai3, tsum, tchain, rinv, rai_inv, lij_inv, rai_inv2;
845 real log_term, prod, sk2_rinv, sk_ai, sk2_ai;
846 real ix1, iy1, iz1, jx1, jy1, jz1, dx11, dy11, dz11;
847 real lij2, uij2, lij3, uij3, dlij, duij, t1, t2, t3;
848 real doffset, raj_inv, dadx_val;
851 /* Keep the compiler happy */
856 doffset = born->gb_doffset;
857 gb_radius = born->gb_radius;
859 for (i = 0; i < born->nr; i++)
861 born->gpol_hct_work[i] = 0;
864 for (i = 0; i < nl->nri; i++)
869 nj1 = nl->jindex[i+1];
871 /* Load shifts for this list */
872 shift = nl->shift[i];
873 shX = fr->shift_vec[shift][0];
874 shY = fr->shift_vec[shift][1];
875 shZ = fr->shift_vec[shift][2];
880 sk_ai = born->param[ai];
881 sk2_ai = sk_ai*sk_ai;
883 /* Load atom i coordinates, add shift vectors */
884 ix1 = shX + x[ai][0];
885 iy1 = shY + x[ai][1];
886 iz1 = shZ + x[ai][2];
890 for (k = nj0; k < nj1 && nl->jjnr[k] >= 0; k++)
902 dr2 = dx11*dx11+dy11*dy11+dz11*dz11;
903 rinv = gmx_invsqrt(dr2);
906 /* sk is precalculated in init_gb() */
907 sk = born->param[aj];
910 /* aj -> ai interaction */
930 lij_inv = gmx_invsqrt(lij2);
933 prod = 0.25*sk2_rinv;
935 log_term = log(uij*lij_inv);
937 tmp = lij-uij + 0.25*dr*diff2 + (0.5*rinv)*log_term + prod*(-diff2);
941 tmp = tmp + 2.0 * (rai_inv-lij);
945 t1 = 0.5*lij2 + prod*lij3 - 0.25*(lij*rinv+lij3*dr);
946 t2 = -0.5*uij2 - 0.25*sk2_rinv*uij3 + 0.25*(uij*rinv+uij3*dr);
947 t3 = 0.125*(1.0+sk2_rinv*rinv)*(-diff2)+0.25*log_term*rinv*rinv;
949 dadx_val = (dlij*t1+t2+t3)*rinv; /* rb2 is moved to chainrule */
957 fr->dadx[n++] = dadx_val;
959 /* ai -> aj interaction */
960 if (raj < dr + sk_ai)
962 lij = 1.0/(dr-sk_ai);
975 uij = 1.0/(dr+sk_ai);
981 lij_inv = gmx_invsqrt(lij2);
982 sk2 = sk2_ai; /* sk2_ai = sk_ai * sk_ai in i loop above */
984 prod = 0.25 * sk2_rinv;
986 /* log_term = table_log(uij*lij_inv,born->log_table,LOG_TABLE_ACCURACY); */
987 log_term = log(uij*lij_inv);
989 tmp = lij-uij + 0.25*dr*diff2 + (0.5*rinv)*log_term + prod*(-diff2);
993 tmp = tmp + 2.0 * (raj_inv-lij);
996 t1 = 0.5*lij2 + prod*lij3 - 0.25*(lij*rinv+lij3*dr);
997 t2 = -0.5*uij2 - 0.25*sk2_rinv*uij3 + 0.25*(uij*rinv+uij3*dr);
998 t3 = 0.125*(1.0+sk2_rinv*rinv)*(-diff2)+0.25*log_term*rinv*rinv;
1000 dadx_val = (dlij*t1+t2+t3)*rinv; /* rb2 is moved to chainrule */
1002 born->gpol_hct_work[aj] += 0.5*tmp;
1009 fr->dadx[n++] = dadx_val;
1012 born->gpol_hct_work[ai] += sum_ai;
1016 /* Parallel summations */
1019 gmx_sum(natoms, born->gpol_hct_work, cr);
1021 else if (DOMAINDECOMP(cr))
1023 dd_atom_sum_real(cr->dd, born->gpol_hct_work);
1026 for (i = 0; i < fr->natoms_force; i++) /* PELA born->nr */
1028 if (born->use[i] != 0)
1030 rai = top->atomtypes.gb_radius[md->typeA[i]];
1034 sum_ai = rai * born->gpol_hct_work[i];
1035 sum_ai2 = sum_ai * sum_ai;
1036 sum_ai3 = sum_ai2 * sum_ai;
1038 tsum = tanh(born->obc_alpha*sum_ai-born->obc_beta*sum_ai2+born->obc_gamma*sum_ai3);
