src/gromacs/gmxlib/nonbonded/nb_kernel_c/nb_kernel_ElecEwSw_VdwBhamSw_GeomW3P1_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_ElecEwSw_VdwBhamSw_GeomW3P1_VF_c
  49  * Electrostatics interaction: Ewald
  50  * VdW interaction:            Buckingham
  51  * Geometry:                   Water3-Particle
  52  * Calculate force/pot:        PotentialAndForce
  53  */
  54 void
  55 nb_kernel_ElecEwSw_VdwBhamSw_GeomW3P1_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              vdwjidx0;
  77     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
  78     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
  79     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
  80     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
  81     real             velec,felec,velecsum,facel,crf,krf,krf2;
  82     real             *charge;
  83     int              nvdwtype;
  84     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
  85     int              *vdwtype;
  86     real             *vdwparam;
  87     int              ewitab;
  88     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
  89     real             *ewtab;
  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     nvdwtype         = fr->ntype;
 106     vdwparam         = fr->nbfp;
 107     vdwtype          = mdatoms->typeA;
 108
 109     sh_ewald         = fr->ic->sh_ewald;
 110     ewtab            = fr->ic->tabq_coul_FDV0;
 111     ewtabscale       = fr->ic->tabq_scale;
 112     ewtabhalfspace   = 0.5/ewtabscale;
 113
 114     /* Setup water-specific parameters */
 115     inr              = nlist->iinr[0];
 116     iq0              = facel*charge[inr+0];
 117     iq1              = facel*charge[inr+1];
 118     iq2              = facel*charge[inr+2];
 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->rcoulomb_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
 166         fix0             = 0.0;
 167         fiy0             = 0.0;
 168         fiz0             = 0.0;
 169         fix1             = 0.0;
 170         fiy1             = 0.0;
 171         fiz1             = 0.0;
 172         fix2             = 0.0;
 173         fiy2             = 0.0;
 174         fiz2             = 0.0;
 175
 176         /* Reset potential sums */
 177         velecsum         = 0.0;
 178         vvdwsum          = 0.0;
 179
 180         /* Start inner kernel loop */
 181         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 182         {
 183             /* Get j neighbor index, and coordinate index */
 184             jnr              = jjnr[jidx];
 185             j_coord_offset   = DIM*jnr;
 186
 187             /* load j atom coordinates */
 188             jx0              = x[j_coord_offset+DIM*0+XX];
 189             jy0              = x[j_coord_offset+DIM*0+YY];
 190             jz0              = x[j_coord_offset+DIM*0+ZZ];
 191
 192             /* Calculate displacement vector */
 193             dx00             = ix0 - jx0;
 194             dy00             = iy0 - jy0;
 195             dz00             = iz0 - jz0;
 196             dx10             = ix1 - jx0;
 197             dy10             = iy1 - jy0;
 198             dz10             = iz1 - jz0;
 199             dx20             = ix2 - jx0;
 200             dy20             = iy2 - jy0;
 201             dz20             = iz2 - jz0;
 202
 203             /* Calculate squared distance and things based on it */
 204             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 205             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 206             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 207
 208             rinv00           = gmx_invsqrt(rsq00);
 209             rinv10           = gmx_invsqrt(rsq10);
 210             rinv20           = gmx_invsqrt(rsq20);
 211
 212             rinvsq00         = rinv00*rinv00;
 213             rinvsq10         = rinv10*rinv10;
 214             rinvsq20         = rinv20*rinv20;
 215
 216             /* Load parameters for j particles */
 217             jq0              = charge[jnr+0];
 218             vdwjidx0         = 3*vdwtype[jnr+0];
 219
 220             /**************************
 221              * CALCULATE INTERACTIONS *
 222              **************************/
 223
 224             if (rsq00<rcutoff2)
 225             {
 226
 227             r00              = rsq00*rinv00;
 228
 229             qq00             = iq0*jq0;
 230             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 231             cexp1_00         = vdwparam[vdwioffset0+vdwjidx0+1];
 232             cexp2_00         = vdwparam[vdwioffset0+vdwjidx0+2];
 233
