src/gromacs/gmxlib/nonbonded/nb_kernel_c/nb_kernel_ElecRFCut_VdwLJSw_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 "config.h"
  39
  40 #include <math.h>
  41
  42 #include "../nb_kernel.h"
  43 #include "gromacs/legacyheaders/types/simple.h"
  44 #include "gromacs/math/vec.h"
  45 #include "gromacs/legacyheaders/nrnb.h"
  46
  47 /*
  48  * Gromacs nonbonded kernel:   nb_kernel_ElecRFCut_VdwLJSw_GeomW4P1_VF_c
  49  * Electrostatics interaction: ReactionField
  50  * VdW interaction:            LennardJones
  51  * Geometry:                   Water4-Particle
  52  * Calculate force/pot:        PotentialAndForce
  53  */
  54 void
  55 nb_kernel_ElecRFCut_VdwLJSw_GeomW4P1_VF_c
  56                     (t_nblist                    * gmx_restrict       nlist,
  57                      rvec                        * gmx_restrict          xx,
  58                      rvec                        * gmx_restrict          ff,
  59                      t_forcerec                  * gmx_restrict          fr,
  60                      t_mdatoms                   * gmx_restrict     mdatoms,
  61                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
  62                      t_nrnb                      * gmx_restrict        nrnb)
  63 {
  64     int              i_shift_offset,i_coord_offset,j_coord_offset;
  65     int              j_index_start,j_index_end;
  66     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
  67     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
  68     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
  69     real             *shiftvec,*fshift,*x,*f;
  70     int              vdwioffset0;
  71     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
  72     int              vdwioffset1;
  73     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
  74     int              vdwioffset2;
  75     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
  76     int              vdwioffset3;
  77     real             ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
  78     int              vdwjidx0;
  79     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
  80     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
  81     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
  82     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
  83     real             dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30,cexp1_30,cexp2_30;
  84     real             velec,felec,velecsum,facel,crf,krf,krf2;
  85     real             *charge;
  86     int              nvdwtype;
  87     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
  88     int              *vdwtype;
  89     real             *vdwparam;
  90     real             rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
  91
  92     x                = xx[0];
  93     f                = ff[0];
  94
  95     nri              = nlist->nri;
  96     iinr             = nlist->iinr;
  97     jindex           = nlist->jindex;
  98     jjnr             = nlist->jjnr;
  99     shiftidx         = nlist->shift;
 100     gid              = nlist->gid;
 101     shiftvec         = fr->shift_vec[0];
 102     fshift           = fr->fshift[0];
 103     facel            = fr->epsfac;
 104     charge           = mdatoms->chargeA;
 105     krf              = fr->ic->k_rf;
 106     krf2             = krf*2.0;
 107     crf              = fr->ic->c_rf;
 108     nvdwtype         = fr->ntype;
 109     vdwparam         = fr->nbfp;
 110     vdwtype          = mdatoms->typeA;
 111
 112     /* Setup water-specific parameters */
 113     inr              = nlist->iinr[0];
 114     iq1              = facel*charge[inr+1];
 115     iq2              = facel*charge[inr+2];
 116     iq3              = facel*charge[inr+3];
 117     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 118
 119     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 120     rcutoff          = fr->rcoulomb;
 121     rcutoff2         = rcutoff*rcutoff;
 122
 123     rswitch          = fr->rvdw_switch;
 124     /* Setup switch parameters */
 125     d                = rcutoff-rswitch;
 126     swV3             = -10.0/(d*d*d);
 127     swV4             =  15.0/(d*d*d*d);
 128     swV5             =  -6.0/(d*d*d*d*d);
 129     swF2             = -30.0/(d*d*d);
 130     swF3             =  60.0/(d*d*d*d);
 131     swF4             = -30.0/(d*d*d*d*d);
 132
 133     outeriter        = 0;
 134     inneriter        = 0;