1039 born->bRad[i] = rai_inv - tsum*rai_inv2;
1040 born->bRad[i] = 1.0 / born->bRad[i];
1042 fr->invsqrta[i] = gmx_invsqrt(born->bRad[i]);
1044 tchain = rai * (born->obc_alpha-2*born->obc_beta*sum_ai+3*born->obc_gamma*sum_ai2);
1045 born->drobc[i] = (1.0-tsum*tsum)*tchain*rai_inv2;
1049 /* Extra (local) communication required for DD */
1050 if (DOMAINDECOMP(cr))
1052 dd_atom_spread_real(cr->dd, born->bRad);
1053 dd_atom_spread_real(cr->dd, fr->invsqrta);
1054 dd_atom_spread_real(cr->dd, born->drobc);
1063 int calc_gb_rad(t_commrec *cr, t_forcerec *fr, t_inputrec *ir, gmx_localtop_t *top,
1064 rvec x[], t_nblist *nl, gmx_genborn_t *born, t_mdatoms *md, t_nrnb *nrnb)
1070 if (fr->bAllvsAll && fr->dadx == NULL)
1072 /* We might need up to 8 atoms of padding before and after,
1073 * and another 4 units to guarantee SSE alignment.
1075 fr->nalloc_dadx = 2*(md->homenr+12)*(md->nr/2+1+12);
1076 snew(fr->dadx_rawptr, fr->nalloc_dadx);
1077 fr->dadx = (real *) (((size_t) fr->dadx_rawptr + 16) & (~((size_t) 15)));
1081 /* In the SSE-enabled gb-loops, when writing to dadx, we
1082 * always write 2*4 elements at a time, even in the case with only
1083 * 1-3 j particles, where we only really need to write 2*(1-3)
1084 * elements. This is because we want dadx to be aligned to a 16-
1085 * byte boundary, and being able to use _mm_store/load_ps
1087 ndadx = 2 * (nl->nrj + 3*nl->nri);
1089 /* First, reallocate the dadx array, we need 3 extra for SSE */
1090 if (ndadx + 3 > fr->nalloc_dadx)
1092 fr->nalloc_dadx = over_alloc_large(ndadx) + 3;
1093 srenew(fr->dadx_rawptr, fr->nalloc_dadx);
1094 fr->dadx = (real *) (((size_t) fr->dadx_rawptr + 16) & (~((size_t) 15)));
1100 cnt = md->homenr*(md->nr/2+1);
1102 if (ir->gb_algorithm == egbSTILL)
1104 #if 0 && defined (GMX_X86_SSE2)
1105 if (fr->use_acceleration)
1108 genborn_allvsall_calc_still_radii_sse2_double(fr, md, born, top, x[0], cr, &fr->AllvsAll_workgb);
1110 genborn_allvsall_calc_still_radii_sse2_single(fr, md, born, top, x[0], cr, &fr->AllvsAll_workgb);
1115 genborn_allvsall_calc_still_radii(fr, md, born, top, x[0], cr, &fr->AllvsAll_workgb);
1118 genborn_allvsall_calc_still_radii(fr, md, born, top, x[0], cr, &fr->AllvsAll_workgb);
1120 /* 13 flops in outer loop, 47 flops in inner loop */
1121 inc_nrnb(nrnb, eNR_BORN_AVA_RADII_STILL, md->homenr*13+cnt*47);
1123 else if (ir->gb_algorithm == egbHCT || ir->gb_algorithm == egbOBC)
1125 #if 0 && defined (GMX_X86_SSE2)
1126 if (fr->use_acceleration)
1129 genborn_allvsall_calc_hct_obc_radii_sse2_double(fr, md, born, ir->gb_algorithm, top, x[0], cr, &fr->AllvsAll_workgb);
1131 genborn_allvsall_calc_hct_obc_radii_sse2_single(fr, md, born, ir->gb_algorithm, top, x[0], cr, &fr->AllvsAll_workgb);
1136 genborn_allvsall_calc_hct_obc_radii(fr, md, born, ir->gb_algorithm, top, x[0], cr, &fr->AllvsAll_workgb);
1139 genborn_allvsall_calc_hct_obc_radii(fr, md, born, ir->gb_algorithm, top, x[0], cr, &fr->AllvsAll_workgb);