 234             /* EWALD ELECTROSTATICS */
 235
 236             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 237             ewrt             = r00*ewtabscale;
 238             ewitab           = ewrt;
 239             eweps            = ewrt-ewitab;
 240             ewitab           = 4*ewitab;
 241             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 242             velec            = qq00*(rinv00-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 243             felec            = qq00*rinv00*(rinvsq00-felec);
 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             felec            = felec*sw - rinv00*velec*dsw;
 263             fvdw             = fvdw*sw - rinv00*vvdw*dsw;
 264             velec           *= sw;
 265             vvdw            *= sw;
 266
 267             /* Update potential sums from outer loop */
 268             velecsum        += velec;
 269             vvdwsum         += vvdw;
 270
 271             fscal            = felec+fvdw;
 272
 273             /* Calculate temporary vectorial force */
 274             tx               = fscal*dx00;
 275             ty               = fscal*dy00;
 276             tz               = fscal*dz00;
 277
 278             /* Update vectorial force */
 279             fix0            += tx;
 280             fiy0            += ty;
 281             fiz0            += tz;
 282             f[j_coord_offset+DIM*0+XX] -= tx;
 283             f[j_coord_offset+DIM*0+YY] -= ty;
 284             f[j_coord_offset+DIM*0+ZZ] -= tz;
 285
 286             }
 287
 288             /**************************
 289              * CALCULATE INTERACTIONS *
 290              **************************/
 291
 292             if (rsq10<rcutoff2)
 293             {
 294
 295             r10              = rsq10*rinv10;
 296
 297             qq10             = iq1*jq0;
 298
 299             /* EWALD ELECTROSTATICS */
 300
 301             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 302             ewrt             = r10*ewtabscale;
 303             ewitab           = ewrt;
 304             eweps            = ewrt-ewitab;
 305             ewitab           = 4*ewitab;
 306             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 307             velec            = qq10*(rinv10-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 308             felec            = qq10*rinv10*(rinvsq10-felec);
 309
 310             d                = r10-rswitch;
 311             d                = (d>0.0) ? d : 0.0;
 312             d2               = d*d;
 313             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 314
 315             dsw              = d2*(swF2+d*(swF3+d*swF4));
 316
 317             /* Evaluate switch function */
 318             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 319             felec            = felec*sw - rinv10*velec*dsw;
 320             velec           *= sw;
 321
 322             /* Update potential sums from outer loop */
 323             velecsum        += velec;
 324
 325             fscal            = felec;
 326
 327             /* Calculate temporary vectorial force */
 328             tx               = fscal*dx10;
 329             ty               = fscal*dy10;
 330             tz               = fscal*dz10;
 331
 332             /* Update vectorial force */
 333             fix1            += tx;
 334             fiy1            += ty;
 335             fiz1            += tz;
 336             f[j_coord_offset+DIM*0+XX] -= tx;
 337             f[j_coord_offset+DIM*0+YY] -= ty;
 338             f[j_coord_offset+DIM*0+ZZ] -= tz;
 339
 340             }
 341
 342             /**************************
 343              * CALCULATE INTERACTIONS *
 344              **************************/
 345
 346             if (rsq20<rcutoff2)
 347             {
 348
 349             r20              = rsq20*rinv20;
 350
 351             qq20             = iq2*jq0;
 352
 353             /* EWALD ELECTROSTATICS */
 354
 355             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 356             ewrt             = r20*ewtabscale;
 357             ewitab           = ewrt;
 358             eweps            = ewrt-ewitab;
 359             ewitab           = 4*ewitab;
 360             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 361             velec            = qq20*(rinv20-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 362             felec            = qq20*rinv20*(rinvsq20-felec);
 363
 364             d                = r20-rswitch;
 365             d                = (d>0.0) ? d : 0.0;
 366             d2               = d*d;
