src/gmxlib/nonbonded/nb_kernel_c/nb_kernel_ElecEwSw_VdwLJSw_GeomW3W3_c.c

   1 /*
   2  * Note: this file was generated by the Gromacs c kernel generator.
   3  *
   4  *                This source code is part of
   5  *
   6  *                 G   R   O   M   A   C   S
   7  *
   8  * Copyright (c) 2001-2012, The GROMACS Development Team
   9  *
  10  * Gromacs is a library for molecular simulation and trajectory analysis,
  11  * written by Erik Lindahl, David van der Spoel, Berk Hess, and others - for
  12  * a full list of developers and information, check out http://www.gromacs.org
  13  *
  14  * This program is free software; you can redistribute it and/or modify it under
  15  * the terms of the GNU Lesser General Public License as published by the Free
  16  * Software Foundation; either version 2 of the License, or (at your option) any
  17  * later version.
  18  *
  19  * To help fund GROMACS development, we humbly ask that you cite
  20  * the papers people have written on it - you can find them on the website.
  21  */
  22 #ifdef HAVE_CONFIG_H
  23 #include <config.h>
  24 #endif
  25
  26 #include <math.h>
  27
  28 #include "../nb_kernel.h"
  29 #include "types/simple.h"
  30 #include "vec.h"
  31 #include "nrnb.h"
  32
  33 /*
  34  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSw_VdwLJSw_GeomW3W3_VF_c
  35  * Electrostatics interaction: Ewald
  36  * VdW interaction:            LennardJones
  37  * Geometry:                   Water3-Water3
  38  * Calculate force/pot:        PotentialAndForce
  39  */
  40 void
  41 nb_kernel_ElecEwSw_VdwLJSw_GeomW3W3_VF_c
  42                     (t_nblist * gmx_restrict                nlist,
  43                      rvec * gmx_restrict                    xx,
  44                      rvec * gmx_restrict                    ff,
  45                      t_forcerec * gmx_restrict              fr,
  46                      t_mdatoms * gmx_restrict               mdatoms,
  47                      nb_kernel_data_t * gmx_restrict        kernel_data,
  48                      t_nrnb * gmx_restrict                  nrnb)
  49 {
  50     int              i_shift_offset,i_coord_offset,j_coord_offset;
  51     int              j_index_start,j_index_end;
  52     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
  53     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
  54     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
  55     real             *shiftvec,*fshift,*x,*f;
  56     int              vdwioffset0;
  57     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
  58     int              vdwioffset1;
  59     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
  60     int              vdwioffset2;
  61     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
  62     int              vdwjidx0;
  63     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
  64     int              vdwjidx1;
  65     real             jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
  66     int              vdwjidx2;
  67     real             jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
  68     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
  69     real             dx01,dy01,dz01,rsq01,rinv01,rinvsq01,r01,qq01,c6_01,c12_01,cexp1_01,cexp2_01;
  70     real             dx02,dy02,dz02,rsq02,rinv02,rinvsq02,r02,qq02,c6_02,c12_02,cexp1_02,cexp2_02;
  71     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
  72     real             dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11,cexp1_11,cexp2_11;
  73     real             dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12,cexp1_12,cexp2_12;
  74     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
  75     real             dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21,cexp1_21,cexp2_21;
  76     real             dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22,cexp1_22,cexp2_22;
  77     real             velec,felec,velecsum,facel,crf,krf,krf2;
  78     real             *charge;
  79     int              nvdwtype;
  80     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
  81     int              *vdwtype;
  82     real             *vdwparam;
  83     int              ewitab;
  84     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
  85     real             *ewtab;
  86     real             rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
  87
  88     x                = xx[0];
  89     f                = ff[0];
  90
  91     nri              = nlist->nri;
  92     iinr             = nlist->iinr;
  93     jindex           = nlist->jindex;
  94     jjnr             = nlist->jjnr;
  95     shiftidx         = nlist->shift;
  96     gid              = nlist->gid;
  97     shiftvec         = fr->shift_vec[0];
  98     fshift           = fr->fshift[0];
  99     facel            = fr->epsfac;
 100     charge           = mdatoms->chargeA;
 101     nvdwtype         = fr->ntype;
 102     vdwparam         = fr->nbfp;
 103     vdwtype          = mdatoms->typeA;
 104
 105     sh_ewald         = fr->ic->sh_ewald;
 106     ewtab            = fr->ic->tabq_coul_FDV0;
 107     ewtabscale       = fr->ic->tabq_scale;
 108     ewtabhalfspace   = 0.5/ewtabscale;
 109
 110     /* Setup water-specific parameters */
 111     inr              = nlist->iinr[0];
 112     iq0              = facel*charge[inr+0];
 113     iq1              = facel*charge[inr+1];
 114     iq2              = facel*charge[inr+2];
 115     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 116
 117     jq0              = charge[inr+0];
 118     jq1              = charge[inr+1];
 119     jq2              = charge[inr+2];
 120     vdwjidx0         = 2*vdwtype[inr+0];
 121     qq00             = iq0*jq0;
 122     c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 123     c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 124     qq01             = iq0*jq1;
 125     qq02             = iq0*jq2;
 126     qq10             = iq1*jq0;
 127     qq11             = iq1*jq1;
 128     qq12             = iq1*jq2;
 129     qq20             = iq2*jq0;
 130     qq21             = iq2*jq1;
 131     qq22             = iq2*jq2;
 132
 133     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 134     rcutoff          = fr->rcoulomb;
 135     rcutoff2         = rcutoff*rcutoff;
 136
 137     rswitch          = fr->rcoulomb_switch;
 138     /* Setup switch parameters */
 139     d                = rcutoff-rswitch;
 140     swV3             = -10.0/(d*d*d);
 141     swV4             =  15.0/(d*d*d*d);
 142     swV5             =  -6.0/(d*d*d*d*d);
 143     swF2             = -30.0/(d*d*d);
 144     swF3             =  60.0/(d*d*d*d);
 145     swF4             = -30.0/(d*d*d*d*d);
 146
 147     outeriter        = 0;
 148     inneriter        = 0;
 149
 150     /* Start outer loop over neighborlists */
 151     for(iidx=0; iidx<nri; iidx++)
 152     {
 153         /* Load shift vector for this list */
 154         i_shift_offset   = DIM*shiftidx[iidx];
 155         shX              = shiftvec[i_shift_offset+XX];
 156         shY              = shiftvec[i_shift_offset+YY];
 157         shZ              = shiftvec[i_shift_offset+ZZ];
 158
 159         /* Load limits for loop over neighbors */
 160         j_index_start    = jindex[iidx];
 161         j_index_end      = jindex[iidx+1];
 162
 163         /* Get outer coordinate index */
 164         inr              = iinr[iidx];
 165         i_coord_offset   = DIM*inr;
 166
 167         /* Load i particle coords and add shift vector */
 168         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 169         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 170         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 171         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 172         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 173         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 174         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 175         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 176         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 177