 135
 136     /* Start outer loop over neighborlists */
 137     for(iidx=0; iidx<nri; iidx++)
 138     {
 139         /* Load shift vector for this list */
 140         i_shift_offset   = DIM*shiftidx[iidx];
 141         shX              = shiftvec[i_shift_offset+XX];
 142         shY              = shiftvec[i_shift_offset+YY];
 143         shZ              = shiftvec[i_shift_offset+ZZ];
 144
 145         /* Load limits for loop over neighbors */
 146         j_index_start    = jindex[iidx];
 147         j_index_end      = jindex[iidx+1];
 148
 149         /* Get outer coordinate index */
 150         inr              = iinr[iidx];
 151         i_coord_offset   = DIM*inr;
 152
 153         /* Load i particle coords and add shift vector */
 154         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 155         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 156         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 157         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 158         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 159         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 160         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 161         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 162         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 163         ix3              = shX + x[i_coord_offset+DIM*3+XX];
 164         iy3              = shY + x[i_coord_offset+DIM*3+YY];
 165         iz3              = shZ + x[i_coord_offset+DIM*3+ZZ];
 166
 167         fix0             = 0.0;
 168         fiy0             = 0.0;
 169         fiz0             = 0.0;
 170         fix1             = 0.0;
 171         fiy1             = 0.0;
 172         fiz1             = 0.0;
 173         fix2             = 0.0;
 174         fiy2             = 0.0;
 175         fiz2             = 0.0;
 176         fix3             = 0.0;
 177         fiy3             = 0.0;
 178         fiz3             = 0.0;
 179
 180         /* Reset potential sums */
 181         velecsum         = 0.0;
 182         vvdwsum          = 0.0;
 183
 184         /* Start inner kernel loop */
 185         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 186         {
 187             /* Get j neighbor index, and coordinate index */
 188             jnr              = jjnr[jidx];
 189             j_coord_offset   = DIM*jnr;
 190
 191             /* load j atom coordinates */
 192             jx0              = x[j_coord_offset+DIM*0+XX];
 193             jy0              = x[j_coord_offset+DIM*0+YY];
 194             jz0              = x[j_coord_offset+DIM*0+ZZ];
 195
 196             /* Calculate displacement vector */
 197             dx00             = ix0 - jx0;
 198             dy00             = iy0 - jy0;
 199             dz00             = iz0 - jz0;
 200             dx10             = ix1 - jx0;
 201             dy10             = iy1 - jy0;
 202             dz10             = iz1 - jz0;
 203             dx20             = ix2 - jx0;
 204             dy20             = iy2 - jy0;
 205             dz20             = iz2 - jz0;
 206             dx30             = ix3 - jx0;
 207             dy30             = iy3 - jy0;
 208             dz30             = iz3 - jz0;
 209
 210             /* Calculate squared distance and things based on it */
 211             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 212             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 213             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 214             rsq30            = dx30*dx30+dy30*dy30+dz30*dz30;
 215
 216             rinv00           = gmx_invsqrt(rsq00);
 217             rinv10           = gmx_invsqrt(rsq10);
 218             rinv20           = gmx_invsqrt(rsq20);
 219             rinv30           = gmx_invsqrt(rsq30);
 220
 221             rinvsq00         = rinv00*rinv00;
 222             rinvsq10         = rinv10*rinv10;
 223             rinvsq20         = rinv20*rinv20;
 224             rinvsq30         = rinv30*rinv30;
 225
 226             /* Load parameters for j particles */
 227             jq0              = charge[jnr+0];
 228             vdwjidx0         = 2*vdwtype[jnr+0];
 229
 230             /**************************
 231              * CALCULATE INTERACTIONS *
 232              **************************/
 233
 234             if (rsq00<rcutoff2)
 235             {
 236
 237             r00              = rsq00*rinv00;