 367             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 368
 369             dsw              = d2*(swF2+d*(swF3+d*swF4));
 370
 371             /* Evaluate switch function */
 372             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 373             felec            = felec*sw - rinv20*velec*dsw;
 374             velec           *= sw;
 375
 376             /* Update potential sums from outer loop */
 377             velecsum        += velec;
 378
 379             fscal            = felec;
 380
 381             /* Calculate temporary vectorial force */
 382             tx               = fscal*dx20;
 383             ty               = fscal*dy20;
 384             tz               = fscal*dz20;
 385
 386             /* Update vectorial force */
 387             fix2            += tx;
 388             fiy2            += ty;
 389             fiz2            += tz;
 390             f[j_coord_offset+DIM*0+XX] -= tx;
 391             f[j_coord_offset+DIM*0+YY] -= ty;
 392             f[j_coord_offset+DIM*0+ZZ] -= tz;
 393
 394             }
 395
 396             /* Inner loop uses 219 flops */
 397         }
 398         /* End of innermost loop */
 399
 400         tx = ty = tz = 0;
 401         f[i_coord_offset+DIM*0+XX] += fix0;
 402         f[i_coord_offset+DIM*0+YY] += fiy0;
 403         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 404         tx                         += fix0;
 405         ty                         += fiy0;
 406         tz                         += fiz0;
 407         f[i_coord_offset+DIM*1+XX] += fix1;
 408         f[i_coord_offset+DIM*1+YY] += fiy1;
 409         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 410         tx                         += fix1;
 411         ty                         += fiy1;
 412         tz                         += fiz1;
 413         f[i_coord_offset+DIM*2+XX] += fix2;
 414         f[i_coord_offset+DIM*2+YY] += fiy2;
 415         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 416         tx                         += fix2;
 417         ty                         += fiy2;
 418         tz                         += fiz2;
 419         fshift[i_shift_offset+XX]  += tx;
 420         fshift[i_shift_offset+YY]  += ty;
 421         fshift[i_shift_offset+ZZ]  += tz;
 422
 423         ggid                        = gid[iidx];
 424         /* Update potential energies */
 425         kernel_data->energygrp_elec[ggid] += velecsum;
 426         kernel_data->energygrp_vdw[ggid] += vvdwsum;
 427
 428         /* Increment number of inner iterations */
 429         inneriter                  += j_index_end - j_index_start;
 430
 431         /* Outer loop uses 32 flops */
 432     }
 433
 434     /* Increment number of outer iterations */
 435     outeriter        += nri;
 436
 437     /* Update outer/inner flops */
 438
 439     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3_VF,outeriter*32 + inneriter*219);
 440 }
 441 /*
 442  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSw_VdwBhamSw_GeomW3P1_F_c
 443  * Electrostatics interaction: Ewald
 444  * VdW interaction:            Buckingham
 445  * Geometry:                   Water3-Particle
 446  * Calculate force/pot:        Force
 447  */
 448 void
 449 nb_kernel_ElecEwSw_VdwBhamSw_GeomW3P1_F_c
 450                     (t_nblist                    * gmx_restrict       nlist,
 451                      rvec                        * gmx_restrict          xx,
 452                      rvec                        * gmx_restrict          ff,
 453                      t_forcerec                  * gmx_restrict          fr,
 454                      t_mdatoms                   * gmx_restrict     mdatoms,
 455                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
 456                      t_nrnb                      * gmx_restrict        nrnb)
 457 {
 458     int              i_shift_offset,i_coord_offset,j_coord_offset;
 459     int              j_index_start,j_index_end;
 460     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
 461     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
 462     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 463     real             *shiftvec,*fshift,*x,*f;
 464     int              vdwioffset0;
 465     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 466     int              vdwioffset1;
 467     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
 468     int              vdwioffset2;
 469     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
 470     int              vdwjidx0;
 471     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 472     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