 178         fix0             = 0.0;
 179         fiy0             = 0.0;
 180         fiz0             = 0.0;
 181         fix1             = 0.0;
 182         fiy1             = 0.0;
 183         fiz1             = 0.0;
 184         fix2             = 0.0;
 185         fiy2             = 0.0;
 186         fiz2             = 0.0;
 187
 188         /* Reset potential sums */
 189         velecsum         = 0.0;
 190         vvdwsum          = 0.0;
 191
 192         /* Start inner kernel loop */
 193         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 194         {
 195             /* Get j neighbor index, and coordinate index */
 196             jnr              = jjnr[jidx];
 197             j_coord_offset   = DIM*jnr;
 198
 199             /* load j atom coordinates */
 200             jx0              = x[j_coord_offset+DIM*0+XX];
 201             jy0              = x[j_coord_offset+DIM*0+YY];
 202             jz0              = x[j_coord_offset+DIM*0+ZZ];
 203             jx1              = x[j_coord_offset+DIM*1+XX];
 204             jy1              = x[j_coord_offset+DIM*1+YY];
 205             jz1              = x[j_coord_offset+DIM*1+ZZ];
 206             jx2              = x[j_coord_offset+DIM*2+XX];
 207             jy2              = x[j_coord_offset+DIM*2+YY];
 208             jz2              = x[j_coord_offset+DIM*2+ZZ];
 209
 210             /* Calculate displacement vector */
 211             dx00             = ix0 - jx0;
 212             dy00             = iy0 - jy0;
 213             dz00             = iz0 - jz0;
 214             dx01             = ix0 - jx1;
 215             dy01             = iy0 - jy1;
 216             dz01             = iz0 - jz1;
 217             dx02             = ix0 - jx2;
 218             dy02             = iy0 - jy2;
 219             dz02             = iz0 - jz2;
 220             dx10             = ix1 - jx0;
 221             dy10             = iy1 - jy0;
 222             dz10             = iz1 - jz0;
 223             dx11             = ix1 - jx1;
 224             dy11             = iy1 - jy1;
 225             dz11             = iz1 - jz1;
 226             dx12             = ix1 - jx2;
 227             dy12             = iy1 - jy2;
 228             dz12             = iz1 - jz2;
 229             dx20             = ix2 - jx0;
 230             dy20             = iy2 - jy0;
 231             dz20             = iz2 - jz0;
 232             dx21             = ix2 - jx1;
 233             dy21             = iy2 - jy1;
 234             dz21             = iz2 - jz1;
 235             dx22             = ix2 - jx2;
 236             dy22             = iy2 - jy2;
 237             dz22             = iz2 - jz2;
 238
 239             /* Calculate squared distance and things based on it */
 240             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 241             rsq01            = dx01*dx01+dy01*dy01+dz01*dz01;
 242             rsq02            = dx02*dx02+dy02*dy02+dz02*dz02;
 243             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 244             rsq11            = dx11*dx11+dy11*dy11+dz11*dz11;
 245             rsq12            = dx12*dx12+dy12*dy12+dz12*dz12;
 246             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 247             rsq21            = dx21*dx21+dy21*dy21+dz21*dz21;
 248             rsq22            = dx22*dx22+dy22*dy22+dz22*dz22;
 249
 250             rinv00           = gmx_invsqrt(rsq00);
 251             rinv01           = gmx_invsqrt(rsq01);
 252             rinv02           = gmx_invsqrt(rsq02);
 253             rinv10           = gmx_invsqrt(rsq10);
 254             rinv11           = gmx_invsqrt(rsq11);
 255             rinv12           = gmx_invsqrt(rsq12);
 256             rinv20           = gmx_invsqrt(rsq20);
 257             rinv21           = gmx_invsqrt(rsq21);
 258             rinv22           = gmx_invsqrt(rsq22);
 259
 260             rinvsq00         = rinv00*rinv00;
 261             rinvsq01         = rinv01*rinv01;
 262             rinvsq02         = rinv02*rinv02;
 263             rinvsq10         = rinv10*rinv10;
 264             rinvsq11         = rinv11*rinv11;
 265             rinvsq12         = rinv12*rinv12;
 266             rinvsq20         = rinv20*rinv20;
 267             rinvsq21         = rinv21*rinv21;
 268             rinvsq22         = rinv22*rinv22;
 269
 270             /**************************
 271              * CALCULATE INTERACTIONS *
 272              **************************/
 273
 274             if (rsq00<rcutoff2)
 275             {
 276
 277             r00              = rsq00*rinv00;
 278
 279             /* EWALD ELECTROSTATICS */
 280
 281             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 282             ewrt             = r00*ewtabscale;
 283             ewitab           = ewrt;
 284             eweps            = ewrt-ewitab;
 285             ewitab           = 4*ewitab;
 286             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 287             velec            = qq00*(rinv00-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 288             felec            = qq00*rinv00*(rinvsq00-felec);
 289
 290             /* LENNARD-JONES DISPERSION/REPULSION */
 291
 292             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 293             vvdw6            = c6_00*rinvsix;
 294             vvdw12           = c12_00*rinvsix*rinvsix;
 295             vvdw             = vvdw12*(1.0/12.0) - vvdw6*(1.0/6.0);
 296             fvdw             = (vvdw12-vvdw6)*rinvsq00;
 297
 298             d                = r00-rswitch;
 299             d                = (d>0.0) ? d : 0.0;
 300             d2               = d*d;
 301             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 302
 303             dsw              = d2*(swF2+d*(swF3+d*swF4));
 304
 305             /* Evaluate switch function */
 306             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 307             felec            = felec*sw - rinv00*velec*dsw;
 308             fvdw             = fvdw*sw - rinv00*vvdw*dsw;
 309             velec           *= sw;
 310             vvdw            *= sw;
 311
 312             /* Update potential sums from outer loop */
 313             velecsum        += velec;
 314             vvdwsum         += vvdw;
 315
 316             fscal            = felec+fvdw;
 317
 318             /* Calculate temporary vectorial force */
 319             tx               = fscal*dx00;
 320             ty               = fscal*dy00;
 321             tz               = fscal*dz00;
 322
 323             /* Update vectorial force */
 324             fix0            += tx;
 325             fiy0            += ty;
 326             fiz0            += tz;
 327             f[j_coord_offset+DIM*0+XX] -= tx;
 328             f[j_coord_offset+DIM*0+YY] -= ty;
 329             f[j_coord_offset+DIM*0+ZZ] -= tz;
 330
 331             }
 332
 333             /**************************
 334              * CALCULATE INTERACTIONS *
 335              **************************/
 336
 337             if (rsq01<rcutoff2)
 338             {
 339
 340             r01              = rsq01*rinv01;
 341
 342             /* EWALD ELECTROSTATICS */
 343
 344             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 345             ewrt             = r01*ewtabscale;
 346             ewitab           = ewrt;
 347             eweps            = ewrt-ewitab;
 348             ewitab           = 4*ewitab;
 349             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 350             velec            = qq01*(rinv01-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 351             felec            = qq01*rinv01*(rinvsq01-felec);