1141 /* 24 flops in outer loop, 183 in inner */
1142 inc_nrnb(nrnb, eNR_BORN_AVA_RADII_HCT_OBC, md->homenr*24+cnt*183);
1146 gmx_fatal(FARGS, "Bad gb algorithm for all-vs-all interactions");
1151 /* Switch for determining which algorithm to use for Born radii calculation */
1154 #if 0 && defined (GMX_X86_SSE2)
1155 /* x86 or x86-64 with GCC inline assembly and/or SSE intrinsics */
1156 switch (ir->gb_algorithm)
1159 if (fr->use_acceleration)
1161 calc_gb_rad_still_sse2_double(cr, fr, born->nr, top, atype, x[0], nl, born);
1165 calc_gb_rad_still(cr, fr, born->nr, top, atype, x, nl, born, md);
1169 if (fr->use_acceleration)
1171 calc_gb_rad_hct_obc_sse2_double(cr, fr, born->nr, top, atype, x[0], nl, born, md, ir->gb_algorithm);
1175 calc_gb_rad_hct(cr, fr, born->nr, top, atype, x, nl, born, md);
1179 if (fr->use_acceleration)
1181 calc_gb_rad_hct_obc_sse2_double(cr, fr, born->nr, top, atype, x[0], nl, born, md, ir->gb_algorithm);
1185 calc_gb_rad_obc(cr, fr, born->nr, top, atype, x, nl, born, md);
1190 gmx_fatal(FARGS, "Unknown double precision sse-enabled algorithm for Born radii calculation: %d", ir->gb_algorithm);
1193 switch (ir->gb_algorithm)
1196 calc_gb_rad_still(cr, fr, born->nr, top, x, nl, born, md);
1199 calc_gb_rad_hct(cr, fr, born->nr, top, x, nl, born, md);
1202 calc_gb_rad_obc(cr, fr, born->nr, top, x, nl, born, md);
1206 gmx_fatal(FARGS, "Unknown double precision algorithm for Born radii calculation: %d", ir->gb_algorithm);
1213 #if 0 && defined (GMX_X86_SSE2)
1214 /* x86 or x86-64 with GCC inline assembly and/or SSE intrinsics */
1215 switch (ir->gb_algorithm)
1218 if (fr->use_acceleration)
1220 calc_gb_rad_still_sse2_single(cr, fr, born->nr, top, x[0], nl, born);
1224 calc_gb_rad_still(cr, fr, born->nr, top, x, nl, born, md);
1228 if (fr->use_acceleration)
1230 calc_gb_rad_hct_obc_sse2_single(cr, fr, born->nr, top, x[0], nl, born, md, ir->gb_algorithm);
1234 calc_gb_rad_hct(cr, fr, born->nr, top, x, nl, born, md);
1239 if (fr->use_acceleration)
1241 calc_gb_rad_hct_obc_sse2_single(cr, fr, born->nr, top, x[0], nl, born, md, ir->gb_algorithm);
1245 calc_gb_rad_obc(cr, fr, born->nr, top, x, nl, born, md);
1250 gmx_fatal(FARGS, "Unknown sse-enabled algorithm for Born radii calculation: %d", ir->gb_algorithm);
1254 switch (ir->gb_algorithm)
1257 calc_gb_rad_still(cr, fr, born->nr, top, x, nl, born, md);
1260 calc_gb_rad_hct(cr, fr, born->nr, top, x, nl, born, md);
1263 calc_gb_rad_obc(cr, fr, born->nr, top, x, nl, born, md);
1267 gmx_fatal(FARGS, "Unknown algorithm for Born radii calculation: %d", ir->gb_algorithm);
1270 #endif /* Single precision sse */
1272 #endif /* Double or single precision */
1274 if (fr->bAllvsAll == FALSE)
1276 switch (ir->gb_algorithm)
1279 /* 17 flops per outer loop iteration, 47 flops per inner loop */
1280 inc_nrnb(nrnb, eNR_BORN_RADII_STILL, nl->nri*17+nl->nrj*47);