 238
 239             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 240             c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 241
 242             /* LENNARD-JONES DISPERSION/REPULSION */
 243
 244             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 245             vvdw6            = c6_00*rinvsix;
 246             vvdw12           = c12_00*rinvsix*rinvsix;
 247             vvdw             = vvdw12*(1.0/12.0) - vvdw6*(1.0/6.0);
 248             fvdw             = (vvdw12-vvdw6)*rinvsq00;
 249
 250             d                = r00-rswitch;
 251             d                = (d>0.0) ? d : 0.0;
 252             d2               = d*d;
 253             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 254
 255             dsw              = d2*(swF2+d*(swF3+d*swF4));
 256
 257             /* Evaluate switch function */
 258             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 259             fvdw             = fvdw*sw - rinv00*vvdw*dsw;
 260             vvdw            *= sw;
 261
 262             /* Update potential sums from outer loop */
 263             vvdwsum         += vvdw;
 264
 265             fscal            = fvdw;
 266
 267             /* Calculate temporary vectorial force */
 268             tx               = fscal*dx00;
 269             ty               = fscal*dy00;
 270             tz               = fscal*dz00;
 271
 272             /* Update vectorial force */
 273             fix0            += tx;
 274             fiy0            += ty;
 275             fiz0            += tz;
 276             f[j_coord_offset+DIM*0+XX] -= tx;
 277             f[j_coord_offset+DIM*0+YY] -= ty;
 278             f[j_coord_offset+DIM*0+ZZ] -= tz;
 279
 280             }
 281
 282             /**************************
 283              * CALCULATE INTERACTIONS *
 284              **************************/
 285
 286             if (rsq10<rcutoff2)
 287             {
 288
 289             qq10             = iq1*jq0;
 290
 291             /* REACTION-FIELD ELECTROSTATICS */
 292             velec            = qq10*(rinv10+krf*rsq10-crf);
 293             felec            = qq10*(rinv10*rinvsq10-krf2);
 294
 295             /* Update potential sums from outer loop */
 296             velecsum        += velec;
 297
 298             fscal            = felec;
 299
 300             /* Calculate temporary vectorial force */
 301             tx               = fscal*dx10;
 302             ty               = fscal*dy10;
 303             tz               = fscal*dz10;
 304
 305             /* Update vectorial force */
 306             fix1            += tx;
 307             fiy1            += ty;
 308             fiz1            += tz;
 309             f[j_coord_offset+DIM*0+XX] -= tx;
 310             f[j_coord_offset+DIM*0+YY] -= ty;
 311             f[j_coord_offset+DIM*0+ZZ] -= tz;
 312
 313             }
 314
 315             /**************************
 316              * CALCULATE INTERACTIONS *
 317              **************************/
 318
 319             if (rsq20<rcutoff2)
 320             {
 321
 322             qq20             = iq2*jq0;
 323
 324             /* REACTION-FIELD ELECTROSTATICS */
 325             velec            = qq20*(rinv20+krf*rsq20-crf);
 326             felec            = qq20*(rinv20*rinvsq20-krf2);
 327
 328             /* Update potential sums from outer loop */
 329             velecsum        += velec;
 330
 331             fscal            = felec;
 332
 333             /* Calculate temporary vectorial force */
 334             tx               = fscal*dx20;
 335             ty               = fscal*dy20;
 336             tz               = fscal*dz20;
 337
 338             /* Update vectorial force */
 339             fix2            += tx;
 340             fiy2            += ty;
 341             fiz2            += tz;
 342             f[j_coord_offset+DIM*0+XX] -= tx;
 343             f[j_coord_offset+DIM*0+YY] -= ty;
 344             f[j_coord_offset+DIM*0+ZZ] -= tz;
 345
 346             }
 347
 348             /**************************
 349              * CALCULATE INTERACTIONS *
 350              **************************/
 351
 352             if (rsq30<rcutoff2)
 353             {
 354
 355             qq30             = iq3*jq0;
 356
 357             /* REACTION-FIELD ELECTROSTATICS */
 358             velec            = qq30*(rinv30+krf*rsq30-crf);