 352
 353             d                = r01-rswitch;
 354             d                = (d>0.0) ? d : 0.0;
 355             d2               = d*d;
 356             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 357
 358             dsw              = d2*(swF2+d*(swF3+d*swF4));
 359
 360             /* Evaluate switch function */
 361             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 362             felec            = felec*sw - rinv01*velec*dsw;
 363             velec           *= sw;
 364
 365             /* Update potential sums from outer loop */
 366             velecsum        += velec;
 367
 368             fscal            = felec;
 369
 370             /* Calculate temporary vectorial force */
 371             tx               = fscal*dx01;
 372             ty               = fscal*dy01;
 373             tz               = fscal*dz01;
 374
 375             /* Update vectorial force */
 376             fix0            += tx;
 377             fiy0            += ty;
 378             fiz0            += tz;
 379             f[j_coord_offset+DIM*1+XX] -= tx;
 380             f[j_coord_offset+DIM*1+YY] -= ty;
 381             f[j_coord_offset+DIM*1+ZZ] -= tz;
 382
 383             }
 384
 385             /**************************
 386              * CALCULATE INTERACTIONS *
 387              **************************/
 388
 389             if (rsq02<rcutoff2)
 390             {
 391
 392             r02              = rsq02*rinv02;
 393
 394             /* EWALD ELECTROSTATICS */
 395
 396             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 397             ewrt             = r02*ewtabscale;
 398             ewitab           = ewrt;
 399             eweps            = ewrt-ewitab;
 400             ewitab           = 4*ewitab;
 401             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 402             velec            = qq02*(rinv02-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 403             felec            = qq02*rinv02*(rinvsq02-felec);
 404
 405             d                = r02-rswitch;
 406             d                = (d>0.0) ? d : 0.0;
 407             d2               = d*d;
 408             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 409
 410             dsw              = d2*(swF2+d*(swF3+d*swF4));
 411
 412             /* Evaluate switch function */
 413             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 414             felec            = felec*sw - rinv02*velec*dsw;
 415             velec           *= sw;
 416
 417             /* Update potential sums from outer loop */
 418             velecsum        += velec;
 419
 420             fscal            = felec;
 421
 422             /* Calculate temporary vectorial force */
 423             tx               = fscal*dx02;
 424             ty               = fscal*dy02;
 425             tz               = fscal*dz02;
 426
 427             /* Update vectorial force */
 428             fix0            += tx;
 429             fiy0            += ty;
 430             fiz0            += tz;
 431             f[j_coord_offset+DIM*2+XX] -= tx;
 432             f[j_coord_offset+DIM*2+YY] -= ty;
 433             f[j_coord_offset+DIM*2+ZZ] -= tz;
 434
 435             }
 436
 437             /**************************
 438              * CALCULATE INTERACTIONS *
 439              **************************/
 440
 441             if (rsq10<rcutoff2)
 442             {
 443
 444             r10              = rsq10*rinv10;
 445
 446             /* EWALD ELECTROSTATICS */
 447
 448             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 449             ewrt             = r10*ewtabscale;
 450             ewitab           = ewrt;
 451             eweps            = ewrt-ewitab;
 452             ewitab           = 4*ewitab;
 453             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 454             velec            = qq10*(rinv10-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 455             felec            = qq10*rinv10*(rinvsq10-felec);
 456
 457             d                = r10-rswitch;
 458             d                = (d>0.0) ? d : 0.0;
 459             d2               = d*d;
 460             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 461
 462             dsw              = d2*(swF2+d*(swF3+d*swF4));
 463
 464             /* Evaluate switch function */
 465             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 466             felec            = felec*sw - rinv10*velec*dsw;
 467             velec           *= sw;
 468
 469             /* Update potential sums from outer loop */
 470             velecsum        += velec;
 471
 472             fscal            = felec;
 473
 474             /* Calculate temporary vectorial force */
 475             tx               = fscal*dx10;
 476             ty               = fscal*dy10;
 477             tz               = fscal*dz10;
 478
 479             /* Update vectorial force */
 480             fix1            += tx;
 481             fiy1            += ty;
 482             fiz1            += tz;
 483             f[j_coord_offset+DIM*0+XX] -= tx;
 484             f[j_coord_offset+DIM*0+YY] -= ty;
 485             f[j_coord_offset+DIM*0+ZZ] -= tz;
 486
 487             }
 488
 489             /**************************
 490              * CALCULATE INTERACTIONS *
 491              **************************/
 492
 493             if (rsq11<rcutoff2)
 494             {
 495
 496             r11              = rsq11*rinv11;
 497
 498             /* EWALD ELECTROSTATICS */
 499
 500             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 501             ewrt             = r11*ewtabscale;
 502             ewitab           = ewrt;
 503             eweps            = ewrt-ewitab;
 504             ewitab           = 4*ewitab;
 505             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 506             velec            = qq11*(rinv11-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 507             felec            = qq11*rinv11*(rinvsq11-felec);
 508
 509             d                = r11-rswitch;
 510             d                = (d>0.0) ? d : 0.0;
 511             d2               = d*d;
 512             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 513
 514             dsw              = d2*(swF2+d*(swF3+d*swF4));
 515
 516             /* Evaluate switch function */
 517             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 518             felec            = felec*sw - rinv11*velec*dsw;
 519             velec           *= sw;
 520
 521             /* Update potential sums from outer loop */
 522             velecsum        += velec;
 523
 524             fscal            = felec;
 525
 526             /* Calculate temporary vectorial force */
 527             tx               = fscal*dx11;
 528             ty               = fscal*dy11;
 529             tz               = fscal*dz11;
 530
 531             /* Update vectorial force */
 532             fix1            += tx;
 533             fiy1            += ty;
 534             fiz1            += tz;
 535             f[j_coord_offset+DIM*1+XX] -= tx;
 536             f[j_coord_offset+DIM*1+YY] -= ty;
 537             f[j_coord_offset+DIM*1+ZZ] -= tz;
 538
 539             }
 540
 541             /**************************
 542              * CALCULATE INTERACTIONS *
 543              **************************/
 544
 545             if (rsq12<rcutoff2)
 546             {
 547
 548             r12              = rsq12*rinv12;
 549
 550             /* EWALD ELECTROSTATICS */
 551
 552             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 553             ewrt             = r12*ewtabscale;
 554             ewitab           = ewrt;
 555             eweps            = ewrt-ewitab;
 556             ewitab           = 4*ewitab;