 473     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
 474     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
 475     real             velec,felec,velecsum,facel,crf,krf,krf2;
 476     real             *charge;
 477     int              nvdwtype;
 478     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
 479     int              *vdwtype;
 480     real             *vdwparam;
 481     int              ewitab;
 482     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
 483     real             *ewtab;
 484     real             rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
 485
 486     x                = xx[0];
 487     f                = ff[0];
 488
 489     nri              = nlist->nri;
 490     iinr             = nlist->iinr;
 491     jindex           = nlist->jindex;
 492     jjnr             = nlist->jjnr;
 493     shiftidx         = nlist->shift;
 494     gid              = nlist->gid;
 495     shiftvec         = fr->shift_vec[0];
 496     fshift           = fr->fshift[0];
 497     facel            = fr->epsfac;
 498     charge           = mdatoms->chargeA;
 499     nvdwtype         = fr->ntype;
 500     vdwparam         = fr->nbfp;
 501     vdwtype          = mdatoms->typeA;
 502
 503     sh_ewald         = fr->ic->sh_ewald;
 504     ewtab            = fr->ic->tabq_coul_FDV0;
 505     ewtabscale       = fr->ic->tabq_scale;
 506     ewtabhalfspace   = 0.5/ewtabscale;
 507
 508     /* Setup water-specific parameters */
 509     inr              = nlist->iinr[0];
 510     iq0              = facel*charge[inr+0];
 511     iq1              = facel*charge[inr+1];
 512     iq2              = facel*charge[inr+2];
 513     vdwioffset0      = 3*nvdwtype*vdwtype[inr+0];
 514
 515     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 516     rcutoff          = fr->rcoulomb;
 517     rcutoff2         = rcutoff*rcutoff;
 518
 519     rswitch          = fr->rcoulomb_switch;
 520     /* Setup switch parameters */
 521     d                = rcutoff-rswitch;
 522     swV3             = -10.0/(d*d*d);
 523     swV4             =  15.0/(d*d*d*d);
 524     swV5             =  -6.0/(d*d*d*d*d);
 525     swF2             = -30.0/(d*d*d);
 526     swF3             =  60.0/(d*d*d*d);
 527     swF4             = -30.0/(d*d*d*d*d);
 528
 529     outeriter        = 0;
 530     inneriter        = 0;
 531
 532     /* Start outer loop over neighborlists */
 533     for(iidx=0; iidx<nri; iidx++)
 534     {
 535         /* Load shift vector for this list */
 536         i_shift_offset   = DIM*shiftidx[iidx];
 537         shX              = shiftvec[i_shift_offset+XX];
 538         shY              = shiftvec[i_shift_offset+YY];
 539         shZ              = shiftvec[i_shift_offset+ZZ];
 540
 541         /* Load limits for loop over neighbors */
 542         j_index_start    = jindex[iidx];
 543         j_index_end      = jindex[iidx+1];
 544
 545         /* Get outer coordinate index */
 546         inr              = iinr[iidx];
 547         i_coord_offset   = DIM*inr;
 548
 549         /* Load i particle coords and add shift vector */
 550         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 551         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 552         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 553         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 554         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 555         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 556         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 557         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 558         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 559
 560         fix0             = 0.0;
 561         fiy0             = 0.0;
 562         fiz0             = 0.0;
 563         fix1             = 0.0;
 564         fiy1             = 0.0;
 565         fiz1             = 0.0;
 566         fix2             = 0.0;
 567         fiy2             = 0.0;
 568         fiz2             = 0.0;
 569
 570         /* Start inner kernel loop */
 571         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 572         {
 573             /* Get j neighbor index, and coordinate index */
 574             jnr              = jjnr[jidx];
 575             j_coord_offset   = DIM*jnr;
 576
 577             /* load j atom coordinates */
 578             jx0              = x[j_coord_offset+DIM*0+XX];
 579             jy0              = x[j_coord_offset+DIM*0+YY];
 580             jz0              = x[j_coord_offset+DIM*0+ZZ];
 581
 582             /* Calculate displacement vector */
 583             dx00             = ix0 - jx0;
