src/gromacs/gmxlib/nonbonded/nb_kernel_c/nb_kernel_ElecRFCut_VdwBhamSw_GeomW4P1_c.c

   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 /*
  36  * Note: this file was generated by the GROMACS c kernel generator.
  37  */
  38 #include "gmxpre.h"
  39
  40 #include "config.h"
  41
  42 #include <math.h>
  43
  44 #include "../nb_kernel.h"
  45 #include "gromacs/legacyheaders/types/simple.h"
  46 #include "gromacs/math/vec.h"
  47 #include "gromacs/legacyheaders/nrnb.h"
  48
  49 /*
  50  * Gromacs nonbonded kernel:   nb_kernel_ElecRFCut_VdwBhamSw_GeomW4P1_VF_c
  51  * Electrostatics interaction: ReactionField
  52  * VdW interaction:            Buckingham
  53  * Geometry:                   Water4-Particle
  54  * Calculate force/pot:        PotentialAndForce
  55  */
  56 void
  57 nb_kernel_ElecRFCut_VdwBhamSw_GeomW4P1_VF_c
  58                     (t_nblist                    * gmx_restrict       nlist,
  59                      rvec                        * gmx_restrict          xx,
  60                      rvec                        * gmx_restrict          ff,
  61                      t_forcerec                  * gmx_restrict          fr,
  62                      t_mdatoms                   * gmx_restrict     mdatoms,
  63                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
  64                      t_nrnb                      * gmx_restrict        nrnb)
  65 {
  66     int              i_shift_offset,i_coord_offset,j_coord_offset;
  67     int              j_index_start,j_index_end;
  68     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
  69     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
  70     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
  71     real             *shiftvec,*fshift,*x,*f;
  72     int              vdwioffset0;
  73     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
  74     int              vdwioffset1;
  75     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
  76     int              vdwioffset2;
  77     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
  78     int              vdwioffset3;
  79     real             ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
  80     int              vdwjidx0;
  81     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
  82     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
  83     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
  84     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
  85     real             dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30,cexp1_30,cexp2_30;
  86     real             velec,felec,velecsum,facel,crf,krf,krf2;
  87     real             *charge;
  88     int              nvdwtype;
  89     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
  90     int              *vdwtype;
  91     real             *vdwparam;
  92     real             rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
  93
  94     x                = xx[0];
  95     f                = ff[0];
  96
  97     nri              = nlist->nri;
  98     iinr             = nlist->iinr;
  99     jindex           = nlist->jindex;
 100     jjnr             = nlist->jjnr;
 101     shiftidx         = nlist->shift;
 102     gid              = nlist->gid;
 103     shiftvec         = fr->shift_vec[0];
 104     fshift           = fr->fshift[0];
 105     facel            = fr->epsfac;
 106     charge           = mdatoms->chargeA;
 107     krf              = fr->ic->k_rf;
 108     krf2             = krf*2.0;
 109     crf              = fr->ic->c_rf;
 110     nvdwtype         = fr->ntype;
 111     vdwparam         = fr->nbfp;
 112     vdwtype          = mdatoms->typeA;
 113
 114     /* Setup water-specific parameters */
 115     inr              = nlist->iinr[0];
 116     iq1              = facel*charge[inr+1];
 117     iq2              = facel*charge[inr+2];
 118     iq3              = facel*charge[inr+3];
 119     vdwioffset0      = 3*nvdwtype*vdwtype[inr+0];
 120
 121     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 122     rcutoff          = fr->rcoulomb;
 123     rcutoff2         = rcutoff*rcutoff;
 124
 125     rswitch          = fr->rvdw_switch;
 126     /* Setup switch parameters */
 127     d                = rcutoff-rswitch;
 128     swV3             = -10.0/(d*d*d);
 129     swV4             =  15.0/(d*d*d*d);
 130     swV5             =  -6.0/(d*d*d*d*d);
 131     swF2             = -30.0/(d*d*d);
 132     swF3             =  60.0/(d*d*d*d);