1284 /* 61 (assuming 10 for tanh) flops for outer loop iteration, 183 flops per inner loop */
1285 inc_nrnb(nrnb, eNR_BORN_RADII_HCT_OBC, nl->nri*61+nl->nrj*183);
1298 real gb_bonds_tab(rvec x[], rvec f[], rvec fshift[], real *charge, real *p_gbtabscale,
1299 real *invsqrta, real *dvda, real *GBtab, t_idef *idef, real epsilon_r,
1300 real gb_epsilon_solvent, real facel, const t_pbc *pbc, const t_graph *graph)
1302 int i, j, n0, m, nnn, type, ai, aj;
1308 real isaprod, qq, gbscale, gbtabscale, Y, F, Geps, Heps2, Fp, VV, FF, rt, eps, eps2;
1309 real vgb, fgb, vcoul, fijC, dvdatmp, fscal, dvdaj;
1315 t_iatom *forceatoms;
1317 /* Scale the electrostatics by gb_epsilon_solvent */
1318 facel = facel * ((1.0/epsilon_r) - 1.0/gb_epsilon_solvent);
1320 gbtabscale = *p_gbtabscale;
1323 for (j = F_GB12; j <= F_GB14; j++)
1325 forceatoms = idef->il[j].iatoms;
1327 for (i = 0; i < idef->il[j].nr; )
1329 /* To avoid reading in the interaction type, we just increment i to pass over
1330 * the types in the forceatoms array, this saves some memory accesses
1333 ai = forceatoms[i++];
1334 aj = forceatoms[i++];
1336 ki = pbc_rvec_sub(pbc, x[ai], x[aj], dx);
1337 rsq11 = iprod(dx, dx);
1339 isai = invsqrta[ai];
1340 iq = (-1)*facel*charge[ai];
1342 rinv11 = gmx_invsqrt(rsq11);
1343 isaj = invsqrta[aj];
1344 isaprod = isai*isaj;
1345 qq = isaprod*iq*charge[aj];
1346 gbscale = isaprod*gbtabscale;
1355 Geps = eps*GBtab[nnn+2];
1356 Heps2 = eps2*GBtab[nnn+3];
1359 FF = Fp+Geps+2.0*Heps2;
1361 fijC = qq*FF*gbscale;
1362 dvdatmp = -(vgb+fijC*r)*0.5;
1363 dvda[aj] = dvda[aj] + dvdatmp*isaj*isaj;
1364 dvda[ai] = dvda[ai] + dvdatmp*isai*isai;
1365 vctot = vctot + vgb;
1366 fgb = -(fijC)*rinv11;
1370 ivec_sub(SHIFT_IVEC(graph, ai), SHIFT_IVEC(graph, aj), dt);
1374 for (m = 0; (m < DIM); m++) /* 15 */
1379 fshift[ki][m] += fscal;
1380 fshift[CENTRAL][m] -= fscal;
1388 real calc_gb_selfcorrections(t_commrec *cr, int natoms,
1389 real *charge, gmx_genborn_t *born, real *dvda, double facel)
1391 int i, ai, at0, at1;
1392 real rai, e, derb, q, q2, fi, rai_inv, vtot;
1396 pd_at_range(cr, &at0, &at1);
1398 else if (DOMAINDECOMP(cr))
1401 at1 = cr->dd->nat_home;
1410 /* Scale the electrostatics by gb_epsilon_solvent */
1411 facel = facel * ((1.0/born->epsilon_r) - 1.0/born->gb_epsilon_solvent);
1415 /* Apply self corrections */
1416 for (i = at0; i < at1; i++)
1420 if (born->use[ai] == 1)
1422 rai = born->bRad[ai];
1428 derb = 0.5*e*rai_inv*rai_inv;
1429 dvda[ai] += derb*rai;
1438 real calc_gb_nonpolar(t_commrec *cr, t_forcerec *fr, int natoms, gmx_genborn_t *born, gmx_localtop_t *top,
1439 real *dvda, t_mdatoms *md)
1441 int ai, i, at0, at1;
1442 real e, es, rai, rbi, term, probe, tmp, factor;
1443 real rbi_inv, rbi_inv2;
1445 /* To keep the compiler happy */
1450 pd_at_range(cr, &at0, &at1);
1452 else if (DOMAINDECOMP(cr))
1455 at1 = cr->dd->nat_home;
1463 /* factor is the surface tension */