 557             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 558             velec            = qq12*(rinv12-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 559             felec            = qq12*rinv12*(rinvsq12-felec);
 560
 561             d                = r12-rswitch;
 562             d                = (d>0.0) ? d : 0.0;
 563             d2               = d*d;
 564             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 565
 566             dsw              = d2*(swF2+d*(swF3+d*swF4));
 567
 568             /* Evaluate switch function */
 569             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 570             felec            = felec*sw - rinv12*velec*dsw;
 571             velec           *= sw;
 572
 573             /* Update potential sums from outer loop */
 574             velecsum        += velec;
 575
 576             fscal            = felec;
 577
 578             /* Calculate temporary vectorial force */
 579             tx               = fscal*dx12;
 580             ty               = fscal*dy12;
 581             tz               = fscal*dz12;
 582
 583             /* Update vectorial force */
 584             fix1            += tx;
 585             fiy1            += ty;
 586             fiz1            += tz;
 587             f[j_coord_offset+DIM*2+XX] -= tx;
 588             f[j_coord_offset+DIM*2+YY] -= ty;
 589             f[j_coord_offset+DIM*2+ZZ] -= tz;
 590
 591             }
 592
 593             /**************************
 594              * CALCULATE INTERACTIONS *
 595              **************************/
 596
 597             if (rsq20<rcutoff2)
 598             {
 599
 600             r20              = rsq20*rinv20;
 601
 602             /* EWALD ELECTROSTATICS */
 603
 604             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 605             ewrt             = r20*ewtabscale;
 606             ewitab           = ewrt;
 607             eweps            = ewrt-ewitab;
 608             ewitab           = 4*ewitab;
 609             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 610             velec            = qq20*(rinv20-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 611             felec            = qq20*rinv20*(rinvsq20-felec);
 612
 613             d                = r20-rswitch;
 614             d                = (d>0.0) ? d : 0.0;
 615             d2               = d*d;
 616             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 617
 618             dsw              = d2*(swF2+d*(swF3+d*swF4));
 619
 620             /* Evaluate switch function */
 621             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 622             felec            = felec*sw - rinv20*velec*dsw;
 623             velec           *= sw;
 624
 625             /* Update potential sums from outer loop */
 626             velecsum        += velec;
 627
 628             fscal            = felec;
 629
 630             /* Calculate temporary vectorial force */
 631             tx               = fscal*dx20;
 632             ty               = fscal*dy20;
 633             tz               = fscal*dz20;
 634
 635             /* Update vectorial force */
 636             fix2            += tx;
 637             fiy2            += ty;
 638             fiz2            += tz;
 639             f[j_coord_offset+DIM*0+XX] -= tx;
 640             f[j_coord_offset+DIM*0+YY] -= ty;
 641             f[j_coord_offset+DIM*0+ZZ] -= tz;
 642
 643             }
 644
 645             /**************************
 646              * CALCULATE INTERACTIONS *
 647              **************************/
 648
 649             if (rsq21<rcutoff2)
 650             {
 651
 652             r21              = rsq21*rinv21;
 653
 654             /* EWALD ELECTROSTATICS */
 655
 656             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 657             ewrt             = r21*ewtabscale;
 658             ewitab           = ewrt;
 659             eweps            = ewrt-ewitab;
 660             ewitab           = 4*ewitab;
 661             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 662             velec            = qq21*(rinv21-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 663             felec            = qq21*rinv21*(rinvsq21-felec);
 664
 665             d                = r21-rswitch;
 666             d                = (d>0.0) ? d : 0.0;
 667             d2               = d*d;
 668             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 669
 670             dsw              = d2*(swF2+d*(swF3+d*swF4));
 671
 672             /* Evaluate switch function */
 673             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 674             felec            = felec*sw - rinv21*velec*dsw;
 675             velec           *= sw;
 676
 677             /* Update potential sums from outer loop */
 678             velecsum        += velec;
 679
 680             fscal            = felec;
 681
 682             /* Calculate temporary vectorial force */
 683             tx               = fscal*dx21;
 684             ty               = fscal*dy21;
 685             tz               = fscal*dz21;
 686
 687             /* Update vectorial force */
 688             fix2            += tx;
 689             fiy2            += ty;
 690             fiz2            += tz;
 691             f[j_coord_offset+DIM*1+XX] -= tx;
 692             f[j_coord_offset+DIM*1+YY] -= ty;
 693             f[j_coord_offset+DIM*1+ZZ] -= tz;
 694
 695             }
 696
 697             /**************************
 698              * CALCULATE INTERACTIONS *
 699              **************************/
 700
 701             if (rsq22<rcutoff2)
 702             {
 703
 704             r22              = rsq22*rinv22;
 705
 706             /* EWALD ELECTROSTATICS */
 707
 708             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 709             ewrt             = r22*ewtabscale;
 710             ewitab           = ewrt;
 711             eweps            = ewrt-ewitab;
 712             ewitab           = 4*ewitab;
 713             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 714             velec            = qq22*(rinv22-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 715             felec            = qq22*rinv22*(rinvsq22-felec);
 716
 717             d                = r22-rswitch;
 718             d                = (d>0.0) ? d : 0.0;
 719             d2               = d*d;
 720             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 721
 722             dsw              = d2*(swF2+d*(swF3+d*swF4));
 723
 724             /* Evaluate switch function */
 725             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 726             felec            = felec*sw - rinv22*velec*dsw;
 727             velec           *= sw;
 728
 729             /* Update potential sums from outer loop */
 730             velecsum        += velec;
 731
 732             fscal            = felec;
 733
 734             /* Calculate temporary vectorial force */
 735             tx               = fscal*dx22;
 736             ty               = fscal*dy22;
 737             tz               = fscal*dz22;
 738
 739             /* Update vectorial force */
 740             fix2            += tx;
 741             fiy2            += ty;
 742             fiz2            += tz;
 743             f[j_coord_offset+DIM*2+XX] -= tx;
 744             f[j_coord_offset+DIM*2+YY] -= ty;
 745             f[j_coord_offset+DIM*2+ZZ] -= tz;
 746
 747             }
 748
 749             /* Inner loop uses 538 flops */
 750         }
 751         /* End of innermost loop */
 752
 753         tx = ty = tz = 0;
 754         f[i_coord_offset+DIM*0+XX] += fix0;
 755         f[i_coord_offset+DIM*0+YY] += fiy0;
 756         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 757         tx                         += fix0;
