src/gmxlib/nonbonded/nb_kernel_c/nb_kernel_ElecEwSw_VdwLJSw_GeomP1P1_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_GeomP1P1_VF_c
  35  * Electrostatics interaction: Ewald
  36  * VdW interaction:            LennardJones
  37  * Geometry:                   Particle-Particle
  38  * Calculate force/pot:        PotentialAndForce
  39  */
  40 void
  41 nb_kernel_ElecEwSw_VdwLJSw_GeomP1P1_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              vdwjidx0;
  59     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
  60     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
  61     real             velec,felec,velecsum,facel,crf,krf,krf2;
  62     real             *charge;
  63     int              nvdwtype;
  64     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
  65     int              *vdwtype;
  66     real             *vdwparam;
  67     int              ewitab;
  68     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
  69     real             *ewtab;
  70     real             rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
  71
  72     x                = xx[0];
  73     f                = ff[0];
  74
  75     nri              = nlist->nri;
  76     iinr             = nlist->iinr;
  77     jindex           = nlist->jindex;
  78     jjnr             = nlist->jjnr;
  79     shiftidx         = nlist->shift;
  80     gid              = nlist->gid;
  81     shiftvec         = fr->shift_vec[0];
  82     fshift           = fr->fshift[0];
  83     facel            = fr->epsfac;
  84     charge           = mdatoms->chargeA;
  85     nvdwtype         = fr->ntype;
  86     vdwparam         = fr->nbfp;
  87     vdwtype          = mdatoms->typeA;
  88
  89     sh_ewald         = fr->ic->sh_ewald;
  90     ewtab            = fr->ic->tabq_coul_FDV0;
  91     ewtabscale       = fr->ic->tabq_scale;
  92     ewtabhalfspace   = 0.5/ewtabscale;
  93
  94     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
  95     rcutoff          = fr->rcoulomb;
  96     rcutoff2         = rcutoff*rcutoff;
  97
  98     rswitch          = fr->rcoulomb_switch;
  99     /* Setup switch parameters */
 100     d                = rcutoff-rswitch;
 101     swV3             = -10.0/(d*d*d);
 102     swV4             =  15.0/(d*d*d*d);
 103     swV5             =  -6.0/(d*d*d*d*d);
 104     swF2             = -30.0/(d*d*d);
 105     swF3             =  60.0/(d*d*d*d);
 106     swF4             = -30.0/(d*d*d*d*d);
 107
 108     outeriter        = 0;
 109     inneriter        = 0;
 110
 111     /* Start outer loop over neighborlists */
 112     for(iidx=0; iidx<nri; iidx++)
 113     {
 114         /* Load shift vector for this list */
 115         i_shift_offset   = DIM*shiftidx[iidx];
 116         shX              = shiftvec[i_shift_offset+XX];
 117         shY              = shiftvec[i_shift_offset+YY];
 118         shZ              = shiftvec[i_shift_offset+ZZ];
 119
 120         /* Load limits for loop over neighbors */
 121         j_index_start    = jindex[iidx];
 122         j_index_end      = jindex[iidx+1];
 123
 124         /* Get outer coordinate index */
 125         inr              = iinr[iidx];
 126         i_coord_offset   = DIM*inr;
 127
 128         /* Load i particle coords and add shift vector */
 129         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 130         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 131         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 132
 133         fix0             = 0.0;
 134         fiy0             = 0.0;
 135         fiz0             = 0.0;
 136
 137         /* Load parameters for i particles */
 138         iq0              = facel*charge[inr+0];
 139         vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 140
 141         /* Reset potential sums */
 142         velecsum         = 0.0;
 143         vvdwsum          = 0.0;
 144
 145         /* Start inner kernel loop */
 146         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 147         {
 148             /* Get j neighbor index, and coordinate index */
 149             jnr              = jjnr[jidx];
 150             j_coord_offset   = DIM*jnr;
 151
 152             /* load j atom coordinates */
 153             jx0              = x[j_coord_offset+DIM*0+XX];
 154             jy0              = x[j_coord_offset+DIM*0+YY];
 155             jz0              = x[j_coord_offset+DIM*0+ZZ];
 156
 157             /* Calculate displacement vector */
 158             dx00             = ix0 - jx0;
 159             dy00             = iy0 - jy0;
 160             dz00             = iz0 - jz0;
