src/gromacs/gmxlib/nonbonded/nb_kernel_c/nb_kernel_ElecEwSw_VdwNone_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_VdwNone_GeomP1P1_VF_c
  35  * Electrostatics interaction: Ewald
  36  * VdW interaction:            None
  37  * Geometry:                   Particle-Particle
  38  * Calculate force/pot:        PotentialAndForce
  39  */
  40 void
  41 nb_kernel_ElecEwSw_VdwNone_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              ewitab;
  64     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
  65     real             *ewtab;
  66     real             rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
  67
  68     x                = xx[0];
  69     f                = ff[0];
  70
  71     nri              = nlist->nri;
  72     iinr             = nlist->iinr;
  73     jindex           = nlist->jindex;
  74     jjnr             = nlist->jjnr;
  75     shiftidx         = nlist->shift;
  76     gid              = nlist->gid;
  77     shiftvec         = fr->shift_vec[0];
  78     fshift           = fr->fshift[0];
  79     facel            = fr->epsfac;
  80     charge           = mdatoms->chargeA;
  81
  82     sh_ewald         = fr->ic->sh_ewald;
  83     ewtab            = fr->ic->tabq_coul_FDV0;
  84     ewtabscale       = fr->ic->tabq_scale;
  85     ewtabhalfspace   = 0.5/ewtabscale;
  86
  87     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
  88     rcutoff          = fr->rcoulomb;
  89     rcutoff2         = rcutoff*rcutoff;
  90
  91     rswitch          = fr->rcoulomb_switch;
  92     /* Setup switch parameters */
  93     d                = rcutoff-rswitch;
  94     swV3             = -10.0/(d*d*d);
  95     swV4             =  15.0/(d*d*d*d);
  96     swV5             =  -6.0/(d*d*d*d*d);
  97     swF2             = -30.0/(d*d*d);
  98     swF3             =  60.0/(d*d*d*d);
  99     swF4             = -30.0/(d*d*d*d*d);
 100
 101     outeriter        = 0;
 102     inneriter        = 0;
 103
 104     /* Start outer loop over neighborlists */
 105     for(iidx=0; iidx<nri; iidx++)
 106     {
 107         /* Load shift vector for this list */
 108         i_shift_offset   = DIM*shiftidx[iidx];
 109         shX              = shiftvec[i_shift_offset+XX];
 110         shY              = shiftvec[i_shift_offset+YY];
 111         shZ              = shiftvec[i_shift_offset+ZZ];
 112
 113         /* Load limits for loop over neighbors */
 114         j_index_start    = jindex[iidx];
 115         j_index_end      = jindex[iidx+1];
 116
 117         /* Get outer coordinate index */
 118         inr              = iinr[iidx];
 119         i_coord_offset   = DIM*inr;
 120
 121         /* Load i particle coords and add shift vector */
 122         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 123         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 124         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 125
 126         fix0             = 0.0;
 127         fiy0             = 0.0;
 128         fiz0             = 0.0;
 129
 130         /* Load parameters for i particles */
 131         iq0              = facel*charge[inr+0];
 132
 133         /* Reset potential sums */
 134         velecsum         = 0.0;
 135
 136         /* Start inner kernel loop */
 137         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 138         {
 139             /* Get j neighbor index, and coordinate index */
 140             jnr              = jjnr[jidx];
 141             j_coord_offset   = DIM*jnr;
 142
 143             /* load j atom coordinates */
 144             jx0              = x[j_coord_offset+DIM*0+XX];
 145             jy0              = x[j_coord_offset+DIM*0+YY];
 146             jz0              = x[j_coord_offset+DIM*0+ZZ];
 147
 148             /* Calculate displacement vector */
 149             dx00             = ix0 - jx0;
 150             dy00             = iy0 - jy0;
 151             dz00             = iz0 - jz0;
 152
 153             /* Calculate squared distance and things based on it */
 154             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 155
 156             rinv00           = gmx_invsqrt(rsq00);
 157
 158             rinvsq00         = rinv00*rinv00;
 159
 160             /* Load parameters for j particles */
 161             jq0              = charge[jnr+0];
 162
 163             /**************************
 164              * CALCULATE INTERACTIONS *
 165              **************************/
 166
 167             if (rsq00<rcutoff2)
 168             {
 169
 170             r00              = rsq00*rinv00;
 171
 172             qq00             = iq0*jq0;
 173
 174             /* EWALD ELECTROSTATICS */
 175
 176             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 177             ewrt             = r00*ewtabscale;
 178             ewitab           = ewrt;
 179             eweps            = ewrt-ewitab;
 180             ewitab           = 4*ewitab;
 181             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 182             velec            = qq00*(rinv00-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 183             felec            = qq00*rinv00*(rinvsq00-felec);
 184
 185             d                = r00-rswitch;
 186             d                = (d>0.0) ? d : 0.0;
 187             d2               = d*d;
