src/gromacs/gmxlib/nonbonded/nb_kernel_c/nb_kernel_ElecEwSw_VdwLJSw_GeomP1P1_c.c

   1 /*
   2  * This file is part of the GROMACS molecular simulation package.
   3  *
   4  * Copyright (c) 2012,2013,2014, by the GROMACS development team, led by
   5  * Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl,
   6  * and including many others, as listed in the AUTHORS file in the
   7  * top-level source directory and at http://www.gromacs.org.
   8  *
   9  * GROMACS is free software; you can redistribute it and/or
  10  * modify it under the terms of the GNU Lesser General Public License
  11  * as published by the Free Software Foundation; either version 2.1
  12  * of the License, or (at your option) any later version.
  13  *
  14  * GROMACS is distributed in the hope that it will be useful,
  15  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  16  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  17  * Lesser General Public License for more details.
  18  *
  19  * You should have received a copy of the GNU Lesser General Public
  20  * License along with GROMACS; if not, see
  21  * http://www.gnu.org/licenses, or write to the Free Software Foundation,
  22  * Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA.
  23  *
  24  * If you want to redistribute modifications to GROMACS, please
  25  * consider that scientific software is very special. Version
  26  * control is crucial - bugs must be traceable. We will be happy to
  27  * consider code for inclusion in the official distribution, but
  28  * derived work must not be called official GROMACS. Details are found
  29  * in the README & COPYING files - if they are missing, get the
  30  * official version at http://www.gromacs.org.
  31  *
  32  * To help us fund GROMACS development, we humbly ask that you cite
  33  * the research papers on the package. Check out http://www.gromacs.org.
  34  */
  35 /*
  36  * Note: this file was generated by the GROMACS c kernel generator.
  37  */
  38 #include "config.h"
  39
  40 #include <math.h>
  41
  42 #include "../nb_kernel.h"
  43 #include "gromacs/legacyheaders/types/simple.h"
  44 #include "gromacs/math/vec.h"
  45 #include "gromacs/legacyheaders/nrnb.h"
  46
  47 /*
  48  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSw_VdwLJSw_GeomP1P1_VF_c
  49  * Electrostatics interaction: Ewald
  50  * VdW interaction:            LennardJones
  51  * Geometry:                   Particle-Particle
  52  * Calculate force/pot:        PotentialAndForce
  53  */
  54 void
  55 nb_kernel_ElecEwSw_VdwLJSw_GeomP1P1_VF_c
  56                     (t_nblist                    * gmx_restrict       nlist,
  57                      rvec                        * gmx_restrict          xx,
  58                      rvec                        * gmx_restrict          ff,
  59                      t_forcerec                  * gmx_restrict          fr,
  60                      t_mdatoms                   * gmx_restrict     mdatoms,
  61                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
  62                      t_nrnb                      * gmx_restrict        nrnb)
  63 {
  64     int              i_shift_offset,i_coord_offset,j_coord_offset;
  65     int              j_index_start,j_index_end;
  66     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
  67     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
  68     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
  69     real             *shiftvec,*fshift,*x,*f;
  70     int              vdwioffset0;
  71     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
  72     int              vdwjidx0;
  73     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
  74     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
  75     real             velec,felec,velecsum,facel,crf,krf,krf2;
  76     real             *charge;
  77     int              nvdwtype;
  78     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
  79     int              *vdwtype;
  80     real             *vdwparam;
  81     int              ewitab;
  82     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
  83     real             *ewtab;
  84     real             rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
  85
  86     x                = xx[0];
  87     f                = ff[0];
  88
  89     nri              = nlist->nri;
  90     iinr             = nlist->iinr;
  91     jindex           = nlist->jindex;
  92     jjnr             = nlist->jjnr;
  93     shiftidx         = nlist->shift;
  94     gid              = nlist->gid;
  95     shiftvec         = fr->shift_vec[0];
  96     fshift           = fr->fshift[0];
  97     facel            = fr->epsfac;
  98     charge           = mdatoms->chargeA;
  99     nvdwtype         = fr->ntype;
 100     vdwparam         = fr->nbfp;
 101     vdwtype          = mdatoms->typeA;
 102
 103     sh_ewald         = fr->ic->sh_ewald;
 104     ewtab            = fr->ic->tabq_coul_FDV0;
 105     ewtabscale       = fr->ic->tabq_scale;
 106     ewtabhalfspace   = 0.5/ewtabscale;
 107
 108     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 109     rcutoff          = fr->rcoulomb;
 110     rcutoff2         = rcutoff*rcutoff;
 111
 112     rswitch          = fr->rcoulomb_switch;
 113     /* Setup switch parameters */
 114     d                = rcutoff-rswitch;
 115     swV3             = -10.0/(d*d*d);