 758         ty                         += fiy0;
 759         tz                         += fiz0;
 760         f[i_coord_offset+DIM*1+XX] += fix1;
 761         f[i_coord_offset+DIM*1+YY] += fiy1;
 762         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 763         tx                         += fix1;
 764         ty                         += fiy1;
 765         tz                         += fiz1;
 766         f[i_coord_offset+DIM*2+XX] += fix2;
 767         f[i_coord_offset+DIM*2+YY] += fiy2;
 768         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 769         tx                         += fix2;
 770         ty                         += fiy2;
 771         tz                         += fiz2;
 772         fshift[i_shift_offset+XX]  += tx;
 773         fshift[i_shift_offset+YY]  += ty;
 774         fshift[i_shift_offset+ZZ]  += tz;
 775
 776         ggid                        = gid[iidx];
 777         /* Update potential energies */
 778         kernel_data->energygrp_elec[ggid] += velecsum;
 779         kernel_data->energygrp_vdw[ggid] += vvdwsum;
 780
 781         /* Increment number of inner iterations */
 782         inneriter                  += j_index_end - j_index_start;
 783
 784         /* Outer loop uses 32 flops */
 785     }
 786
 787     /* Increment number of outer iterations */
 788     outeriter        += nri;
 789
 790     /* Update outer/inner flops */
 791
 792     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3W3_VF,outeriter*32 + inneriter*538);
 793 }
 794 /*
 795  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSw_VdwLJSw_GeomW3W3_F_c
 796  * Electrostatics interaction: Ewald
 797  * VdW interaction:            LennardJones
 798  * Geometry:                   Water3-Water3
 799  * Calculate force/pot:        Force
 800  */
 801 void
 802 nb_kernel_ElecEwSw_VdwLJSw_GeomW3W3_F_c
 803                     (t_nblist * gmx_restrict                nlist,
 804                      rvec * gmx_restrict                    xx,
 805                      rvec * gmx_restrict                    ff,
 806                      t_forcerec * gmx_restrict              fr,
 807                      t_mdatoms * gmx_restrict               mdatoms,
 808                      nb_kernel_data_t * gmx_restrict        kernel_data,
 809                      t_nrnb * gmx_restrict                  nrnb)
 810 {
 811     int              i_shift_offset,i_coord_offset,j_coord_offset;
 812     int              j_index_start,j_index_end;
 813     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
 814     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
 815     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 816     real             *shiftvec,*fshift,*x,*f;
 817     int              vdwioffset0;
 818     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 819     int              vdwioffset1;
 820     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
 821     int              vdwioffset2;
 822     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
 823     int              vdwjidx0;
 824     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 825     int              vdwjidx1;
 826     real             jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
 827     int              vdwjidx2;
 828     real             jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
 829     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
 830     real             dx01,dy01,dz01,rsq01,rinv01,rinvsq01,r01,qq01,c6_01,c12_01,cexp1_01,cexp2_01;
 831     real             dx02,dy02,dz02,rsq02,rinv02,rinvsq02,r02,qq02,c6_02,c12_02,cexp1_02,cexp2_02;
 832     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
 833     real             dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11,cexp1_11,cexp2_11;
 834     real             dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12,cexp1_12,cexp2_12;
 835     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
 836     real             dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21,cexp1_21,cexp2_21;
 837     real             dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22,cexp1_22,cexp2_22;
 838     real             velec,felec,velecsum,facel,crf,krf,krf2;
 839     real             *charge;
 840     int              nvdwtype;
 841     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
 842     int              *vdwtype;
 843     real             *vdwparam;
 844     int              ewitab;
 845     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
 846     real             *ewtab;
 847     real             rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
 848
 849     x                = xx[0];
 850     f                = ff[0];
 851
 852     nri              = nlist->nri;
 853     iinr             = nlist->iinr;
 854     jindex           = nlist->jindex;
 855     jjnr             = nlist->jjnr;
 856     shiftidx         = nlist->shift;
 857     gid              = nlist->gid;
 858     shiftvec         = fr->shift_vec[0];
 859     fshift           = fr->fshift[0];
 860     facel            = fr->epsfac;
 861     charge           = mdatoms->chargeA;
 862     nvdwtype         = fr->ntype;
 863     vdwparam         = fr->nbfp;
 864     vdwtype          = mdatoms->typeA;
 865
 866     sh_ewald         = fr->ic->sh_ewald;
 867     ewtab            = fr->ic->tabq_coul_FDV0;
 868     ewtabscale       = fr->ic->tabq_scale;
 869     ewtabhalfspace   = 0.5/ewtabscale;
 870
 871     /* Setup water-specific parameters */
 872     inr              = nlist->iinr[0];
 873     iq0              = facel*charge[inr+0];
 874     iq1              = facel*charge[inr+1];
 875     iq2              = facel*charge[inr+2];
 876     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 877
 878     jq0              = charge[inr+0];
 879     jq1              = charge[inr+1];
 880     jq2              = charge[inr+2];
 881     vdwjidx0         = 2*vdwtype[inr+0];
 882     qq00             = iq0*jq0;
 883     c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 884     c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 885     qq01             = iq0*jq1;
 886     qq02             = iq0*jq2;
 887     qq10             = iq1*jq0;
 888     qq11             = iq1*jq1;
 889     qq12             = iq1*jq2;
 890     qq20             = iq2*jq0;
 891     qq21             = iq2*jq1;
 892     qq22             = iq2*jq2;
 893
 894     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 895     rcutoff          = fr->rcoulomb;
 896     rcutoff2         = rcutoff*rcutoff;
 897
 898     rswitch          = fr->rcoulomb_switch;
 899     /* Setup switch parameters */
 900     d                = rcutoff-rswitch;
 901     swV3             = -10.0/(d*d*d);
 902     swV4             =  15.0/(d*d*d*d);
 903     swV5             =  -6.0/(d*d*d*d*d);
 904     swF2             = -30.0/(d*d*d);
 905     swF3             =  60.0/(d*d*d*d);
 906     swF4             = -30.0/(d*d*d*d*d);
 907
 908     outeriter        = 0;
 909     inneriter        = 0;
 910
 911     /* Start outer loop over neighborlists */
 912     for(iidx=0; iidx<nri; iidx++)
 913     {
 914         /* Load shift vector for this list */
 915         i_shift_offset   = DIM*shiftidx[iidx];
 916         shX              = shiftvec[i_shift_offset+XX];
 917         shY              = shiftvec[i_shift_offset+YY];
 918         shZ              = shiftvec[i_shift_offset+ZZ];
 919
 920         /* Load limits for loop over neighbors */
 921         j_index_start    = jindex[iidx];
 922         j_index_end      = jindex[iidx+1];
 923
 924         /* Get outer coordinate index */
 925         inr              = iinr[iidx];
 926         i_coord_offset   = DIM*inr;
 927
 928         /* Load i particle coords and add shift vector */
