src/gromacs/gmxlib/nonbonded/nb_kernel_c/nb_kernel_ElecEwSw_VdwBhamSw_GeomW4P1_c.c

   1 /*
   2  * This file is part of the GROMACS molecular simulation package.
   3  *
   4  * Copyright (c) 2012,2013,2014, by the GROMACS development team, led by
   5  * Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl,
   6  * and including many others, as listed in the AUTHORS file in the
   7  * top-level source directory and at http://www.gromacs.org.
   8  *
   9  * GROMACS is free software; you can redistribute it and/or
  10  * modify it under the terms of the GNU Lesser General Public License
  11  * as published by the Free Software Foundation; either version 2.1
  12  * of the License, or (at your option) any later version.
  13  *
  14  * GROMACS is distributed in the hope that it will be useful,
  15  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  16  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  17  * Lesser General Public License for more details.
  18  *
  19  * You should have received a copy of the GNU Lesser General Public
  20  * License along with GROMACS; if not, see
  21  * http://www.gnu.org/licenses, or write to the Free Software Foundation,
  22  * Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA.
  23  *
  24  * If you want to redistribute modifications to GROMACS, please
  25  * consider that scientific software is very special. Version
  26  * control is crucial - bugs must be traceable. We will be happy to
  27  * consider code for inclusion in the official distribution, but
  28  * derived work must not be called official GROMACS. Details are found
  29  * in the README & COPYING files - if they are missing, get the
  30  * official version at http://www.gromacs.org.
  31  *
  32  * To help us fund GROMACS development, we humbly ask that you cite
  33  * the research papers on the package. Check out http://www.gromacs.org.
  34  */
  35 /*
  36  * Note: this file was generated by the GROMACS c kernel generator.
  37  */
  38 #include "gmxpre.h"
  39
  40 #include "config.h"
  41
  42 #include <math.h>
  43
  44 #include "../nb_kernel.h"
  45 #include "gromacs/legacyheaders/types/simple.h"
  46 #include "gromacs/math/vec.h"
  47 #include "gromacs/legacyheaders/nrnb.h"
  48
  49 /*
  50  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSw_VdwBhamSw_GeomW4P1_VF_c
  51  * Electrostatics interaction: Ewald
  52  * VdW interaction:            Buckingham
  53  * Geometry:                   Water4-Particle
  54  * Calculate force/pot:        PotentialAndForce
  55  */
  56 void
  57 nb_kernel_ElecEwSw_VdwBhamSw_GeomW4P1_VF_c
  58                     (t_nblist                    * gmx_restrict       nlist,
  59                      rvec                        * gmx_restrict          xx,
  60                      rvec                        * gmx_restrict          ff,
  61                      t_forcerec                  * gmx_restrict          fr,
  62                      t_mdatoms                   * gmx_restrict     mdatoms,
  63                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
  64                      t_nrnb                      * gmx_restrict        nrnb)
  65 {
  66     int              i_shift_offset,i_coord_offset,j_coord_offset;
  67     int              j_index_start,j_index_end;
  68     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
  69     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
  70     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
  71     real             *shiftvec,*fshift,*x,*f;
  72     int              vdwioffset0;
  73     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
  74     int              vdwioffset1;
  75     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
  76     int              vdwioffset2;
  77     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
  78     int              vdwioffset3;
  79     real             ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
  80     int              vdwjidx0;
  81     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
  82     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
  83     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
  84     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
  85     real             dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30,cexp1_30,cexp2_30;
  86     real             velec,felec,velecsum,facel,crf,krf,krf2;
  87     real             *charge;
  88     int              nvdwtype;
  89     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
  90     int              *vdwtype;
  91     real             *vdwparam;
  92     int              ewitab;
  93     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
  94     real             *ewtab;
  95     real             rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
  96
  97     x                = xx[0];
  98     f                = ff[0];
  99