 133     swF4             = -30.0/(d*d*d*d*d);
 134
 135     outeriter        = 0;
 136     inneriter        = 0;
 137
 138     /* Start outer loop over neighborlists */
 139     for(iidx=0; iidx<nri; iidx++)
 140     {
 141         /* Load shift vector for this list */
 142         i_shift_offset   = DIM*shiftidx[iidx];
 143         shX              = shiftvec[i_shift_offset+XX];
 144         shY              = shiftvec[i_shift_offset+YY];
 145         shZ              = shiftvec[i_shift_offset+ZZ];
 146
 147         /* Load limits for loop over neighbors */
 148         j_index_start    = jindex[iidx];
 149         j_index_end      = jindex[iidx+1];
 150
 151         /* Get outer coordinate index */
 152         inr              = iinr[iidx];
 153         i_coord_offset   = DIM*inr;
 154
 155         /* Load i particle coords and add shift vector */
 156         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 157         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 158         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 159         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 160         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 161         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 162         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 163         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 164         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 165         ix3              = shX + x[i_coord_offset+DIM*3+XX];
 166         iy3              = shY + x[i_coord_offset+DIM*3+YY];
 167         iz3              = shZ + x[i_coord_offset+DIM*3+ZZ];
 168
 169         fix0             = 0.0;
 170         fiy0             = 0.0;
 171         fiz0             = 0.0;
 172         fix1             = 0.0;
 173         fiy1             = 0.0;
 174         fiz1             = 0.0;
 175         fix2             = 0.0;
 176         fiy2             = 0.0;
 177         fiz2             = 0.0;
 178         fix3             = 0.0;
 179         fiy3             = 0.0;
 180         fiz3             = 0.0;
 181
 182         /* Reset potential sums */
 183         velecsum         = 0.0;
 184         vvdwsum          = 0.0;
 185
 186         /* Start inner kernel loop */
 187         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 188         {
 189             /* Get j neighbor index, and coordinate index */
 190             jnr              = jjnr[jidx];
 191             j_coord_offset   = DIM*jnr;
 192
 193             /* load j atom coordinates */
 194             jx0              = x[j_coord_offset+DIM*0+XX];
 195             jy0              = x[j_coord_offset+DIM*0+YY];
 196             jz0              = x[j_coord_offset+DIM*0+ZZ];
 197
 198             /* Calculate displacement vector */
 199             dx00             = ix0 - jx0;
 200             dy00             = iy0 - jy0;
 201             dz00             = iz0 - jz0;
 202             dx10             = ix1 - jx0;
 203             dy10             = iy1 - jy0;
 204             dz10             = iz1 - jz0;
 205             dx20             = ix2 - jx0;
 206             dy20             = iy2 - jy0;
 207             dz20             = iz2 - jz0;
 208             dx30             = ix3 - jx0;
 209             dy30             = iy3 - jy0;
 210             dz30             = iz3 - jz0;
 211
 212             /* Calculate squared distance and things based on it */
 213             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 214             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 215             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 216             rsq30            = dx30*dx30+dy30*dy30+dz30*dz30;
 217
 218             rinv00           = gmx_invsqrt(rsq00);
 219             rinv10           = gmx_invsqrt(rsq10);
 220             rinv20           = gmx_invsqrt(rsq20);
 221             rinv30           = gmx_invsqrt(rsq30);
 222
 223             rinvsq00         = rinv00*rinv00;
 224             rinvsq10         = rinv10*rinv10;
 225             rinvsq20         = rinv20*rinv20;
 226             rinvsq30         = rinv30*rinv30;
 227
 228             /* Load parameters for j particles */
 229             jq0              = charge[jnr+0];
 230             vdwjidx0         = 3*vdwtype[jnr+0];
 231
 232             /**************************
 233              * CALCULATE INTERACTIONS *
 234              **************************/
 235
 236             if (rsq00<rcutoff2)
 237             {