1464 factor = born->sa_surface_tension;
1467 // The surface tension factor is 0.0049 for Still model, 0.0054 for HCT/OBC
1468 if(gb_algorithm==egbSTILL)
1470 factor=0.0049*100*CAL2JOULE;
1474 factor=0.0054*100*CAL2JOULE;
1477 /* if(gb_algorithm==egbHCT || gb_algorithm==egbOBC) */
1483 for (i = at0; i < at1; i++)
1487 if (born->use[ai] == 1)
1489 rai = top->atomtypes.gb_radius[md->typeA[ai]];
1490 rbi_inv = fr->invsqrta[ai];
1491 rbi_inv2 = rbi_inv * rbi_inv;
1492 tmp = (rai*rbi_inv2)*(rai*rbi_inv2);
1494 e = factor*term*(rai+probe)*(rai+probe)*tmp;
1495 dvda[ai] = dvda[ai] - 6*e*rbi_inv2;
1505 real calc_gb_chainrule(int natoms, t_nblist *nl, real *dadx, real *dvda, rvec x[], rvec t[], rvec fshift[],
1506 rvec shift_vec[], int gb_algorithm, gmx_genborn_t *born)
1508 int i, k, n, ai, aj, nj0, nj1, n0, n1;
1511 real fgb, fij, rb2, rbi, fix1, fiy1, fiz1;
1512 real ix1, iy1, iz1, jx1, jy1, jz1, dx11, dy11, dz11, rsq11;
1513 real rinv11, tx, ty, tz, rbai, rbaj, fgb_ai;
1523 if (gb_algorithm == egbSTILL)
1525 for (i = n0; i < n1; i++)
1527 rbi = born->bRad[i];
1528 rb[i] = (2 * rbi * rbi * dvda[i])/ONE_4PI_EPS0;
1531 else if (gb_algorithm == egbHCT)
1533 for (i = n0; i < n1; i++)
1535 rbi = born->bRad[i];
1536 rb[i] = rbi * rbi * dvda[i];
1539 else if (gb_algorithm == egbOBC)
1541 for (i = n0; i < n1; i++)
1543 rbi = born->bRad[i];
1544 rb[i] = rbi * rbi * born->drobc[i] * dvda[i];
1548 for (i = 0; i < nl->nri; i++)
1552 nj0 = nl->jindex[i];
1553 nj1 = nl->jindex[i+1];
1555 /* Load shifts for this list */
1556 shift = nl->shift[i];
1557 shX = shift_vec[shift][0];
1558 shY = shift_vec[shift][1];
1559 shZ = shift_vec[shift][2];
1561 /* Load atom i coordinates, add shift vectors */
1562 ix1 = shX + x[ai][0];
1563 iy1 = shY + x[ai][1];
1564 iz1 = shZ + x[ai][2];
1572 for (k = nj0; k < nj1 && nl->jjnr[k] >= 0; k++)
1586 fgb = rbai*dadx[n++];
1587 fgb_ai = rbaj*dadx[n++];
1589 /* Total force between ai and aj is the sum of ai->aj and aj->ai */
1600 /* Update force on atom aj */
1601 t[aj][0] = t[aj][0] - tx;
1602 t[aj][1] = t[aj][1] - ty;
1603 t[aj][2] = t[aj][2] - tz;
1606 /* Update force and shift forces on atom ai */
1607 t[ai][0] = t[ai][0] + fix1;
1608 t[ai][1] = t[ai][1] + fiy1;
1609 t[ai][2] = t[ai][2] + fiz1;
1611 fshift[shift][0] = fshift[shift][0] + fix1;
1612 fshift[shift][1] = fshift[shift][1] + fiy1;
1613 fshift[shift][2] = fshift[shift][2] + fiz1;
1622 calc_gb_forces(t_commrec *cr, t_mdatoms *md, gmx_genborn_t *born, gmx_localtop_t *top,
1623 rvec x[], rvec f[], t_forcerec *fr, t_idef *idef, int gb_algorithm, int sa_algorithm, t_nrnb *nrnb,
1624 const t_pbc *pbc, const t_graph *graph, gmx_enerdata_t *enerd)
1631 const t_pbc *pbc_null;
1642 if (sa_algorithm == esaAPPROX)
1644 /* Do a simple ACE type approximation for the non-polar solvation */
1645 enerd->term[F_NPSOLVATION] += calc_gb_nonpolar(cr, fr, born->nr, born, top, fr->dvda, md);