 100     nri              = nlist->nri;
 101     iinr             = nlist->iinr;
 102     jindex           = nlist->jindex;
 103     jjnr             = nlist->jjnr;
 104     shiftidx         = nlist->shift;
 105     gid              = nlist->gid;
 106     shiftvec         = fr->shift_vec[0];
 107     fshift           = fr->fshift[0];
 108     facel            = fr->epsfac;
 109     charge           = mdatoms->chargeA;
 110     nvdwtype         = fr->ntype;
 111     vdwparam         = fr->nbfp;
 112     vdwtype          = mdatoms->typeA;
 113
 114     sh_ewald         = fr->ic->sh_ewald;
 115     ewtab            = fr->ic->tabq_coul_FDV0;
 116     ewtabscale       = fr->ic->tabq_scale;
 117     ewtabhalfspace   = 0.5/ewtabscale;
 118
 119     /* Setup water-specific parameters */
 120     inr              = nlist->iinr[0];
 121     iq1              = facel*charge[inr+1];
 122     iq2              = facel*charge[inr+2];
 123     iq3              = facel*charge[inr+3];
 124     vdwioffset0      = 3*nvdwtype*vdwtype[inr+0];
 125
 126     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 127     rcutoff          = fr->rcoulomb;
 128     rcutoff2         = rcutoff*rcutoff;
 129
 130     rswitch          = fr->rcoulomb_switch;
 131     /* Setup switch parameters */
 132     d                = rcutoff-rswitch;
 133     swV3             = -10.0/(d*d*d);
 134     swV4             =  15.0/(d*d*d*d);
 135     swV5             =  -6.0/(d*d*d*d*d);
 136     swF2             = -30.0/(d*d*d);
 137     swF3             =  60.0/(d*d*d*d);
 138     swF4             = -30.0/(d*d*d*d*d);
 139
 140     outeriter        = 0;
 141     inneriter        = 0;
 142
 143     /* Start outer loop over neighborlists */
 144     for(iidx=0; iidx<nri; iidx++)
 145     {
 146         /* Load shift vector for this list */
 147         i_shift_offset   = DIM*shiftidx[iidx];
 148         shX              = shiftvec[i_shift_offset+XX];
 149         shY              = shiftvec[i_shift_offset+YY];
 150         shZ              = shiftvec[i_shift_offset+ZZ];
 151
 152         /* Load limits for loop over neighbors */
 153         j_index_start    = jindex[iidx];
 154         j_index_end      = jindex[iidx+1];
 155
 156         /* Get outer coordinate index */
 157         inr              = iinr[iidx];
 158         i_coord_offset   = DIM*inr;
 159
 160         /* Load i particle coords and add shift vector */
 161         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 162         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 163         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 164         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 165         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 166         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 167         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 168         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 169         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 170         ix3              = shX + x[i_coord_offset+DIM*3+XX];
 171         iy3              = shY + x[i_coord_offset+DIM*3+YY];
 172         iz3              = shZ + x[i_coord_offset+DIM*3+ZZ];
 173
 174         fix0             = 0.0;
 175         fiy0             = 0.0;
 176         fiz0             = 0.0;
 177         fix1             = 0.0;
 178         fiy1             = 0.0;
 179         fiz1             = 0.0;
 180         fix2             = 0.0;
 181         fiy2             = 0.0;
 182         fiz2             = 0.0;
 183         fix3             = 0.0;
 184         fiy3             = 0.0;
 185         fiz3             = 0.0;
 186
 187         /* Reset potential sums */
 188         velecsum         = 0.0;
 189         vvdwsum          = 0.0;
 190
 191         /* Start inner kernel loop */
 192         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 193         {
 194             /* Get j neighbor index, and coordinate index */
 195             jnr              = jjnr[jidx];
 196             j_coord_offset   = DIM*jnr;
 197
 198             /* load j atom coordinates */
 199             jx0              = x[j_coord_offset+DIM*0+XX];
 200             jy0              = x[j_coord_offset+DIM*0+YY];
 201             jz0              = x[j_coord_offset+DIM*0+ZZ];
 202
 203             /* Calculate displacement vector */
 204             dx00             = ix0 - jx0;
 205             dy00             = iy0 - jy0;
 206             dz00             = iz0 - jz0;
 207             dx10             = ix1 - jx0;
 208             dy10             = iy1 - jy0;
 209             dz10             = iz1 - jz0;
 210             dx20             = ix2 - jx0;
 211             dy20             = iy2 - jy0;
 212             dz20             = iz2 - jz0;
 213             dx30             = ix3 - jx0;
 214             dy30             = iy3 - jy0;
 215             dz30             = iz3 - jz0;
 216