 584             dy00             = iy0 - jy0;
 585             dz00             = iz0 - jz0;
 586             dx10             = ix1 - jx0;
 587             dy10             = iy1 - jy0;
 588             dz10             = iz1 - jz0;
 589             dx20             = ix2 - jx0;
 590             dy20             = iy2 - jy0;
 591             dz20             = iz2 - jz0;
 592
 593             /* Calculate squared distance and things based on it */
 594             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 595             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 596             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 597
 598             rinv00           = gmx_invsqrt(rsq00);
 599             rinv10           = gmx_invsqrt(rsq10);
 600             rinv20           = gmx_invsqrt(rsq20);
 601
 602             rinvsq00         = rinv00*rinv00;
 603             rinvsq10         = rinv10*rinv10;
 604             rinvsq20         = rinv20*rinv20;
 605
 606             /* Load parameters for j particles */
 607             jq0              = charge[jnr+0];
 608             vdwjidx0         = 3*vdwtype[jnr+0];
 609
 610             /**************************
 611              * CALCULATE INTERACTIONS *
 612              **************************/
 613
 614             if (rsq00<rcutoff2)
 615             {
 616
 617             r00              = rsq00*rinv00;
 618
 619             qq00             = iq0*jq0;
 620             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 621             cexp1_00         = vdwparam[vdwioffset0+vdwjidx0+1];
 622             cexp2_00         = vdwparam[vdwioffset0+vdwjidx0+2];
 623
 624             /* EWALD ELECTROSTATICS */
 625
 626             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 627             ewrt             = r00*ewtabscale;
 628             ewitab           = ewrt;
 629             eweps            = ewrt-ewitab;
 630             ewitab           = 4*ewitab;
 631             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 632             velec            = qq00*(rinv00-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 633             felec            = qq00*rinv00*(rinvsq00-felec);
 634
 635             /* BUCKINGHAM DISPERSION/REPULSION */
 636             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 637             vvdw6            = c6_00*rinvsix;
 638             br               = cexp2_00*r00;
 639             vvdwexp          = cexp1_00*exp(-br);
 640             vvdw             = vvdwexp - vvdw6*(1.0/6.0);
 641             fvdw             = (br*vvdwexp-vvdw6)*rinvsq00;
 642
 643             d                = r00-rswitch;
 644             d                = (d>0.0) ? d : 0.0;
 645             d2               = d*d;
 646             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 647
 648             dsw              = d2*(swF2+d*(swF3+d*swF4));
 649
 650             /* Evaluate switch function */
 651             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 652             felec            = felec*sw - rinv00*velec*dsw;
 653             fvdw             = fvdw*sw - rinv00*vvdw*dsw;
 654
 655             fscal            = felec+fvdw;
 656
 657             /* Calculate temporary vectorial force */
 658             tx               = fscal*dx00;
 659             ty               = fscal*dy00;
 660             tz               = fscal*dz00;
 661
 662             /* Update vectorial force */
 663             fix0            += tx;
 664             fiy0            += ty;
 665             fiz0            += tz;
 666             f[j_coord_offset+DIM*0+XX] -= tx;
 667             f[j_coord_offset+DIM*0+YY] -= ty;
 668             f[j_coord_offset+DIM*0+ZZ] -= tz;
 669
 670             }
 671
 672             /**************************
 673              * CALCULATE INTERACTIONS *
 674              **************************/
 675
 676             if (rsq10<rcutoff2)
 677             {
 678
 679             r10              = rsq10*rinv10;
 680
 681             qq10             = iq1*jq0;
 682
 683             /* EWALD ELECTROSTATICS */
 684
 685             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 686             ewrt             = r10*ewtabscale;
 687             ewitab           = ewrt;
 688             eweps            = ewrt-ewitab;
 689             ewitab           = 4*ewitab;
 690             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 691             velec            = qq10*(rinv10-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 692             felec            = qq10*rinv10*(rinvsq10-felec);
 693
