src/gromacs/gmxlib/nonbonded/nb_kernel_c/nb_kernel_ElecEwSw_VdwLJSw_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 "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_ElecEwSw_VdwLJSw_GeomW3P1_VF_c
  51  * Electrostatics interaction: Ewald
  52  * VdW interaction:            LennardJones
  53  * Geometry:                   Water3-Particle
  54  * Calculate force/pot:        PotentialAndForce
  55  */
  56 void
  57 nb_kernel_ElecEwSw_VdwLJSw_GeomW3P1_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              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             velec,felec,velecsum,facel,crf,krf,krf2;
  84     real             *charge;
  85     int              nvdwtype;
  86     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
  87     int              *vdwtype;
  88     real             *vdwparam;
  89     int              ewitab;
  90     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
  91     real             *ewtab;
  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     nvdwtype         = fr->ntype;
 108     vdwparam         = fr->nbfp;
 109     vdwtype          = mdatoms->typeA;
 110
 111     sh_ewald         = fr->ic->sh_ewald;
 112     ewtab            = fr->ic->tabq_coul_FDV0;
 113     ewtabscale       = fr->ic->tabq_scale;
 114     ewtabhalfspace   = 0.5/ewtabscale;
 115
 116     /* Setup water-specific parameters */
 117     inr              = nlist->iinr[0];
 118     iq0              = facel*charge[inr+0];
 119     iq1              = facel*charge[inr+1];
 120     iq2              = facel*charge[inr+2];
 121     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 122
 123     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 124     rcutoff          = fr->rcoulomb;
 125     rcutoff2         = rcutoff*rcutoff;
 126
 127     rswitch          = fr->rcoulomb_switch;
 128     /* Setup switch parameters */
 129     d                = rcutoff-rswitch;
 130     swV3             = -10.0/(d*d*d);
 131     swV4             =  15.0/(d*d*d*d);
 132     swV5             =  -6.0/(d*d*d*d*d);
 133     swF2             = -30.0/(d*d*d);
 134     swF3             =  60.0/(d*d*d*d);
 135     swF4             = -30.0/(d*d*d*d*d);
 136
 137     outeriter        = 0;
 138     inneriter        = 0;
 139
 140     /* Start outer loop over neighborlists */
 141     for(iidx=0; iidx<nri; iidx++)
 142     {
 143         /* Load shift vector for this list */
 144         i_shift_offset   = DIM*shiftidx[iidx];
 145         shX              = shiftvec[i_shift_offset+XX];
 146         shY              = shiftvec[i_shift_offset+YY];
 147         shZ              = shiftvec[i_shift_offset+ZZ];
 148
 149         /* Load limits for loop over neighbors */
 150         j_index_start    = jindex[iidx];
 151         j_index_end      = jindex[iidx+1];
 152
 153         /* Get outer coordinate index */
 154         inr              = iinr[iidx];
 155         i_coord_offset   = DIM*inr;
 156
 157         /* Load i particle coords and add shift vector */
 158         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 159         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 160         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 161         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 162         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 163         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 164         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 165         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 166         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 167
 168         fix0             = 0.0;
 169         fiy0             = 0.0;
 170         fiz0             = 0.0;
 171         fix1             = 0.0;
 172         fiy1             = 0.0;
 173         fiz1             = 0.0;
 174         fix2             = 0.0;
 175         fiy2             = 0.0;
 176         fiz2             = 0.0;
 177
 178         /* Reset potential sums */
 179         velecsum         = 0.0;
 180         vvdwsum          = 0.0;
 181
 182         /* Start inner kernel loop */
 183         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 184         {
 185             /* Get j neighbor index, and coordinate index */
 186             jnr              = jjnr[jidx];
 187             j_coord_offset   = DIM*jnr;
 188
 189             /* load j atom coordinates */
 190             jx0              = x[j_coord_offset+DIM*0+XX];
 191             jy0              = x[j_coord_offset+DIM*0+YY];
 192             jz0              = x[j_coord_offset+DIM*0+ZZ];
 193
 194             /* Calculate displacement vector */
 195             dx00             = ix0 - jx0;
 196             dy00             = iy0 - jy0;
 197             dz00             = iz0 - jz0;
 198             dx10             = ix1 - jx0;
 199             dy10             = iy1 - jy0;
 200             dz10             = iz1 - jz0;
 201             dx20             = ix2 - jx0;
 202             dy20             = iy2 - jy0;
 203             dz20             = iz2 - jz0;
 204
 205             /* Calculate squared distance and things based on it */
 206             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 207             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 208             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 209
 210             rinv00           = gmx_invsqrt(rsq00);
 211             rinv10           = gmx_invsqrt(rsq10);
 212             rinv20           = gmx_invsqrt(rsq20);
 213
 214             rinvsq00         = rinv00*rinv00;
 215             rinvsq10         = rinv10*rinv10;