 217             /* Calculate squared distance and things based on it */
 218             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 219             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 220             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 221             rsq30            = dx30*dx30+dy30*dy30+dz30*dz30;
 222
 223             rinv00           = gmx_invsqrt(rsq00);
 224             rinv10           = gmx_invsqrt(rsq10);
 225             rinv20           = gmx_invsqrt(rsq20);
 226             rinv30           = gmx_invsqrt(rsq30);
 227
 228             rinvsq00         = rinv00*rinv00;
 229             rinvsq10         = rinv10*rinv10;
 230             rinvsq20         = rinv20*rinv20;
 231             rinvsq30         = rinv30*rinv30;
 232
 233             /* Load parameters for j particles */
 234             jq0              = charge[jnr+0];
 235             vdwjidx0         = 3*vdwtype[jnr+0];
 236
 237             /**************************
 238              * CALCULATE INTERACTIONS *
 239              **************************/
 240
 241             if (rsq00<rcutoff2)
 242             {
 243
 244             r00              = rsq00*rinv00;
 245
 246             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 247             cexp1_00         = vdwparam[vdwioffset0+vdwjidx0+1];
 248             cexp2_00         = vdwparam[vdwioffset0+vdwjidx0+2];
 249
 250             /* BUCKINGHAM DISPERSION/REPULSION */
 251             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 252             vvdw6            = c6_00*rinvsix;
 253             br               = cexp2_00*r00;
 254             vvdwexp          = cexp1_00*exp(-br);
 255             vvdw             = vvdwexp - vvdw6*(1.0/6.0);
 256             fvdw             = (br*vvdwexp-vvdw6)*rinvsq00;
 257
 258             d                = r00-rswitch;
 259             d                = (d>0.0) ? d : 0.0;
 260             d2               = d*d;
 261             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 262
 263             dsw              = d2*(swF2+d*(swF3+d*swF4));
 264
 265             /* Evaluate switch function */
 266             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 267             fvdw             = fvdw*sw - rinv00*vvdw*dsw;
 268             vvdw            *= sw;
 269
 270             /* Update potential sums from outer loop */
 271             vvdwsum         += vvdw;
 272
 273             fscal            = fvdw;
 274
 275             /* Calculate temporary vectorial force */
 276             tx               = fscal*dx00;
 277             ty               = fscal*dy00;
 278             tz               = fscal*dz00;
 279
 280             /* Update vectorial force */
 281             fix0            += tx;
 282             fiy0            += ty;
 283             fiz0            += tz;
 284             f[j_coord_offset+DIM*0+XX] -= tx;
 285             f[j_coord_offset+DIM*0+YY] -= ty;
 286             f[j_coord_offset+DIM*0+ZZ] -= tz;
 287
 288             }
 289
 290             /**************************
 291              * CALCULATE INTERACTIONS *
 292              **************************/
 293
 294             if (rsq10<rcutoff2)
 295             {
 296
 297             r10              = rsq10*rinv10;
 298
 299             qq10             = iq1*jq0;
 300
 301             /* EWALD ELECTROSTATICS */
 302
 303             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 304             ewrt             = r10*ewtabscale;
 305             ewitab           = ewrt;
 306             eweps            = ewrt-ewitab;
 307             ewitab           = 4*ewitab;
 308             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 309             velec            = qq10*(rinv10-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 310             felec            = qq10*rinv10*(rinvsq10-felec);
 311
 312             d                = r10-rswitch;
 313             d                = (d>0.0) ? d : 0.0;
 314             d2               = d*d;
 315             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 316
 317             dsw              = d2*(swF2+d*(swF3+d*swF4));
 318
 319             /* Evaluate switch function */
 320             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 321             felec            = felec*sw - rinv10*velec*dsw;
 322             velec           *= sw;
 323
 324             /* Update potential sums from outer loop */
 325             velecsum        += velec;
 326
 327             fscal            = felec;
 328
 329             /* Calculate temporary vectorial force */
 330             tx               = fscal*dx10;
 331             ty               = fscal*dy10;
 332             tz               = fscal*dz10;
 333
 334             /* Update vectorial force */
 335             fix1            += tx;
 336             fiy1            += ty;
 337             fiz1            += tz;