 116     swV4             =  15.0/(d*d*d*d);
 117     swV5             =  -6.0/(d*d*d*d*d);
 118     swF2             = -30.0/(d*d*d);
 119     swF3             =  60.0/(d*d*d*d);
 120     swF4             = -30.0/(d*d*d*d*d);
 121
 122     outeriter        = 0;
 123     inneriter        = 0;
 124
 125     /* Start outer loop over neighborlists */
 126     for(iidx=0; iidx<nri; iidx++)
 127     {
 128         /* Load shift vector for this list */
 129         i_shift_offset   = DIM*shiftidx[iidx];
 130         shX              = shiftvec[i_shift_offset+XX];
 131         shY              = shiftvec[i_shift_offset+YY];
 132         shZ              = shiftvec[i_shift_offset+ZZ];
 133
 134         /* Load limits for loop over neighbors */
 135         j_index_start    = jindex[iidx];
 136         j_index_end      = jindex[iidx+1];
 137
 138         /* Get outer coordinate index */
 139         inr              = iinr[iidx];
 140         i_coord_offset   = DIM*inr;
 141
 142         /* Load i particle coords and add shift vector */
 143         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 144         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 145         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 146
 147         fix0             = 0.0;
 148         fiy0             = 0.0;
 149         fiz0             = 0.0;
 150
 151         /* Load parameters for i particles */
 152         iq0              = facel*charge[inr+0];
 153         vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 154
 155         /* Reset potential sums */
 156         velecsum         = 0.0;
 157         vvdwsum          = 0.0;
 158
 159         /* Start inner kernel loop */
 160         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 161         {
 162             /* Get j neighbor index, and coordinate index */
 163             jnr              = jjnr[jidx];
 164             j_coord_offset   = DIM*jnr;
 165
 166             /* load j atom coordinates */
 167             jx0              = x[j_coord_offset+DIM*0+XX];
 168             jy0              = x[j_coord_offset+DIM*0+YY];
 169             jz0              = x[j_coord_offset+DIM*0+ZZ];
 170
 171             /* Calculate displacement vector */
 172             dx00             = ix0 - jx0;
 173             dy00             = iy0 - jy0;
 174             dz00             = iz0 - jz0;
 175
 176             /* Calculate squared distance and things based on it */
 177             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 178
 179             rinv00           = gmx_invsqrt(rsq00);
 180
 181             rinvsq00         = rinv00*rinv00;
 182
 183             /* Load parameters for j particles */
 184             jq0              = charge[jnr+0];
 185             vdwjidx0         = 2*vdwtype[jnr+0];
 186
 187             /**************************
 188              * CALCULATE INTERACTIONS *
 189              **************************/
 190
 191             if (rsq00<rcutoff2)
 192             {
 193
 194             r00              = rsq00*rinv00;
 195
 196             qq00             = iq0*jq0;
 197             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 198             c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 199
 200             /* EWALD ELECTROSTATICS */
 201
 202             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 203             ewrt             = r00*ewtabscale;
 204             ewitab           = ewrt;
 205             eweps            = ewrt-ewitab;
 206             ewitab           = 4*ewitab;
 207             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 208             velec            = qq00*(rinv00-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 209             felec            = qq00*rinv00*(rinvsq00-felec);
 210
 211             /* LENNARD-JONES DISPERSION/REPULSION */
 212
 213             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 214             vvdw6            = c6_00*rinvsix;
 215             vvdw12           = c12_00*rinvsix*rinvsix;
 216             vvdw             = vvdw12*(1.0/12.0) - vvdw6*(1.0/6.0);
 217             fvdw             = (vvdw12-vvdw6)*rinvsq00;
 218
 219             d                = r00-rswitch;
 220             d                = (d>0.0) ? d : 0.0;
 221             d2               = d*d;
 222             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 223
 224             dsw              = d2*(swF2+d*(swF3+d*swF4));
 225
 226             /* Evaluate switch function */
 227             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 228             felec            = felec*sw - rinv00*velec*dsw;
 229             fvdw             = fvdw*sw - rinv00*vvdw*dsw;
 230             velec           *= sw;
 231             vvdw            *= sw;
 232
 233             /* Update potential sums from outer loop */
 234             velecsum        += velec;
 235             vvdwsum         += vvdw;
 236
 237             fscal            = felec+fvdw;
 238
 239             /* Calculate temporary vectorial force */
 240             tx               = fscal*dx00;
 241             ty               = fscal*dy00;
 242             tz               = fscal*dz00;
 243
 244             /* Update vectorial force */
 245             fix0            += tx;
 246             fiy0            += ty;
 247             fiz0            += tz;
 248             f[j_coord_offset+DIM*0+XX] -= tx;