 359             felec            = qq30*(rinv30*rinvsq30-krf2);
 360
 361             /* Update potential sums from outer loop */
 362             velecsum        += velec;
 363
 364             fscal            = felec;
 365
 366             /* Calculate temporary vectorial force */
 367             tx               = fscal*dx30;
 368             ty               = fscal*dy30;
 369             tz               = fscal*dz30;
 370
 371             /* Update vectorial force */
 372             fix3            += tx;
 373             fiy3            += ty;
 374             fiz3            += tz;
 375             f[j_coord_offset+DIM*0+XX] -= tx;
 376             f[j_coord_offset+DIM*0+YY] -= ty;
 377             f[j_coord_offset+DIM*0+ZZ] -= tz;
 378
 379             }
 380
 381             /* Inner loop uses 149 flops */
 382         }
 383         /* End of innermost loop */
 384
 385         tx = ty = tz = 0;
 386         f[i_coord_offset+DIM*0+XX] += fix0;
 387         f[i_coord_offset+DIM*0+YY] += fiy0;
 388         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 389         tx                         += fix0;
 390         ty                         += fiy0;
 391         tz                         += fiz0;
 392         f[i_coord_offset+DIM*1+XX] += fix1;
 393         f[i_coord_offset+DIM*1+YY] += fiy1;
 394         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 395         tx                         += fix1;
 396         ty                         += fiy1;
 397         tz                         += fiz1;
 398         f[i_coord_offset+DIM*2+XX] += fix2;
 399         f[i_coord_offset+DIM*2+YY] += fiy2;
 400         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 401         tx                         += fix2;
 402         ty                         += fiy2;
 403         tz                         += fiz2;
 404         f[i_coord_offset+DIM*3+XX] += fix3;
 405         f[i_coord_offset+DIM*3+YY] += fiy3;
 406         f[i_coord_offset+DIM*3+ZZ] += fiz3;
 407         tx                         += fix3;
 408         ty                         += fiy3;
 409         tz                         += fiz3;
 410         fshift[i_shift_offset+XX]  += tx;
 411         fshift[i_shift_offset+YY]  += ty;
 412         fshift[i_shift_offset+ZZ]  += tz;
 413
 414         ggid                        = gid[iidx];
 415         /* Update potential energies */
 416         kernel_data->energygrp_elec[ggid] += velecsum;
 417         kernel_data->energygrp_vdw[ggid] += vvdwsum;
 418
 419         /* Increment number of inner iterations */
 420         inneriter                  += j_index_end - j_index_start;
 421
 422         /* Outer loop uses 41 flops */
 423     }
 424
 425     /* Increment number of outer iterations */
 426     outeriter        += nri;
 427
 428     /* Update outer/inner flops */
 429
 430     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_VF,outeriter*41 + inneriter*149);
 431 }
 432 /*
 433  * Gromacs nonbonded kernel:   nb_kernel_ElecRFCut_VdwLJSw_GeomW4P1_F_c
 434  * Electrostatics interaction: ReactionField
 435  * VdW interaction:            LennardJones
 436  * Geometry:                   Water4-Particle
 437  * Calculate force/pot:        Force
 438  */
 439 void
 440 nb_kernel_ElecRFCut_VdwLJSw_GeomW4P1_F_c
 441                     (t_nblist                    * gmx_restrict       nlist,
 442                      rvec                        * gmx_restrict          xx,
 443                      rvec                        * gmx_restrict          ff,
 444                      t_forcerec                  * gmx_restrict          fr,
 445                      t_mdatoms                   * gmx_restrict     mdatoms,
 446                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
 447                      t_nrnb                      * gmx_restrict        nrnb)
 448 {
 449     int              i_shift_offset,i_coord_offset,j_coord_offset;
 450     int              j_index_start,j_index_end;
 451     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
 452     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
 453     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 454     real             *shiftvec,*fshift,*x,*f;
 455     int              vdwioffset0;
 456     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 457     int              vdwioffset1;
 458     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
 459     int              vdwioffset2;
 460     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