 338             f[j_coord_offset+DIM*0+XX] -= tx;
 339             f[j_coord_offset+DIM*0+YY] -= ty;
 340             f[j_coord_offset+DIM*0+ZZ] -= tz;
 341
 342             }
 343
 344             /**************************
 345              * CALCULATE INTERACTIONS *
 346              **************************/
 347
 348             if (rsq20<rcutoff2)
 349             {
 350
 351             r20              = rsq20*rinv20;
 352
 353             qq20             = iq2*jq0;
 354
 355             /* EWALD ELECTROSTATICS */
 356
 357             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 358             ewrt             = r20*ewtabscale;
 359             ewitab           = ewrt;
 360             eweps            = ewrt-ewitab;
 361             ewitab           = 4*ewitab;
 362             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 363             velec            = qq20*(rinv20-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 364             felec            = qq20*rinv20*(rinvsq20-felec);
 365
 366             d                = r20-rswitch;
 367             d                = (d>0.0) ? d : 0.0;
 368             d2               = d*d;
 369             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 370
 371             dsw              = d2*(swF2+d*(swF3+d*swF4));
 372
 373             /* Evaluate switch function */
 374             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 375             felec            = felec*sw - rinv20*velec*dsw;
 376             velec           *= sw;
 377
 378             /* Update potential sums from outer loop */
 379             velecsum        += velec;
 380
 381             fscal            = felec;
 382
 383             /* Calculate temporary vectorial force */
 384             tx               = fscal*dx20;
 385             ty               = fscal*dy20;
 386             tz               = fscal*dz20;
 387
 388             /* Update vectorial force */
 389             fix2            += tx;
 390             fiy2            += ty;
 391             fiz2            += tz;
 392             f[j_coord_offset+DIM*0+XX] -= tx;
 393             f[j_coord_offset+DIM*0+YY] -= ty;
 394             f[j_coord_offset+DIM*0+ZZ] -= tz;
 395
 396             }
 397
 398             /**************************
 399              * CALCULATE INTERACTIONS *
 400              **************************/
 401
 402             if (rsq30<rcutoff2)
 403             {
 404
 405             r30              = rsq30*rinv30;
 406
 407             qq30             = iq3*jq0;
 408
 409             /* EWALD ELECTROSTATICS */
 410
 411             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 412             ewrt             = r30*ewtabscale;
 413             ewitab           = ewrt;
 414             eweps            = ewrt-ewitab;
 415             ewitab           = 4*ewitab;
 416             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 417             velec            = qq30*(rinv30-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 418             felec            = qq30*rinv30*(rinvsq30-felec);
 419
 420             d                = r30-rswitch;
 421             d                = (d>0.0) ? d : 0.0;
 422             d2               = d*d;
 423             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 424
 425             dsw              = d2*(swF2+d*(swF3+d*swF4));
 426
 427             /* Evaluate switch function */
 428             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 429             felec            = felec*sw - rinv30*velec*dsw;
 430             velec           *= sw;
 431
 432             /* Update potential sums from outer loop */
 433             velecsum        += velec;
 434
 435             fscal            = felec;
 436
 437             /* Calculate temporary vectorial force */
 438             tx               = fscal*dx30;
 439             ty               = fscal*dy30;
 440             tz               = fscal*dz30;
 441
 442             /* Update vectorial force */
 443             fix3            += tx;
 444             fiy3            += ty;
 445             fiz3            += tz;
 446             f[j_coord_offset+DIM*0+XX] -= tx;
 447             f[j_coord_offset+DIM*0+YY] -= ty;
 448             f[j_coord_offset+DIM*0+ZZ] -= tz;
 449
 450             }
 451
 452             /* Inner loop uses 256 flops */
 453         }
 454         /* End of innermost loop */
 455
 456         tx = ty = tz = 0;
 457         f[i_coord_offset+DIM*0+XX] += fix0;
 458         f[i_coord_offset+DIM*0+YY] += fiy0;
 459         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 460         tx                         += fix0;
 461         ty                         += fiy0;
 462         tz                         += fiz0;
 463         f[i_coord_offset+DIM*1+XX] += fix1;
 464         f[i_coord_offset+DIM*1+YY] += fiy1;