 249             f[j_coord_offset+DIM*0+YY] -= ty;
 250             f[j_coord_offset+DIM*0+ZZ] -= tz;
 251
 252             }
 253
 254             /* Inner loop uses 75 flops */
 255         }
 256         /* End of innermost loop */
 257
 258         tx = ty = tz = 0;
 259         f[i_coord_offset+DIM*0+XX] += fix0;
 260         f[i_coord_offset+DIM*0+YY] += fiy0;
 261         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 262         tx                         += fix0;
 263         ty                         += fiy0;
 264         tz                         += fiz0;
 265         fshift[i_shift_offset+XX]  += tx;
 266         fshift[i_shift_offset+YY]  += ty;
 267         fshift[i_shift_offset+ZZ]  += tz;
 268
 269         ggid                        = gid[iidx];
 270         /* Update potential energies */
 271         kernel_data->energygrp_elec[ggid] += velecsum;
 272         kernel_data->energygrp_vdw[ggid] += vvdwsum;
 273
 274         /* Increment number of inner iterations */
 275         inneriter                  += j_index_end - j_index_start;
 276
 277         /* Outer loop uses 15 flops */
 278     }
 279
 280     /* Increment number of outer iterations */
 281     outeriter        += nri;
 282
 283     /* Update outer/inner flops */
 284
 285     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_VF,outeriter*15 + inneriter*75);
 286 }
 287 /*
 288  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSw_VdwLJSw_GeomP1P1_F_c
 289  * Electrostatics interaction: Ewald
 290  * VdW interaction:            LennardJones
 291  * Geometry:                   Particle-Particle
 292  * Calculate force/pot:        Force
 293  */
 294 void
 295 nb_kernel_ElecEwSw_VdwLJSw_GeomP1P1_F_c
 296                     (t_nblist                    * gmx_restrict       nlist,
 297                      rvec                        * gmx_restrict          xx,
 298                      rvec                        * gmx_restrict          ff,
 299                      t_forcerec                  * gmx_restrict          fr,
 300                      t_mdatoms                   * gmx_restrict     mdatoms,
 301                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
 302                      t_nrnb                      * gmx_restrict        nrnb)
 303 {
 304     int              i_shift_offset,i_coord_offset,j_coord_offset;
 305     int              j_index_start,j_index_end;
 306     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
 307     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
 308     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 309     real             *shiftvec,*fshift,*x,*f;
 310     int              vdwioffset0;
 311     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 312     int              vdwjidx0;
 313     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 314     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
 315     real             velec,felec,velecsum,facel,crf,krf,krf2;
 316     real             *charge;
 317     int              nvdwtype;
 318     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
 319     int              *vdwtype;
 320     real             *vdwparam;
 321     int              ewitab;
 322     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
 323     real             *ewtab;
 324     real             rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
 325
 326     x                = xx[0];
 327     f                = ff[0];
 328
 329     nri              = nlist->nri;
 330     iinr             = nlist->iinr;
 331     jindex           = nlist->jindex;
 332     jjnr             = nlist->jjnr;
 333     shiftidx         = nlist->shift;
 334     gid              = nlist->gid;
 335     shiftvec         = fr->shift_vec[0];
 336     fshift           = fr->fshift[0];
 337     facel            = fr->epsfac;
 338     charge           = mdatoms->chargeA;
 339     nvdwtype         = fr->ntype;
 340     vdwparam         = fr->nbfp;
 341     vdwtype          = mdatoms->typeA;
 342
 343     sh_ewald         = fr->ic->sh_ewald;
 344     ewtab            = fr->ic->tabq_coul_FDV0;
 345     ewtabscale       = fr->ic->tabq_scale;
 346     ewtabhalfspace   = 0.5/ewtabscale;
 347
 348     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 349     rcutoff          = fr->rcoulomb;
 350     rcutoff2         = rcutoff*rcutoff;
 351
 352     rswitch          = fr->rcoulomb_switch;
 353     /* Setup switch parameters */
 354     d                = rcutoff-rswitch;
 355     swV3             = -10.0/(d*d*d);
 356     swV4             =  15.0/(d*d*d*d);
 357     swV5             =  -6.0/(d*d*d*d*d);
 358     swF2             = -30.0/(d*d*d);
 359     swF3             =  60.0/(d*d*d*d);
 360     swF4             = -30.0/(d*d*d*d*d);
 361
 362     outeriter        = 0;
 363     inneriter        = 0;
 364
 365     /* Start outer loop over neighborlists */
 366     for(iidx=0; iidx<nri; iidx++)
 367     {
 368         /* Load shift vector for this list */
 369         i_shift_offset   = DIM*shiftidx[iidx];
 370         shX              = shiftvec[i_shift_offset+XX];
 371         shY              = shiftvec[i_shift_offset+YY];
 372         shZ              = shiftvec[i_shift_offset+ZZ];
 373
 374         /* Load limits for loop over neighbors */
 375         j_index_start    = jindex[iidx];
 376         j_index_end      = jindex[iidx+1];