1648 /* Calculate the bonded GB-interactions using either table or analytical formula */
1649 enerd->term[F_GBPOL] += gb_bonds_tab(x, f, fr->fshift, md->chargeA, &(fr->gbtabscale),
1650 fr->invsqrta, fr->dvda, fr->gbtab.data, idef, born->epsilon_r, born->gb_epsilon_solvent, fr->epsfac, pbc_null, graph);
1652 /* Calculate self corrections to the GB energies - currently only A state used! (FIXME) */
1653 enerd->term[F_GBPOL] += calc_gb_selfcorrections(cr, born->nr, md->chargeA, born, fr->dvda, fr->epsfac);
1655 /* If parallel, sum the derivative of the potential w.r.t the born radii */
1658 gmx_sum(md->nr, fr->dvda, cr);
1660 else if (DOMAINDECOMP(cr))
1662 dd_atom_sum_real(cr->dd, fr->dvda);
1663 dd_atom_spread_real(cr->dd, fr->dvda);
1668 #if 0 && defined (GMX_X86_SSE2)
1669 if (fr->use_acceleration)
1672 genborn_allvsall_calc_chainrule_sse2_double(fr, md, born, x[0], f[0], gb_algorithm, fr->AllvsAll_workgb);
1674 genborn_allvsall_calc_chainrule_sse2_single(fr, md, born, x[0], f[0], gb_algorithm, fr->AllvsAll_workgb);
1679 genborn_allvsall_calc_chainrule(fr, md, born, x[0], f[0], gb_algorithm, fr->AllvsAll_workgb);
1682 genborn_allvsall_calc_chainrule(fr, md, born, x[0], f[0], gb_algorithm, fr->AllvsAll_workgb);
1684 cnt = md->homenr*(md->nr/2+1);
1685 /* 9 flops for outer loop, 15 for inner */
1686 inc_nrnb(nrnb, eNR_BORN_AVA_CHAINRULE, md->homenr*9+cnt*15);
1690 #if 0 && defined (GMX_X86_SSE2)
1691 if (fr->use_acceleration)
1694 calc_gb_chainrule_sse2_double(fr->natoms_force, &(fr->gblist), fr->dadx, fr->dvda, x[0],
1695 f[0], fr->fshift[0], fr->shift_vec[0], gb_algorithm, born, md);
1697 calc_gb_chainrule_sse2_single(fr->natoms_force, &(fr->gblist), fr->dadx, fr->dvda, x[0],
1698 f[0], fr->fshift[0], fr->shift_vec[0], gb_algorithm, born, md);
1703 calc_gb_chainrule(fr->natoms_force, &(fr->gblist), fr->dadx, fr->dvda,
1704 x, f, fr->fshift, fr->shift_vec, gb_algorithm, born, md);
1707 calc_gb_chainrule(fr->natoms_force, &(fr->gblist), fr->dadx, fr->dvda,
1708 x, f, fr->fshift, fr->shift_vec, gb_algorithm, born);
1713 /* 9 flops for outer loop, 15 for inner */
1714 inc_nrnb(nrnb, eNR_BORN_CHAINRULE, fr->gblist.nri*9+fr->gblist.nrj*15);
1718 static void add_j_to_gblist(gbtmpnbl_t *list, int aj)
1720 if (list->naj >= list->aj_nalloc)
1722 list->aj_nalloc = over_alloc_large(list->naj+1);
1723 srenew(list->aj, list->aj_nalloc);
1726 list->aj[list->naj++] = aj;
1729 static gbtmpnbl_t *find_gbtmplist(struct gbtmpnbls *lists, int shift)
1733 /* Search the list with the same shift, if there is one */
1735 while (ind < lists->nlist && shift != lists->list[ind].shift)
1739 if (ind == lists->nlist)
1741 if (lists->nlist == lists->list_nalloc)
1743 lists->list_nalloc++;
1744 srenew(lists->list, lists->list_nalloc);
1745 for (i = lists->nlist; i < lists->list_nalloc; i++)
1747 lists->list[i].aj = NULL;
1748 lists->list[i].aj_nalloc = 0;