 461     int              vdwioffset3;
 462     real             ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
 463     int              vdwjidx0;
 464     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 465     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
 466     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
 467     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
 468     real             dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30,cexp1_30,cexp2_30;
 469     real             velec,felec,velecsum,facel,crf,krf,krf2;
 470     real             *charge;
 471     int              nvdwtype;
 472     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
 473     int              *vdwtype;
 474     real             *vdwparam;
 475     real             rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
 476
 477     x                = xx[0];
 478     f                = ff[0];
 479
 480     nri              = nlist->nri;
 481     iinr             = nlist->iinr;
 482     jindex           = nlist->jindex;
 483     jjnr             = nlist->jjnr;
 484     shiftidx         = nlist->shift;
 485     gid              = nlist->gid;
 486     shiftvec         = fr->shift_vec[0];
 487     fshift           = fr->fshift[0];
 488     facel            = fr->epsfac;
 489     charge           = mdatoms->chargeA;
 490     krf              = fr->ic->k_rf;
 491     krf2             = krf*2.0;
 492     crf              = fr->ic->c_rf;
 493     nvdwtype         = fr->ntype;
 494     vdwparam         = fr->nbfp;
 495     vdwtype          = mdatoms->typeA;
 496
 497     /* Setup water-specific parameters */
 498     inr              = nlist->iinr[0];
 499     iq1              = facel*charge[inr+1];
 500     iq2              = facel*charge[inr+2];
 501     iq3              = facel*charge[inr+3];
 502     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 503
 504     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 505     rcutoff          = fr->rcoulomb;
 506     rcutoff2         = rcutoff*rcutoff;
 507
 508     rswitch          = fr->rvdw_switch;
 509     /* Setup switch parameters */
 510     d                = rcutoff-rswitch;
 511     swV3             = -10.0/(d*d*d);
 512     swV4             =  15.0/(d*d*d*d);
 513     swV5             =  -6.0/(d*d*d*d*d);
 514     swF2             = -30.0/(d*d*d);
 515     swF3             =  60.0/(d*d*d*d);
 516     swF4             = -30.0/(d*d*d*d*d);
 517
 518     outeriter        = 0;
 519     inneriter        = 0;
 520
 521     /* Start outer loop over neighborlists */
 522     for(iidx=0; iidx<nri; iidx++)
 523     {
 524         /* Load shift vector for this list */
 525         i_shift_offset   = DIM*shiftidx[iidx];
 526         shX              = shiftvec[i_shift_offset+XX];
 527         shY              = shiftvec[i_shift_offset+YY];
 528         shZ              = shiftvec[i_shift_offset+ZZ];
 529
 530         /* Load limits for loop over neighbors */
 531         j_index_start    = jindex[iidx];
 532         j_index_end      = jindex[iidx+1];
 533
 534         /* Get outer coordinate index */
 535         inr              = iinr[iidx];
 536         i_coord_offset   = DIM*inr;
 537
 538         /* Load i particle coords and add shift vector */
 539         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 540         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 541         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 542         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 543         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 544         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 545         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 546         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 547         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 548         ix3              = shX + x[i_coord_offset+DIM*3+XX];
 549         iy3              = shY + x[i_coord_offset+DIM*3+YY];
 550         iz3              = shZ + x[i_coord_offset+DIM*3+ZZ];
 551
 552         fix0             = 0.0;
 553         fiy0             = 0.0;
 554         fiz0             = 0.0;
 555         fix1             = 0.0;
 556         fiy1             = 0.0;
 557         fiz1             = 0.0;
 558         fix2             = 0.0;
 559         fiy2             = 0.0;
 560         fiz2             = 0.0;
 561         fix3             = 0.0;