 929         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 930         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 931         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 932         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 933         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 934         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 935         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 936         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 937         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 938
 939         fix0             = 0.0;
 940         fiy0             = 0.0;
 941         fiz0             = 0.0;
 942         fix1             = 0.0;
 943         fiy1             = 0.0;
 944         fiz1             = 0.0;
 945         fix2             = 0.0;
 946         fiy2             = 0.0;
 947         fiz2             = 0.0;
 948
 949         /* Start inner kernel loop */
 950         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 951         {
 952             /* Get j neighbor index, and coordinate index */
 953             jnr              = jjnr[jidx];
 954             j_coord_offset   = DIM*jnr;
 955
 956             /* load j atom coordinates */
 957             jx0              = x[j_coord_offset+DIM*0+XX];
 958             jy0              = x[j_coord_offset+DIM*0+YY];
 959             jz0              = x[j_coord_offset+DIM*0+ZZ];
 960             jx1              = x[j_coord_offset+DIM*1+XX];
 961             jy1              = x[j_coord_offset+DIM*1+YY];
 962             jz1              = x[j_coord_offset+DIM*1+ZZ];
 963             jx2              = x[j_coord_offset+DIM*2+XX];
 964             jy2              = x[j_coord_offset+DIM*2+YY];
 965             jz2              = x[j_coord_offset+DIM*2+ZZ];
 966
 967             /* Calculate displacement vector */
 968             dx00             = ix0 - jx0;
 969             dy00             = iy0 - jy0;
 970             dz00             = iz0 - jz0;
 971             dx01             = ix0 - jx1;
 972             dy01             = iy0 - jy1;
 973             dz01             = iz0 - jz1;
 974             dx02             = ix0 - jx2;
 975             dy02             = iy0 - jy2;
 976             dz02             = iz0 - jz2;
 977             dx10             = ix1 - jx0;
 978             dy10             = iy1 - jy0;
 979             dz10             = iz1 - jz0;
 980             dx11             = ix1 - jx1;
 981             dy11             = iy1 - jy1;
 982             dz11             = iz1 - jz1;
 983             dx12             = ix1 - jx2;
 984             dy12             = iy1 - jy2;
 985             dz12             = iz1 - jz2;
 986             dx20             = ix2 - jx0;
 987             dy20             = iy2 - jy0;
 988             dz20             = iz2 - jz0;
 989             dx21             = ix2 - jx1;
 990             dy21             = iy2 - jy1;
 991             dz21             = iz2 - jz1;
 992             dx22             = ix2 - jx2;
 993             dy22             = iy2 - jy2;
 994             dz22             = iz2 - jz2;
 995
 996             /* Calculate squared distance and things based on it */
 997             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 998             rsq01            = dx01*dx01+dy01*dy01+dz01*dz01;
 999             rsq02            = dx02*dx02+dy02*dy02+dz02*dz02;
1000             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
1001             rsq11            = dx11*dx11+dy11*dy11+dz11*dz11;
1002             rsq12            = dx12*dx12+dy12*dy12+dz12*dz12;
1003             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
1004             rsq21            = dx21*dx21+dy21*dy21+dz21*dz21;
1005             rsq22            = dx22*dx22+dy22*dy22+dz22*dz22;
1006
1007             rinv00           = gmx_invsqrt(rsq00);
1008             rinv01           = gmx_invsqrt(rsq01);
1009             rinv02           = gmx_invsqrt(rsq02);
1010             rinv10           = gmx_invsqrt(rsq10);
1011             rinv11           = gmx_invsqrt(rsq11);
1012             rinv12           = gmx_invsqrt(rsq12);
1013             rinv20           = gmx_invsqrt(rsq20);
1014             rinv21           = gmx_invsqrt(rsq21);
1015             rinv22           = gmx_invsqrt(rsq22);
1016
1017             rinvsq00         = rinv00*rinv00;
1018             rinvsq01         = rinv01*rinv01;
1019             rinvsq02         = rinv02*rinv02;
1020             rinvsq10         = rinv10*rinv10;
1021             rinvsq11         = rinv11*rinv11;
1022             rinvsq12         = rinv12*rinv12;
1023             rinvsq20         = rinv20*rinv20;
1024             rinvsq21         = rinv21*rinv21;
1025             rinvsq22         = rinv22*rinv22;
1026
1027             /**************************
1028              * CALCULATE INTERACTIONS *
1029              **************************/
1030
1031             if (rsq00<rcutoff2)
1032             {
1033
1034             r00              = rsq00*rinv00;
1035
1036             /* EWALD ELECTROSTATICS */
1037
1038             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1039             ewrt             = r00*ewtabscale;
1040             ewitab           = ewrt;
1041             eweps            = ewrt-ewitab;
1042             ewitab           = 4*ewitab;
1043             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1044             velec            = qq00*(rinv00-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1045             felec            = qq00*rinv00*(rinvsq00-felec);
1046
1047             /* LENNARD-JONES DISPERSION/REPULSION */
1048
1049             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
1050             vvdw6            = c6_00*rinvsix;
1051             vvdw12           = c12_00*rinvsix*rinvsix;
1052             vvdw             = vvdw12*(1.0/12.0) - vvdw6*(1.0/6.0);
1053             fvdw             = (vvdw12-vvdw6)*rinvsq00;
1054
1055             d                = r00-rswitch;
1056             d                = (d>0.0) ? d : 0.0;
1057             d2               = d*d;
1058             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1059
1060             dsw              = d2*(swF2+d*(swF3+d*swF4));
1061
1062             /* Evaluate switch function */
1063             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1064             felec            = felec*sw - rinv00*velec*dsw;
1065             fvdw             = fvdw*sw - rinv00*vvdw*dsw;
1066
1067             fscal            = felec+fvdw;
1068
1069             /* Calculate temporary vectorial force */
1070             tx               = fscal*dx00;
1071             ty               = fscal*dy00;
1072             tz               = fscal*dz00;
1073
1074             /* Update vectorial force */
1075             fix0            += tx;
1076             fiy0            += ty;
1077             fiz0            += tz;
1078             f[j_coord_offset+DIM*0+XX] -= tx;
1079             f[j_coord_offset+DIM*0+YY] -= ty;
1080             f[j_coord_offset+DIM*0+ZZ] -= tz;
1081
1082             }
1083
1084             /**************************
1085              * CALCULATE INTERACTIONS *
1086              **************************/
1087
1088             if (rsq01<rcutoff2)
1089             {
1090
1091             r01              = rsq01*rinv01;
1092
1093             /* EWALD ELECTROSTATICS */
1094
1095             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1096             ewrt             = r01*ewtabscale;
1097             ewitab           = ewrt;
1098             eweps            = ewrt-ewitab;
1099             ewitab           = 4*ewitab;
1100             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1101             velec            = qq01*(rinv01-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1102             felec            = qq01*rinv01*(rinvsq01-felec);
1103
1104             d                = r01-rswitch;