 694             d                = r10-rswitch;
 695             d                = (d>0.0) ? d : 0.0;
 696             d2               = d*d;
 697             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 698
 699             dsw              = d2*(swF2+d*(swF3+d*swF4));
 700
 701             /* Evaluate switch function */
 702             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 703             felec            = felec*sw - rinv10*velec*dsw;
 704
 705             fscal            = felec;
 706
 707             /* Calculate temporary vectorial force */
 708             tx               = fscal*dx10;
 709             ty               = fscal*dy10;
 710             tz               = fscal*dz10;
 711
 712             /* Update vectorial force */
 713             fix1            += tx;
 714             fiy1            += ty;
 715             fiz1            += tz;
 716             f[j_coord_offset+DIM*0+XX] -= tx;
 717             f[j_coord_offset+DIM*0+YY] -= ty;
 718             f[j_coord_offset+DIM*0+ZZ] -= tz;
 719
 720             }
 721
 722             /**************************
 723              * CALCULATE INTERACTIONS *
 724              **************************/
 725
 726             if (rsq20<rcutoff2)
 727             {
 728
 729             r20              = rsq20*rinv20;
 730
 731             qq20             = iq2*jq0;
 732
 733             /* EWALD ELECTROSTATICS */
 734
 735             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 736             ewrt             = r20*ewtabscale;
 737             ewitab           = ewrt;
 738             eweps            = ewrt-ewitab;
 739             ewitab           = 4*ewitab;
 740             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 741             velec            = qq20*(rinv20-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 742             felec            = qq20*rinv20*(rinvsq20-felec);
 743
 744             d                = r20-rswitch;
 745             d                = (d>0.0) ? d : 0.0;
 746             d2               = d*d;
 747             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 748
 749             dsw              = d2*(swF2+d*(swF3+d*swF4));
 750
 751             /* Evaluate switch function */
 752             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 753             felec            = felec*sw - rinv20*velec*dsw;
 754
 755             fscal            = felec;
 756
 757             /* Calculate temporary vectorial force */
 758             tx               = fscal*dx20;
 759             ty               = fscal*dy20;
 760             tz               = fscal*dz20;
 761
 762             /* Update vectorial force */
 763             fix2            += tx;
 764             fiy2            += ty;
 765             fiz2            += tz;
 766             f[j_coord_offset+DIM*0+XX] -= tx;
 767             f[j_coord_offset+DIM*0+YY] -= ty;
 768             f[j_coord_offset+DIM*0+ZZ] -= tz;
 769
 770             }
 771
 772             /* Inner loop uses 211 flops */
 773         }
 774         /* End of innermost loop */
 775
 776         tx = ty = tz = 0;
 777         f[i_coord_offset+DIM*0+XX] += fix0;
 778         f[i_coord_offset+DIM*0+YY] += fiy0;
 779         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 780         tx                         += fix0;
 781         ty                         += fiy0;
 782         tz                         += fiz0;
 783         f[i_coord_offset+DIM*1+XX] += fix1;
 784         f[i_coord_offset+DIM*1+YY] += fiy1;
 785         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 786         tx                         += fix1;
 787         ty                         += fiy1;
 788         tz                         += fiz1;
 789         f[i_coord_offset+DIM*2+XX] += fix2;
 790         f[i_coord_offset+DIM*2+YY] += fiy2;
 791         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 792         tx                         += fix2;
 793         ty                         += fiy2;
 794         tz                         += fiz2;
 795         fshift[i_shift_offset+XX]  += tx;
 796         fshift[i_shift_offset+YY]  += ty;
 797         fshift[i_shift_offset+ZZ]  += tz;
 798
 799         /* Increment number of inner iterations */
 800         inneriter                  += j_index_end - j_index_start;
 801
 802         /* Outer loop uses 30 flops */
 803     }
 804
 805     /* Increment number of outer iterations */
 806     outeriter        += nri;
 807
 808     /* Update outer/inner flops */
 809
 810     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3_F,outeriter*30 + inneriter*211);
 811 }