 465         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 466         tx                         += fix1;
 467         ty                         += fiy1;
 468         tz                         += fiz1;
 469         f[i_coord_offset+DIM*2+XX] += fix2;
 470         f[i_coord_offset+DIM*2+YY] += fiy2;
 471         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 472         tx                         += fix2;
 473         ty                         += fiy2;
 474         tz                         += fiz2;
 475         f[i_coord_offset+DIM*3+XX] += fix3;
 476         f[i_coord_offset+DIM*3+YY] += fiy3;
 477         f[i_coord_offset+DIM*3+ZZ] += fiz3;
 478         tx                         += fix3;
 479         ty                         += fiy3;
 480         tz                         += fiz3;
 481         fshift[i_shift_offset+XX]  += tx;
 482         fshift[i_shift_offset+YY]  += ty;
 483         fshift[i_shift_offset+ZZ]  += tz;
 484
 485         ggid                        = gid[iidx];
 486         /* Update potential energies */
 487         kernel_data->energygrp_elec[ggid] += velecsum;
 488         kernel_data->energygrp_vdw[ggid] += vvdwsum;
 489
 490         /* Increment number of inner iterations */
 491         inneriter                  += j_index_end - j_index_start;
 492
 493         /* Outer loop uses 41 flops */
 494     }
 495
 496     /* Increment number of outer iterations */
 497     outeriter        += nri;
 498
 499     /* Update outer/inner flops */
 500
 501     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_VF,outeriter*41 + inneriter*256);
 502 }
 503 /*
 504  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSw_VdwBhamSw_GeomW4P1_F_c
 505  * Electrostatics interaction: Ewald
 506  * VdW interaction:            Buckingham
 507  * Geometry:                   Water4-Particle
 508  * Calculate force/pot:        Force
 509  */
 510 void
 511 nb_kernel_ElecEwSw_VdwBhamSw_GeomW4P1_F_c
 512                     (t_nblist                    * gmx_restrict       nlist,
 513                      rvec                        * gmx_restrict          xx,
 514                      rvec                        * gmx_restrict          ff,
 515                      t_forcerec                  * gmx_restrict          fr,
 516                      t_mdatoms                   * gmx_restrict     mdatoms,
 517                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
 518                      t_nrnb                      * gmx_restrict        nrnb)
 519 {
 520     int              i_shift_offset,i_coord_offset,j_coord_offset;
 521     int              j_index_start,j_index_end;
 522     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
 523     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
 524     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 525     real             *shiftvec,*fshift,*x,*f;
 526     int              vdwioffset0;
 527     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 528     int              vdwioffset1;
 529     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
 530     int              vdwioffset2;
 531     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
 532     int              vdwioffset3;
 533     real             ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
 534     int              vdwjidx0;
 535     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 536     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
 537     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
 538     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
 539     real             dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30,cexp1_30,cexp2_30;
 540     real             velec,felec,velecsum,facel,crf,krf,krf2;
 541     real             *charge;
 542     int              nvdwtype;
 543     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
 544     int              *vdwtype;
 545     real             *vdwparam;
 546     int              ewitab;
 547     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
 548     real             *ewtab;
 549     real             rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
 550
 551     x                = xx[0];
 552     f                = ff[0];
 553
 554     nri              = nlist->nri;
 555     iinr             = nlist->iinr;
 556     jindex           = nlist->jindex;
 557     jjnr             = nlist->jjnr;
 558     shiftidx         = nlist->shift;
 559     gid              = nlist->gid;
 560     shiftvec         = fr->shift_vec[0];
 561     fshift           = fr->fshift[0];
 562     facel            = fr->epsfac;
 563     charge           = mdatoms->chargeA;
 564     nvdwtype         = fr->ntype;
 565     vdwparam         = fr->nbfp;
 566     vdwtype          = mdatoms->typeA;
 567
 568     sh_ewald         = fr->ic->sh_ewald;
 569     ewtab            = fr->ic->tabq_coul_FDV0;
 570     ewtabscale       = fr->ic->tabq_scale;