1105             d                = (d>0.0) ? d : 0.0;
1106             d2               = d*d;
1107             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1108
1109             dsw              = d2*(swF2+d*(swF3+d*swF4));
1110
1111             /* Evaluate switch function */
1112             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1113             felec            = felec*sw - rinv01*velec*dsw;
1114
1115             fscal            = felec;
1116
1117             /* Calculate temporary vectorial force */
1118             tx               = fscal*dx01;
1119             ty               = fscal*dy01;
1120             tz               = fscal*dz01;
1121
1122             /* Update vectorial force */
1123             fix0            += tx;
1124             fiy0            += ty;
1125             fiz0            += tz;
1126             f[j_coord_offset+DIM*1+XX] -= tx;
1127             f[j_coord_offset+DIM*1+YY] -= ty;
1128             f[j_coord_offset+DIM*1+ZZ] -= tz;
1129
1130             }
1131
1132             /**************************
1133              * CALCULATE INTERACTIONS *
1134              **************************/
1135
1136             if (rsq02<rcutoff2)
1137             {
1138
1139             r02              = rsq02*rinv02;
1140
1141             /* EWALD ELECTROSTATICS */
1142
1143             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1144             ewrt             = r02*ewtabscale;
1145             ewitab           = ewrt;
1146             eweps            = ewrt-ewitab;
1147             ewitab           = 4*ewitab;
1148             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1149             velec            = qq02*(rinv02-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1150             felec            = qq02*rinv02*(rinvsq02-felec);
1151
1152             d                = r02-rswitch;
1153             d                = (d>0.0) ? d : 0.0;
1154             d2               = d*d;
1155             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1156
1157             dsw              = d2*(swF2+d*(swF3+d*swF4));
1158
1159             /* Evaluate switch function */
1160             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1161             felec            = felec*sw - rinv02*velec*dsw;
1162
1163             fscal            = felec;
1164
1165             /* Calculate temporary vectorial force */
1166             tx               = fscal*dx02;
1167             ty               = fscal*dy02;
1168             tz               = fscal*dz02;
1169
1170             /* Update vectorial force */
1171             fix0            += tx;
1172             fiy0            += ty;
1173             fiz0            += tz;
1174             f[j_coord_offset+DIM*2+XX] -= tx;
1175             f[j_coord_offset+DIM*2+YY] -= ty;
1176             f[j_coord_offset+DIM*2+ZZ] -= tz;
1177
1178             }
1179
1180             /**************************
1181              * CALCULATE INTERACTIONS *
1182              **************************/
1183
1184             if (rsq10<rcutoff2)
1185             {
1186
1187             r10              = rsq10*rinv10;
1188
1189             /* EWALD ELECTROSTATICS */
1190
1191             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1192             ewrt             = r10*ewtabscale;
1193             ewitab           = ewrt;
1194             eweps            = ewrt-ewitab;
1195             ewitab           = 4*ewitab;
1196             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1197             velec            = qq10*(rinv10-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1198             felec            = qq10*rinv10*(rinvsq10-felec);
1199
1200             d                = r10-rswitch;
1201             d                = (d>0.0) ? d : 0.0;
1202             d2               = d*d;
1203             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1204
1205             dsw              = d2*(swF2+d*(swF3+d*swF4));
1206
1207             /* Evaluate switch function */
1208             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1209             felec            = felec*sw - rinv10*velec*dsw;
1210
1211             fscal            = felec;
1212
1213             /* Calculate temporary vectorial force */
1214             tx               = fscal*dx10;
1215             ty               = fscal*dy10;
1216             tz               = fscal*dz10;
1217
1218             /* Update vectorial force */
1219             fix1            += tx;
1220             fiy1            += ty;
1221             fiz1            += tz;
1222             f[j_coord_offset+DIM*0+XX] -= tx;
1223             f[j_coord_offset+DIM*0+YY] -= ty;
1224             f[j_coord_offset+DIM*0+ZZ] -= tz;
1225
1226             }
1227
1228             /**************************
1229              * CALCULATE INTERACTIONS *
1230              **************************/
1231
1232             if (rsq11<rcutoff2)
1233             {
1234
1235             r11              = rsq11*rinv11;
1236
1237             /* EWALD ELECTROSTATICS */
1238
1239             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1240             ewrt             = r11*ewtabscale;
1241             ewitab           = ewrt;
1242             eweps            = ewrt-ewitab;
1243             ewitab           = 4*ewitab;
1244             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1245             velec            = qq11*(rinv11-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1246             felec            = qq11*rinv11*(rinvsq11-felec);
1247
1248             d                = r11-rswitch;
1249             d                = (d>0.0) ? d : 0.0;
1250             d2               = d*d;
1251             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1252
1253             dsw              = d2*(swF2+d*(swF3+d*swF4));
1254
1255             /* Evaluate switch function */
1256             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1257             felec            = felec*sw - rinv11*velec*dsw;
1258
1259             fscal            = felec;
1260
1261             /* Calculate temporary vectorial force */
1262             tx               = fscal*dx11;
1263             ty               = fscal*dy11;
1264             tz               = fscal*dz11;
1265
1266             /* Update vectorial force */
1267             fix1            += tx;
1268             fiy1            += ty;
1269             fiz1            += tz;
1270             f[j_coord_offset+DIM*1+XX] -= tx;
1271             f[j_coord_offset+DIM*1+YY] -= ty;
1272             f[j_coord_offset+DIM*1+ZZ] -= tz;
1273
1274             }
1275
1276             /**************************
1277              * CALCULATE INTERACTIONS *
1278              **************************/
1279
1280             if (rsq12<rcutoff2)
1281             {
1282
1283             r12              = rsq12*rinv12;
1284
1285             /* EWALD ELECTROSTATICS */
1286
1287             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1288             ewrt             = r12*ewtabscale;
1289             ewitab           = ewrt;
1290             eweps            = ewrt-ewitab;
1291             ewitab           = 4*ewitab;
1292             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1293             velec            = qq12*(rinv12-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1294             felec            = qq12*rinv12*(rinvsq12-felec);
1295
1296             d                = r12-rswitch;
1297             d                = (d>0.0) ? d : 0.0;
1298             d2               = d*d;
1299             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1300
1301             dsw              = d2*(swF2+d*(swF3+d*swF4));
1302
1303             /* Evaluate switch function */
1304             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1305             felec            = felec*sw - rinv12*velec*dsw;