 161
 162             /* Calculate squared distance and things based on it */
 163             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 164
 165             rinv00           = gmx_invsqrt(rsq00);
 166
 167             rinvsq00         = rinv00*rinv00;
 168
 169             /* Load parameters for j particles */
 170             jq0              = charge[jnr+0];
 171             vdwjidx0         = 2*vdwtype[jnr+0];
 172
 173             /**************************
 174              * CALCULATE INTERACTIONS *
 175              **************************/
 176
 177             if (rsq00<rcutoff2)
 178             {
 179
 180             r00              = rsq00*rinv00;
 181
 182             qq00             = iq0*jq0;
 183             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 184             c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 185
 186             /* EWALD ELECTROSTATICS */
 187
 188             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 189             ewrt             = r00*ewtabscale;
 190             ewitab           = ewrt;
 191             eweps            = ewrt-ewitab;
 192             ewitab           = 4*ewitab;
 193             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 194             velec            = qq00*(rinv00-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 195             felec            = qq00*rinv00*(rinvsq00-felec);
 196
 197             /* LENNARD-JONES DISPERSION/REPULSION */
 198
 199             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 200             vvdw6            = c6_00*rinvsix;
 201             vvdw12           = c12_00*rinvsix*rinvsix;
 202             vvdw             = vvdw12*(1.0/12.0) - vvdw6*(1.0/6.0);
 203             fvdw             = (vvdw12-vvdw6)*rinvsq00;
 204
 205             d                = r00-rswitch;
 206             d                = (d>0.0) ? d : 0.0;
 207             d2               = d*d;
 208             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 209
 210             dsw              = d2*(swF2+d*(swF3+d*swF4));
 211
 212             /* Evaluate switch function */
 213             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 214             felec            = felec*sw - rinv00*velec*dsw;
 215             fvdw             = fvdw*sw - rinv00*vvdw*dsw;
 216             velec           *= sw;
 217             vvdw            *= sw;
 218
 219             /* Update potential sums from outer loop */
 220             velecsum        += velec;
 221             vvdwsum         += vvdw;
 222
 223             fscal            = felec+fvdw;
 224
 225             /* Calculate temporary vectorial force */
 226             tx               = fscal*dx00;
 227             ty               = fscal*dy00;
 228             tz               = fscal*dz00;
 229
 230             /* Update vectorial force */
 231             fix0            += tx;
 232             fiy0            += ty;
 233             fiz0            += tz;
 234             f[j_coord_offset+DIM*0+XX] -= tx;
 235             f[j_coord_offset+DIM*0+YY] -= ty;
 236             f[j_coord_offset+DIM*0+ZZ] -= tz;
 237
 238             }
 239
 240             /* Inner loop uses 75 flops */
 241         }
 242         /* End of innermost loop */
 243
 244         tx = ty = tz = 0;
 245         f[i_coord_offset+DIM*0+XX] += fix0;
 246         f[i_coord_offset+DIM*0+YY] += fiy0;
 247         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 248         tx                         += fix0;
 249         ty                         += fiy0;
 250         tz                         += fiz0;
 251         fshift[i_shift_offset+XX]  += tx;
 252         fshift[i_shift_offset+YY]  += ty;
 253         fshift[i_shift_offset+ZZ]  += tz;
 254
 255         ggid                        = gid[iidx];
 256         /* Update potential energies */
 257         kernel_data->energygrp_elec[ggid] += velecsum;
 258         kernel_data->energygrp_vdw[ggid] += vvdwsum;
 259
 260         /* Increment number of inner iterations */
 261         inneriter                  += j_index_end - j_index_start;
 262
 263         /* Outer loop uses 15 flops */
 264     }
 265
 266     /* Increment number of outer iterations */
 267     outeriter        += nri;
 268
 269     /* Update outer/inner flops */
 270
 271     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_VF,outeriter*15 + inneriter*75);
 272 }
 273 /*
 274  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSw_VdwLJSw_GeomP1P1_F_c
 275  * Electrostatics interaction: Ewald