 188             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 189
 190             dsw              = d2*(swF2+d*(swF3+d*swF4));
 191
 192             /* Evaluate switch function */
 193             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 194             felec            = felec*sw - rinv00*velec*dsw;
 195             velec           *= sw;
 196
 197             /* Update potential sums from outer loop */
 198             velecsum        += velec;
 199
 200             fscal            = felec;
 201
 202             /* Calculate temporary vectorial force */
 203             tx               = fscal*dx00;
 204             ty               = fscal*dy00;
 205             tz               = fscal*dz00;
 206
 207             /* Update vectorial force */
 208             fix0            += tx;
 209             fiy0            += ty;
 210             fiz0            += tz;
 211             f[j_coord_offset+DIM*0+XX] -= tx;
 212             f[j_coord_offset+DIM*0+YY] -= ty;
 213             f[j_coord_offset+DIM*0+ZZ] -= tz;
 214
 215             }
 216
 217             /* Inner loop uses 59 flops */
 218         }
 219         /* End of innermost loop */
 220
 221         tx = ty = tz = 0;
 222         f[i_coord_offset+DIM*0+XX] += fix0;
 223         f[i_coord_offset+DIM*0+YY] += fiy0;
 224         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 225         tx                         += fix0;
 226         ty                         += fiy0;
 227         tz                         += fiz0;
 228         fshift[i_shift_offset+XX]  += tx;
 229         fshift[i_shift_offset+YY]  += ty;
 230         fshift[i_shift_offset+ZZ]  += tz;
 231
 232         ggid                        = gid[iidx];
 233         /* Update potential energies */
 234         kernel_data->energygrp_elec[ggid] += velecsum;
 235
 236         /* Increment number of inner iterations */
 237         inneriter                  += j_index_end - j_index_start;
 238
 239         /* Outer loop uses 14 flops */
 240     }
 241
 242     /* Increment number of outer iterations */
 243     outeriter        += nri;
 244
 245     /* Update outer/inner flops */
 246
 247     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VF,outeriter*14 + inneriter*59);
 248 }
 249 /*
 250  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSw_VdwNone_GeomP1P1_F_c
 251  * Electrostatics interaction: Ewald
 252  * VdW interaction:            None
 253  * Geometry:                   Particle-Particle
 254  * Calculate force/pot:        Force
 255  */
 256 void
 257 nb_kernel_ElecEwSw_VdwNone_GeomP1P1_F_c
 258                     (t_nblist * gmx_restrict                nlist,
 259                      rvec * gmx_restrict                    xx,
 260                      rvec * gmx_restrict                    ff,
 261                      t_forcerec * gmx_restrict              fr,
 262                      t_mdatoms * gmx_restrict               mdatoms,
 263                      nb_kernel_data_t * gmx_restrict        kernel_data,
 264                      t_nrnb * gmx_restrict                  nrnb)
 265 {
 266     int              i_shift_offset,i_coord_offset,j_coord_offset;
 267     int              j_index_start,j_index_end;
 268     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
 269     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
 270     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 271     real             *shiftvec,*fshift,*x,*f;
 272     int              vdwioffset0;
 273     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 274     int              vdwjidx0;
 275     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 276     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
 277     real             velec,felec,velecsum,facel,crf,krf,krf2;
 278     real             *charge;
 279     int              ewitab;
 280     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
 281     real             *ewtab;
 282     real             rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
 283
 284     x                = xx[0];
 285     f                = ff[0];
 286
 287     nri              = nlist->nri;
 288     iinr             = nlist->iinr;
 289     jindex           = nlist->jindex;
 290     jjnr             = nlist->jjnr;
 291     shiftidx         = nlist->shift;
 292     gid              = nlist->gid;
 293     shiftvec         = fr->shift_vec[0];
 294     fshift           = fr->fshift[0];
 295     facel            = fr->epsfac;
 296     charge           = mdatoms->chargeA;
 297
 298     sh_ewald         = fr->ic->sh_ewald;
 299     ewtab            = fr->ic->tabq_coul_FDV0;
 300     ewtabscale       = fr->ic->tabq_scale;
 301     ewtabhalfspace   = 0.5/ewtabscale;
 302
 303     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 304     rcutoff          = fr->rcoulomb;
 305     rcutoff2         = rcutoff*rcutoff;
 306
 307     rswitch          = fr->rcoulomb_switch;
 308     /* Setup switch parameters */
 309     d                = rcutoff-rswitch;
 310     swV3             = -10.0/(d*d*d);
 311     swV4             =  15.0/(d*d*d*d);
 312     swV5             =  -6.0/(d*d*d*d*d);
 313     swF2             = -30.0/(d*d*d);
 314     swF3             =  60.0/(d*d*d*d);
 315     swF4             = -30.0/(d*d*d*d*d);