1306
1307             fscal            = felec;
1308
1309             /* Calculate temporary vectorial force */
1310             tx               = fscal*dx12;
1311             ty               = fscal*dy12;
1312             tz               = fscal*dz12;
1313
1314             /* Update vectorial force */
1315             fix1            += tx;
1316             fiy1            += ty;
1317             fiz1            += tz;
1318             f[j_coord_offset+DIM*2+XX] -= tx;
1319             f[j_coord_offset+DIM*2+YY] -= ty;
1320             f[j_coord_offset+DIM*2+ZZ] -= tz;
1321
1322             }
1323
1324             /**************************
1325              * CALCULATE INTERACTIONS *
1326              **************************/
1327
1328             if (rsq20<rcutoff2)
1329             {
1330
1331             r20              = rsq20*rinv20;
1332
1333             /* EWALD ELECTROSTATICS */
1334
1335             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1336             ewrt             = r20*ewtabscale;
1337             ewitab           = ewrt;
1338             eweps            = ewrt-ewitab;
1339             ewitab           = 4*ewitab;
1340             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1341             velec            = qq20*(rinv20-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1342             felec            = qq20*rinv20*(rinvsq20-felec);
1343
1344             d                = r20-rswitch;
1345             d                = (d>0.0) ? d : 0.0;
1346             d2               = d*d;
1347             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1348
1349             dsw              = d2*(swF2+d*(swF3+d*swF4));
1350
1351             /* Evaluate switch function */
1352             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1353             felec            = felec*sw - rinv20*velec*dsw;
1354
1355             fscal            = felec;
1356
1357             /* Calculate temporary vectorial force */
1358             tx               = fscal*dx20;
1359             ty               = fscal*dy20;
1360             tz               = fscal*dz20;
1361
1362             /* Update vectorial force */
1363             fix2            += tx;
1364             fiy2            += ty;
1365             fiz2            += tz;
1366             f[j_coord_offset+DIM*0+XX] -= tx;
1367             f[j_coord_offset+DIM*0+YY] -= ty;
1368             f[j_coord_offset+DIM*0+ZZ] -= tz;
1369
1370             }
1371
1372             /**************************
1373              * CALCULATE INTERACTIONS *
1374              **************************/
1375
1376             if (rsq21<rcutoff2)
1377             {
1378
1379             r21              = rsq21*rinv21;
1380
1381             /* EWALD ELECTROSTATICS */
1382
1383             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1384             ewrt             = r21*ewtabscale;
1385             ewitab           = ewrt;
1386             eweps            = ewrt-ewitab;
1387             ewitab           = 4*ewitab;
1388             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1389             velec            = qq21*(rinv21-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1390             felec            = qq21*rinv21*(rinvsq21-felec);
1391
1392             d                = r21-rswitch;
1393             d                = (d>0.0) ? d : 0.0;
1394             d2               = d*d;
1395             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1396
1397             dsw              = d2*(swF2+d*(swF3+d*swF4));
1398
1399             /* Evaluate switch function */
1400             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1401             felec            = felec*sw - rinv21*velec*dsw;
1402
1403             fscal            = felec;
1404
1405             /* Calculate temporary vectorial force */
1406             tx               = fscal*dx21;
1407             ty               = fscal*dy21;
1408             tz               = fscal*dz21;
1409
1410             /* Update vectorial force */
1411             fix2            += tx;
1412             fiy2            += ty;
1413             fiz2            += tz;
1414             f[j_coord_offset+DIM*1+XX] -= tx;
1415             f[j_coord_offset+DIM*1+YY] -= ty;
1416             f[j_coord_offset+DIM*1+ZZ] -= tz;
1417
1418             }
1419
1420             /**************************
1421              * CALCULATE INTERACTIONS *
1422              **************************/
1423
1424             if (rsq22<rcutoff2)
1425             {
1426
1427             r22              = rsq22*rinv22;
1428
1429             /* EWALD ELECTROSTATICS */
1430
1431             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1432             ewrt             = r22*ewtabscale;
1433             ewitab           = ewrt;
1434             eweps            = ewrt-ewitab;
1435             ewitab           = 4*ewitab;
1436             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1437             velec            = qq22*(rinv22-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1438             felec            = qq22*rinv22*(rinvsq22-felec);
1439
1440             d                = r22-rswitch;
1441             d                = (d>0.0) ? d : 0.0;
1442             d2               = d*d;
1443             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1444
1445             dsw              = d2*(swF2+d*(swF3+d*swF4));
1446
1447             /* Evaluate switch function */
1448             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1449             felec            = felec*sw - rinv22*velec*dsw;
1450
1451             fscal            = felec;
1452
1453             /* Calculate temporary vectorial force */
1454             tx               = fscal*dx22;
1455             ty               = fscal*dy22;
1456             tz               = fscal*dz22;
1457
1458             /* Update vectorial force */
1459             fix2            += tx;
1460             fiy2            += ty;
1461             fiz2            += tz;
1462             f[j_coord_offset+DIM*2+XX] -= tx;
1463             f[j_coord_offset+DIM*2+YY] -= ty;
1464             f[j_coord_offset+DIM*2+ZZ] -= tz;
1465
1466             }
1467
1468             /* Inner loop uses 518 flops */
1469         }
1470         /* End of innermost loop */
1471
1472         tx = ty = tz = 0;
1473         f[i_coord_offset+DIM*0+XX] += fix0;
1474         f[i_coord_offset+DIM*0+YY] += fiy0;
1475         f[i_coord_offset+DIM*0+ZZ] += fiz0;
1476         tx                         += fix0;
1477         ty                         += fiy0;
1478         tz                         += fiz0;
1479         f[i_coord_offset+DIM*1+XX] += fix1;
1480         f[i_coord_offset+DIM*1+YY] += fiy1;
1481         f[i_coord_offset+DIM*1+ZZ] += fiz1;
1482         tx                         += fix1;
1483         ty                         += fiy1;
1484         tz                         += fiz1;
1485         f[i_coord_offset+DIM*2+XX] += fix2;
1486         f[i_coord_offset+DIM*2+YY] += fiy2;
1487         f[i_coord_offset+DIM*2+ZZ] += fiz2;
1488         tx                         += fix2;
1489         ty                         += fiy2;
1490         tz                         += fiz2;
1491         fshift[i_shift_offset+XX]  += tx;
1492         fshift[i_shift_offset+YY]  += ty;
1493         fshift[i_shift_offset+ZZ]  += tz;
1494
1495         /* Increment number of inner iterations */
1496         inneriter                  += j_index_end - j_index_start;
1497
1498         /* Outer loop uses 30 flops */
1499     }
1500
1501     /* Increment number of outer iterations */
1502     outeriter        += nri;
1503
1504     /* Update outer/inner flops */
1505
1506     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3W3_F,outeriter*30 + inneriter*518);
1507 }