 238
 239             r00              = rsq00*rinv00;
 240
 241             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 242             cexp1_00         = vdwparam[vdwioffset0+vdwjidx0+1];
 243             cexp2_00         = vdwparam[vdwioffset0+vdwjidx0+2];
 244
 245             /* BUCKINGHAM DISPERSION/REPULSION */
 246             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 247             vvdw6            = c6_00*rinvsix;
 248             br               = cexp2_00*r00;
 249             vvdwexp          = cexp1_00*exp(-br);
 250             vvdw             = vvdwexp - vvdw6*(1.0/6.0);
 251             fvdw             = (br*vvdwexp-vvdw6)*rinvsq00;
 252
 253             d                = r00-rswitch;
 254             d                = (d>0.0) ? d : 0.0;
 255             d2               = d*d;
 256             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 257
 258             dsw              = d2*(swF2+d*(swF3+d*swF4));
 259
 260             /* Evaluate switch function */
 261             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 262             fvdw             = fvdw*sw - rinv00*vvdw*dsw;
 263             vvdw            *= sw;
 264
 265             /* Update potential sums from outer loop */
 266             vvdwsum         += vvdw;
 267
 268             fscal            = fvdw;
 269
 270             /* Calculate temporary vectorial force */
 271             tx               = fscal*dx00;
 272             ty               = fscal*dy00;
 273             tz               = fscal*dz00;
 274
 275             /* Update vectorial force */
 276             fix0            += tx;
 277             fiy0            += ty;
 278             fiz0            += tz;
 279             f[j_coord_offset+DIM*0+XX] -= tx;
 280             f[j_coord_offset+DIM*0+YY] -= ty;
 281             f[j_coord_offset+DIM*0+ZZ] -= tz;
 282
 283             }
 284
 285             /**************************
 286              * CALCULATE INTERACTIONS *
 287              **************************/
 288
 289             if (rsq10<rcutoff2)
 290             {
 291
 292             qq10             = iq1*jq0;
 293
 294             /* REACTION-FIELD ELECTROSTATICS */
 295             velec            = qq10*(rinv10+krf*rsq10-crf);
 296             felec            = qq10*(rinv10*rinvsq10-krf2);
 297
 298             /* Update potential sums from outer loop */
 299             velecsum        += velec;
 300
 301             fscal            = felec;
 302
 303             /* Calculate temporary vectorial force */
 304             tx               = fscal*dx10;
 305             ty               = fscal*dy10;
 306             tz               = fscal*dz10;
 307
 308             /* Update vectorial force */
 309             fix1            += tx;
 310             fiy1            += ty;
 311             fiz1            += tz;
 312             f[j_coord_offset+DIM*0+XX] -= tx;
 313             f[j_coord_offset+DIM*0+YY] -= ty;
 314             f[j_coord_offset+DIM*0+ZZ] -= tz;
 315
 316             }
 317
 318             /**************************
 319              * CALCULATE INTERACTIONS *
 320              **************************/
 321
 322             if (rsq20<rcutoff2)
 323             {
 324
 325             qq20             = iq2*jq0;
 326
 327             /* REACTION-FIELD ELECTROSTATICS */
 328             velec            = qq20*(rinv20+krf*rsq20-crf);
 329             felec            = qq20*(rinv20*rinvsq20-krf2);
 330
 331             /* Update potential sums from outer loop */
 332             velecsum        += velec;
 333
 334             fscal            = felec;
 335
 336             /* Calculate temporary vectorial force */
 337             tx               = fscal*dx20;
 338             ty               = fscal*dy20;
 339             tz               = fscal*dz20;
 340
 341             /* Update vectorial force */
 342             fix2            += tx;
 343             fiy2            += ty;
 344             fiz2            += tz;
 345             f[j_coord_offset+DIM*0+XX] -= tx;
 346             f[j_coord_offset+DIM*0+YY] -= ty;
 347             f[j_coord_offset+DIM*0+ZZ] -= tz;
 348
 349             }
 350
 351             /**************************
 352              * CALCULATE INTERACTIONS *
 353              **************************/
 354
 355             if (rsq30<rcutoff2)
 356             {
 357
 358             qq30             = iq3*jq0;
 359
 360             /* REACTION-FIELD ELECTROSTATICS */