 276  * VdW interaction:            LennardJones
 277  * Geometry:                   Particle-Particle
 278  * Calculate force/pot:        Force
 279  */
 280 void
 281 nb_kernel_ElecEwSw_VdwLJSw_GeomP1P1_F_c
 282                     (t_nblist * gmx_restrict                nlist,
 283                      rvec * gmx_restrict                    xx,
 284                      rvec * gmx_restrict                    ff,
 285                      t_forcerec * gmx_restrict              fr,
 286                      t_mdatoms * gmx_restrict               mdatoms,
 287                      nb_kernel_data_t * gmx_restrict        kernel_data,
 288                      t_nrnb * gmx_restrict                  nrnb)
 289 {
 290     int              i_shift_offset,i_coord_offset,j_coord_offset;
 291     int              j_index_start,j_index_end;
 292     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
 293     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
 294     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 295     real             *shiftvec,*fshift,*x,*f;
 296     int              vdwioffset0;
 297     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 298     int              vdwjidx0;
 299     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 300     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
 301     real             velec,felec,velecsum,facel,crf,krf,krf2;
 302     real             *charge;
 303     int              nvdwtype;
 304     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
 305     int              *vdwtype;
 306     real             *vdwparam;
 307     int              ewitab;
 308     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
 309     real             *ewtab;
 310     real             rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
 311
 312     x                = xx[0];
 313     f                = ff[0];
 314
 315     nri              = nlist->nri;
 316     iinr             = nlist->iinr;
 317     jindex           = nlist->jindex;
 318     jjnr             = nlist->jjnr;
 319     shiftidx         = nlist->shift;
 320     gid              = nlist->gid;
 321     shiftvec         = fr->shift_vec[0];
 322     fshift           = fr->fshift[0];
 323     facel            = fr->epsfac;
 324     charge           = mdatoms->chargeA;
 325     nvdwtype         = fr->ntype;
 326     vdwparam         = fr->nbfp;
 327     vdwtype          = mdatoms->typeA;
 328
 329     sh_ewald         = fr->ic->sh_ewald;
 330     ewtab            = fr->ic->tabq_coul_FDV0;
 331     ewtabscale       = fr->ic->tabq_scale;
 332     ewtabhalfspace   = 0.5/ewtabscale;
 333
 334     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 335     rcutoff          = fr->rcoulomb;
 336     rcutoff2         = rcutoff*rcutoff;
 337
 338     rswitch          = fr->rcoulomb_switch;
 339     /* Setup switch parameters */
 340     d                = rcutoff-rswitch;
 341     swV3             = -10.0/(d*d*d);
 342     swV4             =  15.0/(d*d*d*d);
 343     swV5             =  -6.0/(d*d*d*d*d);
 344     swF2             = -30.0/(d*d*d);
 345     swF3             =  60.0/(d*d*d*d);
 346     swF4             = -30.0/(d*d*d*d*d);
 347
 348     outeriter        = 0;
 349     inneriter        = 0;
 350
 351     /* Start outer loop over neighborlists */
 352     for(iidx=0; iidx<nri; iidx++)
 353     {
 354         /* Load shift vector for this list */
 355         i_shift_offset   = DIM*shiftidx[iidx];
 356         shX              = shiftvec[i_shift_offset+XX];
 357         shY              = shiftvec[i_shift_offset+YY];
 358         shZ              = shiftvec[i_shift_offset+ZZ];
 359
 360         /* Load limits for loop over neighbors */
 361         j_index_start    = jindex[iidx];
 362         j_index_end      = jindex[iidx+1];
 363
 364         /* Get outer coordinate index */
 365         inr              = iinr[iidx];
 366         i_coord_offset   = DIM*inr;
 367
 368         /* Load i particle coords and add shift vector */
 369         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 370         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 371         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 372
 373         fix0             = 0.0;
 374         fiy0             = 0.0;
 375         fiz0             = 0.0;
 376
 377         /* Load parameters for i particles */
 378         iq0              = facel*charge[inr+0];
 379         vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 380
 381         /* Start inner kernel loop */
 382         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 383         {