 216             rinvsq20         = rinv20*rinv20;
 217
 218             /* Load parameters for j particles */
 219             jq0              = charge[jnr+0];
 220             vdwjidx0         = 2*vdwtype[jnr+0];
 221
 222             /**************************
 223              * CALCULATE INTERACTIONS *
 224              **************************/
 225
 226             if (rsq00<rcutoff2)
 227             {
 228
 229             r00              = rsq00*rinv00;
 230
 231             qq00             = iq0*jq0;
 232             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 233             c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 234
 235             /* EWALD ELECTROSTATICS */
 236
 237             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 238             ewrt             = r00*ewtabscale;
 239             ewitab           = ewrt;
 240             eweps            = ewrt-ewitab;
 241             ewitab           = 4*ewitab;
 242             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 243             velec            = qq00*(rinv00-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 244             felec            = qq00*rinv00*(rinvsq00-felec);
 245
 246             /* LENNARD-JONES DISPERSION/REPULSION */
 247
 248             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 249             vvdw6            = c6_00*rinvsix;
 250             vvdw12           = c12_00*rinvsix*rinvsix;
 251             vvdw             = vvdw12*(1.0/12.0) - vvdw6*(1.0/6.0);
 252             fvdw             = (vvdw12-vvdw6)*rinvsq00;
 253
 254             d                = r00-rswitch;
 255             d                = (d>0.0) ? d : 0.0;
 256             d2               = d*d;
 257             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 258
 259             dsw              = d2*(swF2+d*(swF3+d*swF4));
 260
 261             /* Evaluate switch function */
 262             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 263             felec            = felec*sw - rinv00*velec*dsw;
 264             fvdw             = fvdw*sw - rinv00*vvdw*dsw;
 265             velec           *= sw;
 266             vvdw            *= sw;
 267
 268             /* Update potential sums from outer loop */
 269             velecsum        += velec;
 270             vvdwsum         += vvdw;
 271
 272             fscal            = felec+fvdw;
 273
 274             /* Calculate temporary vectorial force */
 275             tx               = fscal*dx00;
 276             ty               = fscal*dy00;
 277             tz               = fscal*dz00;
 278
 279             /* Update vectorial force */
 280             fix0            += tx;
 281             fiy0            += ty;
 282             fiz0            += tz;
 283             f[j_coord_offset+DIM*0+XX] -= tx;
 284             f[j_coord_offset+DIM*0+YY] -= ty;
 285             f[j_coord_offset+DIM*0+ZZ] -= tz;
 286
 287             }
 288
 289             /**************************
 290              * CALCULATE INTERACTIONS *
 291              **************************/
 292
 293             if (rsq10<rcutoff2)
 294             {
 295
 296             r10              = rsq10*rinv10;
 297
 298             qq10             = iq1*jq0;
 299
 300             /* EWALD ELECTROSTATICS */
 301
 302             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 303             ewrt             = r10*ewtabscale;
 304             ewitab           = ewrt;
 305             eweps            = ewrt-ewitab;
 306             ewitab           = 4*ewitab;
 307             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 308             velec            = qq10*(rinv10-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 309             felec            = qq10*rinv10*(rinvsq10-felec);
 310
 311             d                = r10-rswitch;
 312             d                = (d>0.0) ? d : 0.0;
 313             d2               = d*d;
 314             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 315
 316             dsw              = d2*(swF2+d*(swF3+d*swF4));
 317
 318             /* Evaluate switch function */
 319             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 320             felec            = felec*sw - rinv10*velec*dsw;
 321             velec           *= sw;
 322
 323             /* Update potential sums from outer loop */
 324             velecsum        += velec;
 325
 326             fscal            = felec;
 327
 328             /* Calculate temporary vectorial force */
 329             tx               = fscal*dx10;
 330             ty               = fscal*dy10;
 331             tz               = fscal*dz10;
 332
 333             /* Update vectorial force */
 334             fix1            += tx;
 335             fiy1            += ty;
 336             fiz1            += tz;
 337             f[j_coord_offset+DIM*0+XX] -= tx;
 338             f[j_coord_offset+DIM*0+YY] -= ty;
 339             f[j_coord_offset+DIM*0+ZZ] -= tz;
 340
 341             }
 342
 343             /**************************
 344              * CALCULATE INTERACTIONS *
 345              **************************/
 346
 347             if (rsq20<rcutoff2)
 348             {
 349
 350             r20              = rsq20*rinv20;
 351
 352             qq20             = iq2*jq0;
 353
 354             /* EWALD ELECTROSTATICS */
 355
 356             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 357             ewrt             = r20*ewtabscale;
 358             ewitab           = ewrt;
 359             eweps            = ewrt-ewitab;