 361             velec            = qq30*(rinv30+krf*rsq30-crf);
 362             felec            = qq30*(rinv30*rinvsq30-krf2);
 363
 364             /* Update potential sums from outer loop */
 365             velecsum        += velec;
 366
 367             fscal            = felec;
 368
 369             /* Calculate temporary vectorial force */
 370             tx               = fscal*dx30;
 371             ty               = fscal*dy30;
 372             tz               = fscal*dz30;
 373
 374             /* Update vectorial force */
 375             fix3            += tx;
 376             fiy3            += ty;
 377             fiz3            += tz;
 378             f[j_coord_offset+DIM*0+XX] -= tx;
 379             f[j_coord_offset+DIM*0+YY] -= ty;
 380             f[j_coord_offset+DIM*0+ZZ] -= tz;
 381
 382             }
 383
 384             /* Inner loop uses 175 flops */
 385         }
 386         /* End of innermost loop */
 387
 388         tx = ty = tz = 0;
 389         f[i_coord_offset+DIM*0+XX] += fix0;
 390         f[i_coord_offset+DIM*0+YY] += fiy0;
 391         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 392         tx                         += fix0;
 393         ty                         += fiy0;
 394         tz                         += fiz0;
 395         f[i_coord_offset+DIM*1+XX] += fix1;
 396         f[i_coord_offset+DIM*1+YY] += fiy1;
 397         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 398         tx                         += fix1;
 399         ty                         += fiy1;
 400         tz                         += fiz1;
 401         f[i_coord_offset+DIM*2+XX] += fix2;
 402         f[i_coord_offset+DIM*2+YY] += fiy2;
 403         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 404         tx                         += fix2;
 405         ty                         += fiy2;
 406         tz                         += fiz2;
 407         f[i_coord_offset+DIM*3+XX] += fix3;
 408         f[i_coord_offset+DIM*3+YY] += fiy3;
 409         f[i_coord_offset+DIM*3+ZZ] += fiz3;
 410         tx                         += fix3;
 411         ty                         += fiy3;
 412         tz                         += fiz3;
 413         fshift[i_shift_offset+XX]  += tx;
 414         fshift[i_shift_offset+YY]  += ty;
 415         fshift[i_shift_offset+ZZ]  += tz;
 416
 417         ggid                        = gid[iidx];
 418         /* Update potential energies */
 419         kernel_data->energygrp_elec[ggid] += velecsum;
 420         kernel_data->energygrp_vdw[ggid] += vvdwsum;
 421
 422         /* Increment number of inner iterations */
 423         inneriter                  += j_index_end - j_index_start;
 424
 425         /* Outer loop uses 41 flops */
 426     }
 427
 428     /* Increment number of outer iterations */
 429     outeriter        += nri;
 430
 431     /* Update outer/inner flops */
 432
 433     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_VF,outeriter*41 + inneriter*175);
 434 }
 435 /*
 436  * Gromacs nonbonded kernel:   nb_kernel_ElecRFCut_VdwBhamSw_GeomW4P1_F_c
 437  * Electrostatics interaction: ReactionField
 438  * VdW interaction:            Buckingham
 439  * Geometry:                   Water4-Particle
 440  * Calculate force/pot:        Force
 441  */
 442 void
 443 nb_kernel_ElecRFCut_VdwBhamSw_GeomW4P1_F_c
 444                     (t_nblist                    * gmx_restrict       nlist,
 445                      rvec                        * gmx_restrict          xx,
 446                      rvec                        * gmx_restrict          ff,
 447                      t_forcerec                  * gmx_restrict          fr,
 448                      t_mdatoms                   * gmx_restrict     mdatoms,
 449                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
 450                      t_nrnb                      * gmx_restrict        nrnb)
 451 {
 452     int              i_shift_offset,i_coord_offset,j_coord_offset;
 453     int              j_index_start,j_index_end;
 454     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
 455     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
 456     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 457     real             *shiftvec,*fshift,*x,*f;
 458     int              vdwioffset0;
 459     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 460     int              vdwioffset1;
 461     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