1753 lists->list[lists->nlist].shift = shift;
1754 lists->list[lists->nlist].naj = 0;
1758 return &lists->list[ind];
1761 static void add_bondeds_to_gblist(t_ilist *il,
1762 gmx_bool bMolPBC, t_pbc *pbc, t_graph *g, rvec *x,
1763 struct gbtmpnbls *nls)
1765 int ind, j, ai, aj, shift, found;
1771 for (ind = 0; ind < il->nr; ind += 3)
1773 ai = il->iatoms[ind+1];
1774 aj = il->iatoms[ind+2];
1779 rvec_sub(x[ai], x[aj], dx);
1780 ivec_sub(SHIFT_IVEC(g, ai), SHIFT_IVEC(g, aj), dt);
1781 shift = IVEC2IS(dt);
1785 shift = pbc_dx_aiuc(pbc, x[ai], x[aj], dx);
1788 /* Find the list for this shift or create one */
1789 list = find_gbtmplist(&nls[ai], shift);
1793 /* So that we do not add the same bond twice.
1794 * This happens with some constraints between 1-3 atoms
1795 * that are in the bond-list but should not be in the GB nb-list */
1796 for (j = 0; j < list->naj; j++)
1798 if (list->aj[j] == aj)
1808 gmx_incons("ai == aj");
1811 add_j_to_gblist(list, aj);
1817 compare_int (const void * a, const void * b)
1819 return ( *(int*)a - *(int*)b );
1824 int make_gb_nblist(t_commrec *cr, int gb_algorithm,
1825 rvec x[], matrix box,
1826 t_forcerec *fr, t_idef *idef, t_graph *graph, gmx_genborn_t *born)
1828 int i, l, ii, j, k, n, nj0, nj1, ai, aj, at0, at1, found, shift, s;
1833 struct gbtmpnbls *nls;
1834 gbtmpnbl_t *list = NULL;
1836 set_pbc(&pbc, fr->ePBC, box);
1837 nls = born->nblist_work;
1839 for (i = 0; i < born->nr; i++)
1846 set_pbc_dd(&pbc, fr->ePBC, cr->dd, TRUE, box);
1849 switch (gb_algorithm)
1853 /* Loop over 1-2, 1-3 and 1-4 interactions */
1854 for (j = F_GB12; j <= F_GB14; j++)
1856 add_bondeds_to_gblist(&idef->il[j], fr->bMolPBC, &pbc, graph, x, nls);
1860 /* Loop over 1-4 interactions */
1861 add_bondeds_to_gblist(&idef->il[F_GB14], fr->bMolPBC, &pbc, graph, x, nls);
1864 gmx_incons("Unknown GB algorithm");
1867 /* Loop over the VDWQQ and VDW nblists to set up the nonbonded part of the GB list */
1868 for (n = 0; (n < fr->nnblists); n++)
1870 for (i = 0; (i < eNL_NR); i++)
1872 nblist = &(fr->nblists[n].nlist_sr[i]);
1874 if (nblist->nri > 0 && (i == eNL_VDWQQ || i == eNL_QQ))
1876 for (j = 0; j < nblist->nri; j++)
1878 ai = nblist->iinr[j];
1879 shift = nblist->shift[j];
1881 /* Find the list for this shift or create one */
1882 list = find_gbtmplist(&nls[ai], shift);
1884 nj0 = nblist->jindex[j];
1885 nj1 = nblist->jindex[j+1];
1887 /* Add all the j-atoms in the non-bonded list to the GB list */
1888 for (k = nj0; k < nj1; k++)
1890 add_j_to_gblist(list, nblist->jjnr[k]);
1897 /* Zero out some counters */
1901 fr->gblist.jindex[0] = fr->gblist.nri;
1903 for (i = 0; i < fr->natoms_force; i++)
1905 for (s = 0; s < nls[i].nlist; s++)
1907 list = &nls[i].list[s];
1909 /* Only add those atoms that actually have neighbours */
1910 if (born->use[i] != 0)
1912 fr->gblist.iinr[fr->gblist.nri] = i;
1913 fr->gblist.shift[fr->gblist.nri] = list->shift;