 562         fiy3             = 0.0;
 563         fiz3             = 0.0;
 564
 565         /* Start inner kernel loop */
 566         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 567         {
 568             /* Get j neighbor index, and coordinate index */
 569             jnr              = jjnr[jidx];
 570             j_coord_offset   = DIM*jnr;
 571
 572             /* load j atom coordinates */
 573             jx0              = x[j_coord_offset+DIM*0+XX];
 574             jy0              = x[j_coord_offset+DIM*0+YY];
 575             jz0              = x[j_coord_offset+DIM*0+ZZ];
 576
 577             /* Calculate displacement vector */
 578             dx00             = ix0 - jx0;
 579             dy00             = iy0 - jy0;
 580             dz00             = iz0 - jz0;
 581             dx10             = ix1 - jx0;
 582             dy10             = iy1 - jy0;
 583             dz10             = iz1 - jz0;
 584             dx20             = ix2 - jx0;
 585             dy20             = iy2 - jy0;
 586             dz20             = iz2 - jz0;
 587             dx30             = ix3 - jx0;
 588             dy30             = iy3 - jy0;
 589             dz30             = iz3 - jz0;
 590
 591             /* Calculate squared distance and things based on it */
 592             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 593             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 594             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 595             rsq30            = dx30*dx30+dy30*dy30+dz30*dz30;
 596
 597             rinv00           = gmx_invsqrt(rsq00);
 598             rinv10           = gmx_invsqrt(rsq10);
 599             rinv20           = gmx_invsqrt(rsq20);
 600             rinv30           = gmx_invsqrt(rsq30);
 601
 602             rinvsq00         = rinv00*rinv00;
 603             rinvsq10         = rinv10*rinv10;
 604             rinvsq20         = rinv20*rinv20;
 605             rinvsq30         = rinv30*rinv30;
 606
 607             /* Load parameters for j particles */
 608             jq0              = charge[jnr+0];
 609             vdwjidx0         = 2*vdwtype[jnr+0];
 610
 611             /**************************
 612              * CALCULATE INTERACTIONS *
 613              **************************/
 614
 615             if (rsq00<rcutoff2)
 616             {
 617
 618             r00              = rsq00*rinv00;
 619
 620             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 621             c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 622
 623             /* LENNARD-JONES DISPERSION/REPULSION */
 624
 625             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 626             vvdw6            = c6_00*rinvsix;
 627             vvdw12           = c12_00*rinvsix*rinvsix;
 628             vvdw             = vvdw12*(1.0/12.0) - vvdw6*(1.0/6.0);
 629             fvdw             = (vvdw12-vvdw6)*rinvsq00;
 630
 631             d                = r00-rswitch;
 632             d                = (d>0.0) ? d : 0.0;
 633             d2               = d*d;
 634             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 635
 636             dsw              = d2*(swF2+d*(swF3+d*swF4));
 637
 638             /* Evaluate switch function */
 639             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 640             fvdw             = fvdw*sw - rinv00*vvdw*dsw;
 641
 642             fscal            = fvdw;
 643
 644             /* Calculate temporary vectorial force */
 645             tx               = fscal*dx00;
 646             ty               = fscal*dy00;
 647             tz               = fscal*dz00;
 648
 649             /* Update vectorial force */
 650             fix0            += tx;
 651             fiy0            += ty;
 652             fiz0            += tz;
 653             f[j_coord_offset+DIM*0+XX] -= tx;
 654             f[j_coord_offset+DIM*0+YY] -= ty;
 655             f[j_coord_offset+DIM*0+ZZ] -= tz;
 656
 657             }
 658
 659             /**************************
 660              * CALCULATE INTERACTIONS *
 661              **************************/
 662
 663             if (rsq10<rcutoff2)
 664             {
 665
 666             qq10             = iq1*jq0;
 667
 668             /* REACTION-FIELD ELECTROSTATICS */
 669             felec            = qq10*(rinv10*rinvsq10-krf2);
 670
 671             fscal            = felec;