 462     int              vdwioffset2;
 463     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
 464     int              vdwioffset3;
 465     real             ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
 466     int              vdwjidx0;
 467     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 468     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
 469     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
 470     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
 471     real             dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30,cexp1_30,cexp2_30;
 472     real             velec,felec,velecsum,facel,crf,krf,krf2;
 473     real             *charge;
 474     int              nvdwtype;
 475     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
 476     int              *vdwtype;
 477     real             *vdwparam;
 478     real             rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
 479
 480     x                = xx[0];
 481     f                = ff[0];
 482
 483     nri              = nlist->nri;
 484     iinr             = nlist->iinr;
 485     jindex           = nlist->jindex;
 486     jjnr             = nlist->jjnr;
 487     shiftidx         = nlist->shift;
 488     gid              = nlist->gid;
 489     shiftvec         = fr->shift_vec[0];
 490     fshift           = fr->fshift[0];
 491     facel            = fr->epsfac;
 492     charge           = mdatoms->chargeA;
 493     krf              = fr->ic->k_rf;
 494     krf2             = krf*2.0;
 495     crf              = fr->ic->c_rf;
 496     nvdwtype         = fr->ntype;
 497     vdwparam         = fr->nbfp;
 498     vdwtype          = mdatoms->typeA;
 499
 500     /* Setup water-specific parameters */
 501     inr              = nlist->iinr[0];
 502     iq1              = facel*charge[inr+1];
 503     iq2              = facel*charge[inr+2];
 504     iq3              = facel*charge[inr+3];
 505     vdwioffset0      = 3*nvdwtype*vdwtype[inr+0];
 506
 507     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 508     rcutoff          = fr->rcoulomb;
 509     rcutoff2         = rcutoff*rcutoff;
 510
 511     rswitch          = fr->rvdw_switch;
 512     /* Setup switch parameters */
 513     d                = rcutoff-rswitch;
 514     swV3             = -10.0/(d*d*d);
 515     swV4             =  15.0/(d*d*d*d);
 516     swV5             =  -6.0/(d*d*d*d*d);
 517     swF2             = -30.0/(d*d*d);
 518     swF3             =  60.0/(d*d*d*d);
 519     swF4             = -30.0/(d*d*d*d*d);
 520
 521     outeriter        = 0;
 522     inneriter        = 0;
 523
 524     /* Start outer loop over neighborlists */
 525     for(iidx=0; iidx<nri; iidx++)
 526     {
 527         /* Load shift vector for this list */
 528         i_shift_offset   = DIM*shiftidx[iidx];
 529         shX              = shiftvec[i_shift_offset+XX];
 530         shY              = shiftvec[i_shift_offset+YY];
 531         shZ              = shiftvec[i_shift_offset+ZZ];
 532
 533         /* Load limits for loop over neighbors */
 534         j_index_start    = jindex[iidx];
 535         j_index_end      = jindex[iidx+1];
 536
 537         /* Get outer coordinate index */
 538         inr              = iinr[iidx];
 539         i_coord_offset   = DIM*inr;
 540
 541         /* Load i particle coords and add shift vector */
 542         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 543         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 544         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 545         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 546         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 547         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 548         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 549         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 550         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 551         ix3              = shX + x[i_coord_offset+DIM*3+XX];
 552         iy3              = shY + x[i_coord_offset+DIM*3+YY];
 553         iz3              = shZ + x[i_coord_offset+DIM*3+ZZ];
 554
 555         fix0             = 0.0;
 556         fiy0             = 0.0;
 557         fiz0             = 0.0;
 558         fix1             = 0.0;
 559         fiy1             = 0.0;
 560         fiz1             = 0.0;
 561         fix2             = 0.0;
 562         fiy2             = 0.0;
 563         fiz2             = 0.0;
 564         fix3             = 0.0;