 377
 378         /* Get outer coordinate index */
 379         inr              = iinr[iidx];
 380         i_coord_offset   = DIM*inr;
 381
 382         /* Load i particle coords and add shift vector */
 383         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 384         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 385         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 386
 387         fix0             = 0.0;
 388         fiy0             = 0.0;
 389         fiz0             = 0.0;
 390
 391         /* Load parameters for i particles */
 392         iq0              = facel*charge[inr+0];
 393         vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 394
 395         /* Start inner kernel loop */
 396         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 397         {
 398             /* Get j neighbor index, and coordinate index */
 399             jnr              = jjnr[jidx];
 400             j_coord_offset   = DIM*jnr;
 401
 402             /* load j atom coordinates */
 403             jx0              = x[j_coord_offset+DIM*0+XX];
 404             jy0              = x[j_coord_offset+DIM*0+YY];
 405             jz0              = x[j_coord_offset+DIM*0+ZZ];
 406
 407             /* Calculate displacement vector */
 408             dx00             = ix0 - jx0;
 409             dy00             = iy0 - jy0;
 410             dz00             = iz0 - jz0;
 411
 412             /* Calculate squared distance and things based on it */
 413             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 414
 415             rinv00           = gmx_invsqrt(rsq00);
 416
 417             rinvsq00         = rinv00*rinv00;
 418
 419             /* Load parameters for j particles */
 420             jq0              = charge[jnr+0];
 421             vdwjidx0         = 2*vdwtype[jnr+0];
 422
 423             /**************************
 424              * CALCULATE INTERACTIONS *
 425              **************************/
 426
 427             if (rsq00<rcutoff2)
 428             {
 429
 430             r00              = rsq00*rinv00;
 431
 432             qq00             = iq0*jq0;
 433             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 434             c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 435
 436             /* EWALD ELECTROSTATICS */
 437
 438             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 439             ewrt             = r00*ewtabscale;
 440             ewitab           = ewrt;
 441             eweps            = ewrt-ewitab;
 442             ewitab           = 4*ewitab;
 443             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 444             velec            = qq00*(rinv00-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 445             felec            = qq00*rinv00*(rinvsq00-felec);
 446
 447             /* LENNARD-JONES DISPERSION/REPULSION */
 448
 449             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 450             vvdw6            = c6_00*rinvsix;
 451             vvdw12           = c12_00*rinvsix*rinvsix;
 452             vvdw             = vvdw12*(1.0/12.0) - vvdw6*(1.0/6.0);
 453             fvdw             = (vvdw12-vvdw6)*rinvsq00;
 454
 455             d                = r00-rswitch;
 456             d                = (d>0.0) ? d : 0.0;
 457             d2               = d*d;
 458             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 459
 460             dsw              = d2*(swF2+d*(swF3+d*swF4));
 461
 462             /* Evaluate switch function */
 463             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 464             felec            = felec*sw - rinv00*velec*dsw;
 465             fvdw             = fvdw*sw - rinv00*vvdw*dsw;
 466
 467             fscal            = felec+fvdw;
 468
 469             /* Calculate temporary vectorial force */
 470             tx               = fscal*dx00;
 471             ty               = fscal*dy00;
 472             tz               = fscal*dz00;
 473
 474             /* Update vectorial force */
 475             fix0            += tx;
 476             fiy0            += ty;
 477             fiz0            += tz;
 478             f[j_coord_offset+DIM*0+XX] -= tx;
 479             f[j_coord_offset+DIM*0+YY] -= ty;
 480             f[j_coord_offset+DIM*0+ZZ] -= tz;
 481
 482             }
 483
 484             /* Inner loop uses 71 flops */
 485         }
 486         /* End of innermost loop */
 487
 488         tx = ty = tz = 0;
 489         f[i_coord_offset+DIM*0+XX] += fix0;
 490         f[i_coord_offset+DIM*0+YY] += fiy0;
 491         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 492         tx                         += fix0;
 493         ty                         += fiy0;
 494         tz                         += fiz0;
 495         fshift[i_shift_offset+XX]  += tx;
 496         fshift[i_shift_offset+YY]  += ty;
 497         fshift[i_shift_offset+ZZ]  += tz;
 498
 499         /* Increment number of inner iterations */
 500         inneriter                  += j_index_end - j_index_start;
 501
 502         /* Outer loop uses 13 flops */
 503     }
 504
 505     /* Increment number of outer iterations */
 506     outeriter        += nri;
 507
 508     /* Update outer/inner flops */
 509
 510     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_F,outeriter*13 + inneriter*71);
 511 }