 316
 317     outeriter        = 0;
 318     inneriter        = 0;
 319
 320     /* Start outer loop over neighborlists */
 321     for(iidx=0; iidx<nri; iidx++)
 322     {
 323         /* Load shift vector for this list */
 324         i_shift_offset   = DIM*shiftidx[iidx];
 325         shX              = shiftvec[i_shift_offset+XX];
 326         shY              = shiftvec[i_shift_offset+YY];
 327         shZ              = shiftvec[i_shift_offset+ZZ];
 328
 329         /* Load limits for loop over neighbors */
 330         j_index_start    = jindex[iidx];
 331         j_index_end      = jindex[iidx+1];
 332
 333         /* Get outer coordinate index */
 334         inr              = iinr[iidx];
 335         i_coord_offset   = DIM*inr;
 336
 337         /* Load i particle coords and add shift vector */
 338         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 339         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 340         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 341
 342         fix0             = 0.0;
 343         fiy0             = 0.0;
 344         fiz0             = 0.0;
 345
 346         /* Load parameters for i particles */
 347         iq0              = facel*charge[inr+0];
 348
 349         /* Start inner kernel loop */
 350         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 351         {
 352             /* Get j neighbor index, and coordinate index */
 353             jnr              = jjnr[jidx];
 354             j_coord_offset   = DIM*jnr;
 355
 356             /* load j atom coordinates */
 357             jx0              = x[j_coord_offset+DIM*0+XX];
 358             jy0              = x[j_coord_offset+DIM*0+YY];
 359             jz0              = x[j_coord_offset+DIM*0+ZZ];
 360
 361             /* Calculate displacement vector */
 362             dx00             = ix0 - jx0;
 363             dy00             = iy0 - jy0;
 364             dz00             = iz0 - jz0;
 365
 366             /* Calculate squared distance and things based on it */
 367             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 368
 369             rinv00           = gmx_invsqrt(rsq00);
 370
 371             rinvsq00         = rinv00*rinv00;
 372
 373             /* Load parameters for j particles */
 374             jq0              = charge[jnr+0];
 375
 376             /**************************
 377              * CALCULATE INTERACTIONS *
 378              **************************/
 379
 380             if (rsq00<rcutoff2)
 381             {
 382
 383             r00              = rsq00*rinv00;
 384
 385             qq00             = iq0*jq0;
 386
 387             /* EWALD ELECTROSTATICS */
 388
 389             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 390             ewrt             = r00*ewtabscale;
 391             ewitab           = ewrt;
 392             eweps            = ewrt-ewitab;
 393             ewitab           = 4*ewitab;
 394             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 395             velec            = qq00*(rinv00-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 396             felec            = qq00*rinv00*(rinvsq00-felec);
 397
 398             d                = r00-rswitch;
 399             d                = (d>0.0) ? d : 0.0;
 400             d2               = d*d;
 401             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 402
 403             dsw              = d2*(swF2+d*(swF3+d*swF4));
 404
 405             /* Evaluate switch function */
 406             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 407             felec            = felec*sw - rinv00*velec*dsw;
 408
 409             fscal            = felec;
 410
 411             /* Calculate temporary vectorial force */
 412             tx               = fscal*dx00;
 413             ty               = fscal*dy00;
 414             tz               = fscal*dz00;
 415
 416             /* Update vectorial force */
 417             fix0            += tx;
 418             fiy0            += ty;
 419             fiz0            += tz;
 420             f[j_coord_offset+DIM*0+XX] -= tx;
 421             f[j_coord_offset+DIM*0+YY] -= ty;
 422             f[j_coord_offset+DIM*0+ZZ] -= tz;
 423
 424             }
 425
 426             /* Inner loop uses 57 flops */
 427         }
 428         /* End of innermost loop */
 429
 430         tx = ty = tz = 0;
 431         f[i_coord_offset+DIM*0+XX] += fix0;
 432         f[i_coord_offset+DIM*0+YY] += fiy0;
 433         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 434         tx                         += fix0;
 435         ty                         += fiy0;
 436         tz                         += fiz0;
 437         fshift[i_shift_offset+XX]  += tx;
 438         fshift[i_shift_offset+YY]  += ty;
 439         fshift[i_shift_offset+ZZ]  += tz;
 440
 441         /* Increment number of inner iterations */
 442         inneriter                  += j_index_end - j_index_start;
 443
 444         /* Outer loop uses 13 flops */
 445     }
 446
 447     /* Increment number of outer iterations */
 448     outeriter        += nri;
 449
 450     /* Update outer/inner flops */
 451
 452     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_F,outeriter*13 + inneriter*57);
 453 }