 384             /* Get j neighbor index, and coordinate index */
 385             jnr              = jjnr[jidx];
 386             j_coord_offset   = DIM*jnr;
 387
 388             /* load j atom coordinates */
 389             jx0              = x[j_coord_offset+DIM*0+XX];
 390             jy0              = x[j_coord_offset+DIM*0+YY];
 391             jz0              = x[j_coord_offset+DIM*0+ZZ];
 392
 393             /* Calculate displacement vector */
 394             dx00             = ix0 - jx0;
 395             dy00             = iy0 - jy0;
 396             dz00             = iz0 - jz0;
 397
 398             /* Calculate squared distance and things based on it */
 399             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 400
 401             rinv00           = gmx_invsqrt(rsq00);
 402
 403             rinvsq00         = rinv00*rinv00;
 404
 405             /* Load parameters for j particles */
 406             jq0              = charge[jnr+0];
 407             vdwjidx0         = 2*vdwtype[jnr+0];
 408
 409             /**************************
 410              * CALCULATE INTERACTIONS *
 411              **************************/
 412
 413             if (rsq00<rcutoff2)
 414             {
 415
 416             r00              = rsq00*rinv00;
 417
 418             qq00             = iq0*jq0;
 419             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 420             c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 421
 422             /* EWALD ELECTROSTATICS */
 423
 424             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 425             ewrt             = r00*ewtabscale;
 426             ewitab           = ewrt;
 427             eweps            = ewrt-ewitab;
 428             ewitab           = 4*ewitab;
 429             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 430             velec            = qq00*(rinv00-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 431             felec            = qq00*rinv00*(rinvsq00-felec);
 432
 433             /* LENNARD-JONES DISPERSION/REPULSION */
 434
 435             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 436             vvdw6            = c6_00*rinvsix;
 437             vvdw12           = c12_00*rinvsix*rinvsix;
 438             vvdw             = vvdw12*(1.0/12.0) - vvdw6*(1.0/6.0);
 439             fvdw             = (vvdw12-vvdw6)*rinvsq00;
 440
 441             d                = r00-rswitch;
 442             d                = (d>0.0) ? d : 0.0;
 443             d2               = d*d;
 444             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 445
 446             dsw              = d2*(swF2+d*(swF3+d*swF4));
 447
 448             /* Evaluate switch function */
 449             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 450             felec            = felec*sw - rinv00*velec*dsw;
 451             fvdw             = fvdw*sw - rinv00*vvdw*dsw;
 452
 453             fscal            = felec+fvdw;
 454
 455             /* Calculate temporary vectorial force */
 456             tx               = fscal*dx00;
 457             ty               = fscal*dy00;
 458             tz               = fscal*dz00;
 459
 460             /* Update vectorial force */
 461             fix0            += tx;
 462             fiy0            += ty;
 463             fiz0            += tz;
 464             f[j_coord_offset+DIM*0+XX] -= tx;
 465             f[j_coord_offset+DIM*0+YY] -= ty;
 466             f[j_coord_offset+DIM*0+ZZ] -= tz;
 467
 468             }
 469
 470             /* Inner loop uses 71 flops */
 471         }
 472         /* End of innermost loop */
 473
 474         tx = ty = tz = 0;
 475         f[i_coord_offset+DIM*0+XX] += fix0;
 476         f[i_coord_offset+DIM*0+YY] += fiy0;
 477         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 478         tx                         += fix0;
 479         ty                         += fiy0;
 480         tz                         += fiz0;
 481         fshift[i_shift_offset+XX]  += tx;
 482         fshift[i_shift_offset+YY]  += ty;
 483         fshift[i_shift_offset+ZZ]  += tz;
 484
 485         /* Increment number of inner iterations */
 486         inneriter                  += j_index_end - j_index_start;
 487
 488         /* Outer loop uses 13 flops */
 489     }
 490
 491     /* Increment number of outer iterations */
 492     outeriter        += nri;
 493
 494     /* Update outer/inner flops */
 495
 496     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_F,outeriter*13 + inneriter*71);
 497 }