 571     ewtabhalfspace   = 0.5/ewtabscale;
 572
 573     /* Setup water-specific parameters */
 574     inr              = nlist->iinr[0];
 575     iq1              = facel*charge[inr+1];
 576     iq2              = facel*charge[inr+2];
 577     iq3              = facel*charge[inr+3];
 578     vdwioffset0      = 3*nvdwtype*vdwtype[inr+0];
 579
 580     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 581     rcutoff          = fr->rcoulomb;
 582     rcutoff2         = rcutoff*rcutoff;
 583
 584     rswitch          = fr->rcoulomb_switch;
 585     /* Setup switch parameters */
 586     d                = rcutoff-rswitch;
 587     swV3             = -10.0/(d*d*d);
 588     swV4             =  15.0/(d*d*d*d);
 589     swV5             =  -6.0/(d*d*d*d*d);
 590     swF2             = -30.0/(d*d*d);
 591     swF3             =  60.0/(d*d*d*d);
 592     swF4             = -30.0/(d*d*d*d*d);
 593
 594     outeriter        = 0;
 595     inneriter        = 0;
 596
 597     /* Start outer loop over neighborlists */
 598     for(iidx=0; iidx<nri; iidx++)
 599     {
 600         /* Load shift vector for this list */
 601         i_shift_offset   = DIM*shiftidx[iidx];
 602         shX              = shiftvec[i_shift_offset+XX];
 603         shY              = shiftvec[i_shift_offset+YY];
 604         shZ              = shiftvec[i_shift_offset+ZZ];
 605
 606         /* Load limits for loop over neighbors */
 607         j_index_start    = jindex[iidx];
 608         j_index_end      = jindex[iidx+1];
 609
 610         /* Get outer coordinate index */
 611         inr              = iinr[iidx];
 612         i_coord_offset   = DIM*inr;
 613
 614         /* Load i particle coords and add shift vector */
 615         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 616         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 617         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 618         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 619         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 620         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 621         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 622         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 623         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 624         ix3              = shX + x[i_coord_offset+DIM*3+XX];
 625         iy3              = shY + x[i_coord_offset+DIM*3+YY];
 626         iz3              = shZ + x[i_coord_offset+DIM*3+ZZ];
 627
 628         fix0             = 0.0;
 629         fiy0             = 0.0;
 630         fiz0             = 0.0;
 631         fix1             = 0.0;
 632         fiy1             = 0.0;
 633         fiz1             = 0.0;
 634         fix2             = 0.0;
 635         fiy2             = 0.0;
 636         fiz2             = 0.0;
 637         fix3             = 0.0;
 638         fiy3             = 0.0;
 639         fiz3             = 0.0;
 640
 641         /* Start inner kernel loop */
 642         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 643         {
 644             /* Get j neighbor index, and coordinate index */
 645             jnr              = jjnr[jidx];
 646             j_coord_offset   = DIM*jnr;
 647
 648             /* load j atom coordinates */
 649             jx0              = x[j_coord_offset+DIM*0+XX];
 650             jy0              = x[j_coord_offset+DIM*0+YY];
 651             jz0              = x[j_coord_offset+DIM*0+ZZ];
 652
 653             /* Calculate displacement vector */
 654             dx00             = ix0 - jx0;
 655             dy00             = iy0 - jy0;
 656             dz00             = iz0 - jz0;
 657             dx10             = ix1 - jx0;
 658             dy10             = iy1 - jy0;
 659             dz10             = iz1 - jz0;
 660             dx20             = ix2 - jx0;
 661             dy20             = iy2 - jy0;
 662             dz20             = iz2 - jz0;
 663             dx30             = ix3 - jx0;
 664             dy30             = iy3 - jy0;
 665             dz30             = iz3 - jz0;
 666
 667             /* Calculate squared distance and things based on it */
 668             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 669             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 670             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 671             rsq30            = dx30*dx30+dy30*dy30+dz30*dz30;
 672
 673             rinv00           = gmx_invsqrt(rsq00);
 674             rinv10           = gmx_invsqrt(rsq10);
 675             rinv20           = gmx_invsqrt(rsq20);
 676             rinv30           = gmx_invsqrt(rsq30);
 677
 678             rinvsq00         = rinv00*rinv00;
 679             rinvsq10         = rinv10*rinv10;
 680             rinvsq20         = rinv20*rinv20;
 681             rinvsq30         = rinv30*rinv30;
 682