 565         fiy3             = 0.0;
 566         fiz3             = 0.0;
 567
 568         /* Start inner kernel loop */
 569         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 570         {
 571             /* Get j neighbor index, and coordinate index */
 572             jnr              = jjnr[jidx];
 573             j_coord_offset   = DIM*jnr;
 574
 575             /* load j atom coordinates */
 576             jx0              = x[j_coord_offset+DIM*0+XX];
 577             jy0              = x[j_coord_offset+DIM*0+YY];
 578             jz0              = x[j_coord_offset+DIM*0+ZZ];
 579
 580             /* Calculate displacement vector */
 581             dx00             = ix0 - jx0;
 582             dy00             = iy0 - jy0;
 583             dz00             = iz0 - jz0;
 584             dx10             = ix1 - jx0;
 585             dy10             = iy1 - jy0;
 586             dz10             = iz1 - jz0;
 587             dx20             = ix2 - jx0;
 588             dy20             = iy2 - jy0;
 589             dz20             = iz2 - jz0;
 590             dx30             = ix3 - jx0;
 591             dy30             = iy3 - jy0;
 592             dz30             = iz3 - jz0;
 593
 594             /* Calculate squared distance and things based on it */
 595             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 596             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 597             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 598             rsq30            = dx30*dx30+dy30*dy30+dz30*dz30;
 599
 600             rinv00           = gmx_invsqrt(rsq00);
 601             rinv10           = gmx_invsqrt(rsq10);
 602             rinv20           = gmx_invsqrt(rsq20);
 603             rinv30           = gmx_invsqrt(rsq30);
 604
 605             rinvsq00         = rinv00*rinv00;
 606             rinvsq10         = rinv10*rinv10;
 607             rinvsq20         = rinv20*rinv20;
 608             rinvsq30         = rinv30*rinv30;
 609
 610             /* Load parameters for j particles */
 611             jq0              = charge[jnr+0];
 612             vdwjidx0         = 3*vdwtype[jnr+0];
 613
 614             /**************************
 615              * CALCULATE INTERACTIONS *
 616              **************************/
 617
 618             if (rsq00<rcutoff2)
 619             {
 620
 621             r00              = rsq00*rinv00;
 622
 623             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 624             cexp1_00         = vdwparam[vdwioffset0+vdwjidx0+1];
 625             cexp2_00         = vdwparam[vdwioffset0+vdwjidx0+2];
 626
 627             /* BUCKINGHAM DISPERSION/REPULSION */
 628             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 629             vvdw6            = c6_00*rinvsix;
 630             br               = cexp2_00*r00;
 631             vvdwexp          = cexp1_00*exp(-br);
 632             vvdw             = vvdwexp - vvdw6*(1.0/6.0);
 633             fvdw             = (br*vvdwexp-vvdw6)*rinvsq00;
 634
 635             d                = r00-rswitch;
 636             d                = (d>0.0) ? d : 0.0;
 637             d2               = d*d;
 638             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 639
 640             dsw              = d2*(swF2+d*(swF3+d*swF4));
 641
 642             /* Evaluate switch function */
 643             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 644             fvdw             = fvdw*sw - rinv00*vvdw*dsw;
 645
 646             fscal            = fvdw;
 647
 648             /* Calculate temporary vectorial force */
 649             tx               = fscal*dx00;
 650             ty               = fscal*dy00;
 651             tz               = fscal*dz00;
 652
 653             /* Update vectorial force */
 654             fix0            += tx;
 655             fiy0            += ty;
 656             fiz0            += tz;
 657             f[j_coord_offset+DIM*0+XX] -= tx;
 658             f[j_coord_offset+DIM*0+YY] -= ty;
 659             f[j_coord_offset+DIM*0+ZZ] -= tz;
 660
 661             }
 662
 663             /**************************
 664              * CALCULATE INTERACTIONS *
 665              **************************/
 666
 667             if (rsq10<rcutoff2)
 668             {
 669
 670             qq10             = iq1*jq0;
 671
 672             /* REACTION-FIELD ELECTROSTATICS */
 673             felec            = qq10*(rinv10*rinvsq10-krf2);