 672
 673             /* Calculate temporary vectorial force */
 674             tx               = fscal*dx10;
 675             ty               = fscal*dy10;
 676             tz               = fscal*dz10;
 677
 678             /* Update vectorial force */
 679             fix1            += tx;
 680             fiy1            += ty;
 681             fiz1            += tz;
 682             f[j_coord_offset+DIM*0+XX] -= tx;
 683             f[j_coord_offset+DIM*0+YY] -= ty;
 684             f[j_coord_offset+DIM*0+ZZ] -= tz;
 685
 686             }
 687
 688             /**************************
 689              * CALCULATE INTERACTIONS *
 690              **************************/
 691
 692             if (rsq20<rcutoff2)
 693             {
 694
 695             qq20             = iq2*jq0;
 696
 697             /* REACTION-FIELD ELECTROSTATICS */
 698             felec            = qq20*(rinv20*rinvsq20-krf2);
 699
 700             fscal            = felec;
 701
 702             /* Calculate temporary vectorial force */
 703             tx               = fscal*dx20;
 704             ty               = fscal*dy20;
 705             tz               = fscal*dz20;
 706
 707             /* Update vectorial force */
 708             fix2            += tx;
 709             fiy2            += ty;
 710             fiz2            += tz;
 711             f[j_coord_offset+DIM*0+XX] -= tx;
 712             f[j_coord_offset+DIM*0+YY] -= ty;
 713             f[j_coord_offset+DIM*0+ZZ] -= tz;
 714
 715             }
 716
 717             /**************************
 718              * CALCULATE INTERACTIONS *
 719              **************************/
 720
 721             if (rsq30<rcutoff2)
 722             {
 723
 724             qq30             = iq3*jq0;
 725
 726             /* REACTION-FIELD ELECTROSTATICS */
 727             felec            = qq30*(rinv30*rinvsq30-krf2);
 728
 729             fscal            = felec;
 730
 731             /* Calculate temporary vectorial force */
 732             tx               = fscal*dx30;
 733             ty               = fscal*dy30;
 734             tz               = fscal*dz30;
 735
 736             /* Update vectorial force */
 737             fix3            += tx;
 738             fiy3            += ty;
 739             fiz3            += tz;
 740             f[j_coord_offset+DIM*0+XX] -= tx;
 741             f[j_coord_offset+DIM*0+YY] -= ty;
 742             f[j_coord_offset+DIM*0+ZZ] -= tz;
 743
 744             }
 745
 746             /* Inner loop uses 132 flops */
 747         }
 748         /* End of innermost loop */
 749
 750         tx = ty = tz = 0;
 751         f[i_coord_offset+DIM*0+XX] += fix0;
 752         f[i_coord_offset+DIM*0+YY] += fiy0;
 753         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 754         tx                         += fix0;
 755         ty                         += fiy0;
 756         tz                         += fiz0;
 757         f[i_coord_offset+DIM*1+XX] += fix1;
 758         f[i_coord_offset+DIM*1+YY] += fiy1;
 759         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 760         tx                         += fix1;
 761         ty                         += fiy1;
 762         tz                         += fiz1;
 763         f[i_coord_offset+DIM*2+XX] += fix2;
 764         f[i_coord_offset+DIM*2+YY] += fiy2;
 765         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 766         tx                         += fix2;
 767         ty                         += fiy2;
 768         tz                         += fiz2;
 769         f[i_coord_offset+DIM*3+XX] += fix3;
 770         f[i_coord_offset+DIM*3+YY] += fiy3;
 771         f[i_coord_offset+DIM*3+ZZ] += fiz3;
 772         tx                         += fix3;
 773         ty                         += fiy3;
 774         tz                         += fiz3;
 775         fshift[i_shift_offset+XX]  += tx;
 776         fshift[i_shift_offset+YY]  += ty;
 777         fshift[i_shift_offset+ZZ]  += tz;
 778
 779         /* Increment number of inner iterations */
 780         inneriter                  += j_index_end - j_index_start;
 781
 782         /* Outer loop uses 39 flops */
 783     }
 784
 785     /* Increment number of outer iterations */
 786     outeriter        += nri;
 787
 788     /* Update outer/inner flops */
 789
 790     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_F,outeriter*39 + inneriter*132);
 791 }