 360             ewitab           = 4*ewitab;
 361             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 362             velec            = qq20*(rinv20-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 363             felec            = qq20*rinv20*(rinvsq20-felec);
 364
 365             d                = r20-rswitch;
 366             d                = (d>0.0) ? d : 0.0;
 367             d2               = d*d;
 368             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 369
 370             dsw              = d2*(swF2+d*(swF3+d*swF4));
 371
 372             /* Evaluate switch function */
 373             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 374             felec            = felec*sw - rinv20*velec*dsw;
 375             velec           *= sw;
 376
 377             /* Update potential sums from outer loop */
 378             velecsum        += velec;
 379
 380             fscal            = felec;
 381
 382             /* Calculate temporary vectorial force */
 383             tx               = fscal*dx20;
 384             ty               = fscal*dy20;
 385             tz               = fscal*dz20;
 386
 387             /* Update vectorial force */
 388             fix2            += tx;
 389             fiy2            += ty;
 390             fiz2            += tz;
 391             f[j_coord_offset+DIM*0+XX] -= tx;
 392             f[j_coord_offset+DIM*0+YY] -= ty;
 393             f[j_coord_offset+DIM*0+ZZ] -= tz;
 394
 395             }
 396
 397             /* Inner loop uses 193 flops */
 398         }
 399         /* End of innermost loop */
 400
 401         tx = ty = tz = 0;
 402         f[i_coord_offset+DIM*0+XX] += fix0;
 403         f[i_coord_offset+DIM*0+YY] += fiy0;
 404         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 405         tx                         += fix0;
 406         ty                         += fiy0;
 407         tz                         += fiz0;
 408         f[i_coord_offset+DIM*1+XX] += fix1;
 409         f[i_coord_offset+DIM*1+YY] += fiy1;
 410         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 411         tx                         += fix1;
 412         ty                         += fiy1;
 413         tz                         += fiz1;
 414         f[i_coord_offset+DIM*2+XX] += fix2;
 415         f[i_coord_offset+DIM*2+YY] += fiy2;
 416         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 417         tx                         += fix2;
 418         ty                         += fiy2;
 419         tz                         += fiz2;
 420         fshift[i_shift_offset+XX]  += tx;
 421         fshift[i_shift_offset+YY]  += ty;
 422         fshift[i_shift_offset+ZZ]  += tz;
 423
 424         ggid                        = gid[iidx];
 425         /* Update potential energies */
 426         kernel_data->energygrp_elec[ggid] += velecsum;
 427         kernel_data->energygrp_vdw[ggid] += vvdwsum;
 428
 429         /* Increment number of inner iterations */
 430         inneriter                  += j_index_end - j_index_start;
 431
 432         /* Outer loop uses 32 flops */
 433     }
 434
 435     /* Increment number of outer iterations */
 436     outeriter        += nri;
 437
 438     /* Update outer/inner flops */
 439
 440     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3_VF,outeriter*32 + inneriter*193);
 441 }
 442 /*
 443  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSw_VdwLJSw_GeomW3P1_F_c
 444  * Electrostatics interaction: Ewald
 445  * VdW interaction:            LennardJones
 446  * Geometry:                   Water3-Particle
 447  * Calculate force/pot:        Force
 448  */
 449 void
 450 nb_kernel_ElecEwSw_VdwLJSw_GeomW3P1_F_c
 451                     (t_nblist                    * gmx_restrict       nlist,
 452                      rvec                        * gmx_restrict          xx,
 453                      rvec                        * gmx_restrict          ff,
 454                      t_forcerec                  * gmx_restrict          fr,
 455                      t_mdatoms                   * gmx_restrict     mdatoms,
 456                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
 457                      t_nrnb                      * gmx_restrict        nrnb)
 458 {
 459     int              i_shift_offset,i_coord_offset,j_coord_offset;
 460     int              j_index_start,j_index_end;
 461     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
 462     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
 463     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 464     real             *shiftvec,*fshift,*x,*f;
 465     int              vdwioffset0;
 466     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 467     int              vdwioffset1;
 468     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
 469     int              vdwioffset2;
 470     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
 471     int              vdwjidx0;
 472     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 473     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
 474     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
 475     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
 476     real             velec,felec,velecsum,facel,crf,krf,krf2;
 477     real             *charge;
 478     int              nvdwtype;
 479     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
 480     int              *vdwtype;
 481     real             *vdwparam;
 482     int              ewitab;
 483     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
 484     real             *ewtab;