 674
 675             fscal            = felec;
 676
 677             /* Calculate temporary vectorial force */
 678             tx               = fscal*dx10;
 679             ty               = fscal*dy10;
 680             tz               = fscal*dz10;
 681
 682             /* Update vectorial force */
 683             fix1            += tx;
 684             fiy1            += ty;
 685             fiz1            += tz;
 686             f[j_coord_offset+DIM*0+XX] -= tx;
 687             f[j_coord_offset+DIM*0+YY] -= ty;
 688             f[j_coord_offset+DIM*0+ZZ] -= tz;
 689
 690             }
 691
 692             /**************************
 693              * CALCULATE INTERACTIONS *
 694              **************************/
 695
 696             if (rsq20<rcutoff2)
 697             {
 698
 699             qq20             = iq2*jq0;
 700
 701             /* REACTION-FIELD ELECTROSTATICS */
 702             felec            = qq20*(rinv20*rinvsq20-krf2);
 703
 704             fscal            = felec;
 705
 706             /* Calculate temporary vectorial force */
 707             tx               = fscal*dx20;
 708             ty               = fscal*dy20;
 709             tz               = fscal*dz20;
 710
 711             /* Update vectorial force */
 712             fix2            += tx;
 713             fiy2            += ty;
 714             fiz2            += tz;
 715             f[j_coord_offset+DIM*0+XX] -= tx;
 716             f[j_coord_offset+DIM*0+YY] -= ty;
 717             f[j_coord_offset+DIM*0+ZZ] -= tz;
 718
 719             }
 720
 721             /**************************
 722              * CALCULATE INTERACTIONS *
 723              **************************/
 724
 725             if (rsq30<rcutoff2)
 726             {
 727
 728             qq30             = iq3*jq0;
 729
 730             /* REACTION-FIELD ELECTROSTATICS */
 731             felec            = qq30*(rinv30*rinvsq30-krf2);
 732
 733             fscal            = felec;
 734
 735             /* Calculate temporary vectorial force */
 736             tx               = fscal*dx30;
 737             ty               = fscal*dy30;
 738             tz               = fscal*dz30;
 739
 740             /* Update vectorial force */
 741             fix3            += tx;
 742             fiy3            += ty;
 743             fiz3            += tz;
 744             f[j_coord_offset+DIM*0+XX] -= tx;
 745             f[j_coord_offset+DIM*0+YY] -= ty;
 746             f[j_coord_offset+DIM*0+ZZ] -= tz;
 747
 748             }
 749
 750             /* Inner loop uses 158 flops */
 751         }
 752         /* End of innermost loop */
 753
 754         tx = ty = tz = 0;
 755         f[i_coord_offset+DIM*0+XX] += fix0;
 756         f[i_coord_offset+DIM*0+YY] += fiy0;
 757         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 758         tx                         += fix0;
 759         ty                         += fiy0;
 760         tz                         += fiz0;
 761         f[i_coord_offset+DIM*1+XX] += fix1;
 762         f[i_coord_offset+DIM*1+YY] += fiy1;
 763         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 764         tx                         += fix1;
 765         ty                         += fiy1;
 766         tz                         += fiz1;
 767         f[i_coord_offset+DIM*2+XX] += fix2;
 768         f[i_coord_offset+DIM*2+YY] += fiy2;
 769         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 770         tx                         += fix2;
 771         ty                         += fiy2;
 772         tz                         += fiz2;
 773         f[i_coord_offset+DIM*3+XX] += fix3;
 774         f[i_coord_offset+DIM*3+YY] += fiy3;
 775         f[i_coord_offset+DIM*3+ZZ] += fiz3;
 776         tx                         += fix3;
 777         ty                         += fiy3;
 778         tz                         += fiz3;
 779         fshift[i_shift_offset+XX]  += tx;
 780         fshift[i_shift_offset+YY]  += ty;
 781         fshift[i_shift_offset+ZZ]  += tz;
 782
 783         /* Increment number of inner iterations */
 784         inneriter                  += j_index_end - j_index_start;
 785
 786         /* Outer loop uses 39 flops */
 787     }
 788
 789     /* Increment number of outer iterations */
 790     outeriter        += nri;
 791
 792     /* Update outer/inner flops */
 793
 794     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_F,outeriter*39 + inneriter*158);
 795 }