1916 for (k = 0; k < list->naj; k++)
1918 /* Memory allocation for jjnr */
1919 if (fr->gblist.nrj >= fr->gblist.maxnrj)
1921 fr->gblist.maxnrj += over_alloc_large(fr->gblist.maxnrj);
1925 fprintf(debug, "Increasing GB neighbourlist j size to %d\n", fr->gblist.maxnrj);
1928 srenew(fr->gblist.jjnr, fr->gblist.maxnrj);
1932 if (i == list->aj[k])
1934 gmx_incons("i == list->aj[k]");
1936 fr->gblist.jjnr[fr->gblist.nrj++] = list->aj[k];
1939 fr->gblist.jindex[fr->gblist.nri] = fr->gblist.nrj;
1946 for (i = 0; i < fr->gblist.nri; i++)
1948 nj0 = fr->gblist.jindex[i];
1949 nj1 = fr->gblist.jindex[i+1];
1950 ai = fr->gblist.iinr[i];
1953 for (j = nj0; j < nj1; j++)
1955 if (fr->gblist.jjnr[j] < ai)
1957 fr->gblist.jjnr[j] += fr->natoms_force;
1960 qsort(fr->gblist.jjnr+nj0, nj1-nj0, sizeof(int), compare_int);
1962 for (j = nj0; j < nj1; j++)
1964 if (fr->gblist.jjnr[j] >= fr->natoms_force)
1966 fr->gblist.jjnr[j] -= fr->natoms_force;
1976 void make_local_gb(const t_commrec *cr, gmx_genborn_t *born, int gb_algorithm)
1979 gmx_domdec_t *dd = NULL;
1981 if (DOMAINDECOMP(cr))
1989 /* Single node or particle decomp (global==local), just copy pointers and return */
1990 if (gb_algorithm == egbSTILL)
1992 born->gpol = born->gpol_globalindex;
1993 born->vsolv = born->vsolv_globalindex;
1994 born->gb_radius = born->gb_radius_globalindex;
1998 born->param = born->param_globalindex;
1999 born->gb_radius = born->gb_radius_globalindex;
2002 born->use = born->use_globalindex;
2007 /* Reallocation of local arrays if necessary */
2008 /* fr->natoms_force is equal to dd->nat_tot */
2009 if (DOMAINDECOMP(cr) && dd->nat_tot > born->nalloc)
2013 nalloc = dd->nat_tot;
2015 /* Arrays specific to different gb algorithms */
2016 if (gb_algorithm == egbSTILL)
2018 srenew(born->gpol, nalloc+3);
2019 srenew(born->vsolv, nalloc+3);
2020 srenew(born->gb_radius, nalloc+3);
2021 for (i = born->nalloc; (i < nalloc+3); i++)
2025 born->gb_radius[i] = 0;
2030 srenew(born->param, nalloc+3);
2031 srenew(born->gb_radius, nalloc+3);
2032 for (i = born->nalloc; (i < nalloc+3); i++)
2035 born->gb_radius[i] = 0;
2039 /* All gb-algorithms use the array for vsites exclusions */
2040 srenew(born->use, nalloc+3);
2041 for (i = born->nalloc; (i < nalloc+3); i++)
2046 born->nalloc = nalloc;
2049 /* With dd, copy algorithm specific arrays */
2050 if (gb_algorithm == egbSTILL)
2052 for (i = at0; i < at1; i++)
2054 born->gpol[i] = born->gpol_globalindex[dd->gatindex[i]];
2055 born->vsolv[i] = born->vsolv_globalindex[dd->gatindex[i]];
2056 born->gb_radius[i] = born->gb_radius_globalindex[dd->gatindex[i]];
2057 born->use[i] = born->use_globalindex[dd->gatindex[i]];
2062 for (i = at0; i < at1; i++)
2064 born->param[i] = born->param_globalindex[dd->gatindex[i]];
2065 born->gb_radius[i] = born->gb_radius_globalindex[dd->gatindex[i]];
2066 born->use[i] = born->use_globalindex[dd->gatindex[i]];