 683             /* Load parameters for j particles */
 684             jq0              = charge[jnr+0];
 685             vdwjidx0         = 3*vdwtype[jnr+0];
 686
 687             /**************************
 688              * CALCULATE INTERACTIONS *
 689              **************************/
 690
 691             if (rsq00<rcutoff2)
 692             {
 693
 694             r00              = rsq00*rinv00;
 695
 696             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 697             cexp1_00         = vdwparam[vdwioffset0+vdwjidx0+1];
 698             cexp2_00         = vdwparam[vdwioffset0+vdwjidx0+2];
 699
 700             /* BUCKINGHAM DISPERSION/REPULSION */
 701             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 702             vvdw6            = c6_00*rinvsix;
 703             br               = cexp2_00*r00;
 704             vvdwexp          = cexp1_00*exp(-br);
 705             vvdw             = vvdwexp - vvdw6*(1.0/6.0);
 706             fvdw             = (br*vvdwexp-vvdw6)*rinvsq00;
 707
 708             d                = r00-rswitch;
 709             d                = (d>0.0) ? d : 0.0;
 710             d2               = d*d;
 711             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 712
 713             dsw              = d2*(swF2+d*(swF3+d*swF4));
 714
 715             /* Evaluate switch function */
 716             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 717             fvdw             = fvdw*sw - rinv00*vvdw*dsw;
 718
 719             fscal            = fvdw;
 720
 721             /* Calculate temporary vectorial force */
 722             tx               = fscal*dx00;
 723             ty               = fscal*dy00;
 724             tz               = fscal*dz00;
 725
 726             /* Update vectorial force */
 727             fix0            += tx;
 728             fiy0            += ty;
 729             fiz0            += tz;
 730             f[j_coord_offset+DIM*0+XX] -= tx;
 731             f[j_coord_offset+DIM*0+YY] -= ty;
 732             f[j_coord_offset+DIM*0+ZZ] -= tz;
 733
 734             }
 735
 736             /**************************
 737              * CALCULATE INTERACTIONS *
 738              **************************/
 739
 740             if (rsq10<rcutoff2)
 741             {
 742
 743             r10              = rsq10*rinv10;
 744
 745             qq10             = iq1*jq0;
 746
 747             /* EWALD ELECTROSTATICS */
 748
 749             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 750             ewrt             = r10*ewtabscale;
 751             ewitab           = ewrt;
 752             eweps            = ewrt-ewitab;
 753             ewitab           = 4*ewitab;
 754             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 755             velec            = qq10*(rinv10-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 756             felec            = qq10*rinv10*(rinvsq10-felec);
 757
 758             d                = r10-rswitch;
 759             d                = (d>0.0) ? d : 0.0;
 760             d2               = d*d;
 761             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 762
 763             dsw              = d2*(swF2+d*(swF3+d*swF4));
 764
 765             /* Evaluate switch function */
 766             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 767             felec            = felec*sw - rinv10*velec*dsw;
 768
 769             fscal            = felec;
 770
 771             /* Calculate temporary vectorial force */
 772             tx               = fscal*dx10;
 773             ty               = fscal*dy10;
 774             tz               = fscal*dz10;
 775
 776             /* Update vectorial force */
 777             fix1            += tx;
 778             fiy1            += ty;
 779             fiz1            += tz;
 780             f[j_coord_offset+DIM*0+XX] -= tx;
 781             f[j_coord_offset+DIM*0+YY] -= ty;
 782             f[j_coord_offset+DIM*0+ZZ] -= tz;
 783
 784             }
 785
 786             /**************************
 787              * CALCULATE INTERACTIONS *
 788              **************************/
 789
 790             if (rsq20<rcutoff2)
 791             {
 792
 793             r20              = rsq20*rinv20;
 794
 795             qq20             = iq2*jq0;
 796
 797             /* EWALD ELECTROSTATICS */
 798
 799             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 800             ewrt             = r20*ewtabscale;
 801             ewitab           = ewrt;
 802             eweps            = ewrt-ewitab;
 803             ewitab           = 4*ewitab;
 804             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 805             velec            = qq20*(rinv20-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 806             felec            = qq20*rinv20*(rinvsq20-felec);
 807