 485     real             rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
 486
 487     x                = xx[0];
 488     f                = ff[0];
 489
 490     nri              = nlist->nri;
 491     iinr             = nlist->iinr;
 492     jindex           = nlist->jindex;
 493     jjnr             = nlist->jjnr;
 494     shiftidx         = nlist->shift;
 495     gid              = nlist->gid;
 496     shiftvec         = fr->shift_vec[0];
 497     fshift           = fr->fshift[0];
 498     facel            = fr->epsfac;
 499     charge           = mdatoms->chargeA;
 500     nvdwtype         = fr->ntype;
 501     vdwparam         = fr->nbfp;
 502     vdwtype          = mdatoms->typeA;
 503
 504     sh_ewald         = fr->ic->sh_ewald;
 505     ewtab            = fr->ic->tabq_coul_FDV0;
 506     ewtabscale       = fr->ic->tabq_scale;
 507     ewtabhalfspace   = 0.5/ewtabscale;
 508
 509     /* Setup water-specific parameters */
 510     inr              = nlist->iinr[0];
 511     iq0              = facel*charge[inr+0];
 512     iq1              = facel*charge[inr+1];
 513     iq2              = facel*charge[inr+2];
 514     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 515
 516     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 517     rcutoff          = fr->rcoulomb;
 518     rcutoff2         = rcutoff*rcutoff;
 519
 520     rswitch          = fr->rcoulomb_switch;
 521     /* Setup switch parameters */
 522     d                = rcutoff-rswitch;
 523     swV3             = -10.0/(d*d*d);
 524     swV4             =  15.0/(d*d*d*d);
 525     swV5             =  -6.0/(d*d*d*d*d);
 526     swF2             = -30.0/(d*d*d);
 527     swF3             =  60.0/(d*d*d*d);
 528     swF4             = -30.0/(d*d*d*d*d);
 529
 530     outeriter        = 0;
 531     inneriter        = 0;
 532
 533     /* Start outer loop over neighborlists */
 534     for(iidx=0; iidx<nri; iidx++)
 535     {
 536         /* Load shift vector for this list */
 537         i_shift_offset   = DIM*shiftidx[iidx];
 538         shX              = shiftvec[i_shift_offset+XX];
 539         shY              = shiftvec[i_shift_offset+YY];
 540         shZ              = shiftvec[i_shift_offset+ZZ];
 541
 542         /* Load limits for loop over neighbors */
 543         j_index_start    = jindex[iidx];
 544         j_index_end      = jindex[iidx+1];
 545
 546         /* Get outer coordinate index */
 547         inr              = iinr[iidx];
 548         i_coord_offset   = DIM*inr;
 549
 550         /* Load i particle coords and add shift vector */
 551         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 552         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 553         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 554         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 555         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 556         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 557         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 558         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 559         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 560
 561         fix0             = 0.0;
 562         fiy0             = 0.0;
 563         fiz0             = 0.0;
 564         fix1             = 0.0;
 565         fiy1             = 0.0;
 566         fiz1             = 0.0;
 567         fix2             = 0.0;
 568         fiy2             = 0.0;
 569         fiz2             = 0.0;
 570
 571         /* Start inner kernel loop */
 572         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 573         {
 574             /* Get j neighbor index, and coordinate index */
 575             jnr              = jjnr[jidx];
 576             j_coord_offset   = DIM*jnr;
 577
 578             /* load j atom coordinates */
 579             jx0              = x[j_coord_offset+DIM*0+XX];
 580             jy0              = x[j_coord_offset+DIM*0+YY];
 581             jz0              = x[j_coord_offset+DIM*0+ZZ];
 582
 583             /* Calculate displacement vector */
 584             dx00             = ix0 - jx0;
 585             dy00             = iy0 - jy0;
 586             dz00             = iz0 - jz0;
 587             dx10             = ix1 - jx0;
 588             dy10             = iy1 - jy0;
 589             dz10             = iz1 - jz0;
 590             dx20             = ix2 - jx0;
 591             dy20             = iy2 - jy0;
 592             dz20             = iz2 - 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
 599             rinv00           = gmx_invsqrt(rsq00);
 600             rinv10           = gmx_invsqrt(rsq10);
 601             rinv20           = gmx_invsqrt(rsq20);
 602
 603             rinvsq00         = rinv00*rinv00;
 604             rinvsq10         = rinv10*rinv10;
 605             rinvsq20         = rinv20*rinv20;
 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             qq00             = iq0*jq0;
 621             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 622             c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 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             /* LENNARD-JONES DISPERSION/REPULSION */
 636
 637             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 638             vvdw6            = c6_00*rinvsix;
 639             vvdw12           = c12_00*rinvsix*rinvsix;
 640             vvdw             = vvdw12*(1.0/12.0) - vvdw6*(1.0/6.0);
 641             fvdw             = (vvdw12-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 185 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*185);
 811 }