 808             d                = r20-rswitch;
 809             d                = (d>0.0) ? d : 0.0;
 810             d2               = d*d;
 811             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 812
 813             dsw              = d2*(swF2+d*(swF3+d*swF4));
 814
 815             /* Evaluate switch function */
 816             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 817             felec            = felec*sw - rinv20*velec*dsw;
 818
 819             fscal            = felec;
 820
 821             /* Calculate temporary vectorial force */
 822             tx               = fscal*dx20;
 823             ty               = fscal*dy20;
 824             tz               = fscal*dz20;
 825
 826             /* Update vectorial force */
 827             fix2            += tx;
 828             fiy2            += ty;
 829             fiz2            += tz;
 830             f[j_coord_offset+DIM*0+XX] -= tx;
 831             f[j_coord_offset+DIM*0+YY] -= ty;
 832             f[j_coord_offset+DIM*0+ZZ] -= tz;
 833
 834             }
 835
 836             /**************************
 837              * CALCULATE INTERACTIONS *
 838              **************************/
 839
 840             if (rsq30<rcutoff2)
 841             {
 842
 843             r30              = rsq30*rinv30;
 844
 845             qq30             = iq3*jq0;
 846
 847             /* EWALD ELECTROSTATICS */
 848
 849             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 850             ewrt             = r30*ewtabscale;
 851             ewitab           = ewrt;
 852             eweps            = ewrt-ewitab;
 853             ewitab           = 4*ewitab;
 854             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 855             velec            = qq30*(rinv30-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 856             felec            = qq30*rinv30*(rinvsq30-felec);
 857
 858             d                = r30-rswitch;
 859             d                = (d>0.0) ? d : 0.0;
 860             d2               = d*d;
 861             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 862
 863             dsw              = d2*(swF2+d*(swF3+d*swF4));
 864
 865             /* Evaluate switch function */
 866             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 867             felec            = felec*sw - rinv30*velec*dsw;
 868
 869             fscal            = felec;
 870
 871             /* Calculate temporary vectorial force */
 872             tx               = fscal*dx30;
 873             ty               = fscal*dy30;
 874             tz               = fscal*dz30;
 875
 876             /* Update vectorial force */
 877             fix3            += tx;
 878             fiy3            += ty;
 879             fiz3            += tz;
 880             f[j_coord_offset+DIM*0+XX] -= tx;
 881             f[j_coord_offset+DIM*0+YY] -= ty;
 882             f[j_coord_offset+DIM*0+ZZ] -= tz;
 883
 884             }
 885
 886             /* Inner loop uses 248 flops */
 887         }
 888         /* End of innermost loop */
 889
 890         tx = ty = tz = 0;
 891         f[i_coord_offset+DIM*0+XX] += fix0;
 892         f[i_coord_offset+DIM*0+YY] += fiy0;
 893         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 894         tx                         += fix0;
 895         ty                         += fiy0;
 896         tz                         += fiz0;
 897         f[i_coord_offset+DIM*1+XX] += fix1;
 898         f[i_coord_offset+DIM*1+YY] += fiy1;
 899         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 900         tx                         += fix1;
 901         ty                         += fiy1;
 902         tz                         += fiz1;
 903         f[i_coord_offset+DIM*2+XX] += fix2;
 904         f[i_coord_offset+DIM*2+YY] += fiy2;
 905         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 906         tx                         += fix2;
 907         ty                         += fiy2;
 908         tz                         += fiz2;
 909         f[i_coord_offset+DIM*3+XX] += fix3;
 910         f[i_coord_offset+DIM*3+YY] += fiy3;
 911         f[i_coord_offset+DIM*3+ZZ] += fiz3;
 912         tx                         += fix3;
 913         ty                         += fiy3;
 914         tz                         += fiz3;
 915         fshift[i_shift_offset+XX]  += tx;
 916         fshift[i_shift_offset+YY]  += ty;
 917         fshift[i_shift_offset+ZZ]  += tz;
 918
 919         /* Increment number of inner iterations */
 920         inneriter                  += j_index_end - j_index_start;
 921
 922         /* Outer loop uses 39 flops */
 923     }
 924
 925     /* Increment number of outer iterations */
 926     outeriter        += nri;
 927
 928     /* Update outer/inner flops */
 929
 930     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_F,outeriter*39 + inneriter*248);
 931 }