src/gromacs/gmxlib/nonbonded/nb_kernel_c/nb_kernel_ElecGB_VdwLJ_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_ElecGB_VdwLJ_GeomP1P1_VF_c
  35  * Electrostatics interaction: GeneralizedBorn
  36  * VdW interaction:            LennardJones
  37  * Geometry:                   Particle-Particle
  38  * Calculate force/pot:        PotentialAndForce
  39  */
  40 void
  41 nb_kernel_ElecGB_VdwLJ_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              gbitab;
  64     real             vgb,fgb,vgbsum,dvdasum,gbscale,gbtabscale,isaprod,gbqqfactor,gbinvepsdiff,dvdaj,gbeps,dvdatmp;
  65     real             *invsqrta,*dvda,*gbtab;
  66     int              nvdwtype;
  67     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
  68     int              *vdwtype;
  69     real             *vdwparam;
  70     int              vfitab;
  71     real             rt,vfeps,vftabscale,Y,F,Geps,Heps2,Fp,VV,FF;
  72     real             *vftab;
  73
  74     x                = xx[0];
  75     f                = ff[0];
  76
  77     nri              = nlist->nri;
  78     iinr             = nlist->iinr;
  79     jindex           = nlist->jindex;
  80     jjnr             = nlist->jjnr;
  81     shiftidx         = nlist->shift;
  82     gid              = nlist->gid;
  83     shiftvec         = fr->shift_vec[0];
  84     fshift           = fr->fshift[0];
  85     facel            = fr->epsfac;
  86     charge           = mdatoms->chargeA;
  87     nvdwtype         = fr->ntype;
  88     vdwparam         = fr->nbfp;
  89     vdwtype          = mdatoms->typeA;
  90
  91     invsqrta         = fr->invsqrta;
  92     dvda             = fr->dvda;
  93     gbtabscale       = fr->gbtab.scale;
  94     gbtab            = fr->gbtab.data;
  95     gbinvepsdiff     = (1.0/fr->epsilon_r) - (1.0/fr->gb_epsilon_solvent);
  96
  97     outeriter        = 0;
  98     inneriter        = 0;
  99
 100     /* Start outer loop over neighborlists */
 101     for(iidx=0; iidx<nri; iidx++)
 102     {
 103         /* Load shift vector for this list */
 104         i_shift_offset   = DIM*shiftidx[iidx];
 105         shX              = shiftvec[i_shift_offset+XX];
 106         shY              = shiftvec[i_shift_offset+YY];
 107         shZ              = shiftvec[i_shift_offset+ZZ];
 108
 109         /* Load limits for loop over neighbors */
 110         j_index_start    = jindex[iidx];
 111         j_index_end      = jindex[iidx+1];
 112
 113         /* Get outer coordinate index */
 114         inr              = iinr[iidx];
 115         i_coord_offset   = DIM*inr;
 116
 117         /* Load i particle coords and add shift vector */
 118         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 119         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 120         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 121
 122         fix0             = 0.0;
 123         fiy0             = 0.0;
 124         fiz0             = 0.0;
 125
 126         /* Load parameters for i particles */
 127         iq0              = facel*charge[inr+0];
 128         isai0            = invsqrta[inr+0];
 129         vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 130
 131         /* Reset potential sums */
 132         velecsum         = 0.0;
 133         vgbsum           = 0.0;
 134         vvdwsum          = 0.0;
 135         dvdasum          = 0.0;
 136         printf("inr=%d\n",inr);
 137         /* Start inner kernel loop */
 138         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 139         {
 140             /* Get j neighbor index, and coordinate index */
 141             jnr              = jjnr[jidx];
 142             j_coord_offset   = DIM*jnr;
 143
 144             /* load j atom coordinates */
 145             jx0              = x[j_coord_offset+DIM*0+XX];
 146             jy0              = x[j_coord_offset+DIM*0+YY];
 147             jz0              = x[j_coord_offset+DIM*0+ZZ];
 148
 149             /* Calculate displacement vector */
 150             dx00             = ix0 - jx0;
 151             dy00             = iy0 - jy0;
 152             dz00             = iz0 - jz0;
 153
 154             /* Calculate squared distance and things based on it */
 155             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 156
 157             rinv00           = gmx_invsqrt(rsq00);
 158
 159             rinvsq00         = rinv00*rinv00;
 160
 161             /* Load parameters for j particles */
 162             jq0              = charge[jnr+0];
 163             isaj0           = invsqrta[jnr+0];
 164             vdwjidx0         = 2*vdwtype[jnr+0];
 165
 166             /**************************
 167              * CALCULATE INTERACTIONS *
 168              **************************/
 169
 170             r00              = rsq00*rinv00;
 171
 172             qq00             = iq0*jq0;
 173             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 174             c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 175
 176             /* GENERALIZED BORN AND COULOMB ELECTROSTATICS */
 177             isaprod          = isai0*isaj0;
 178             gbqqfactor       = isaprod*(-qq00)*gbinvepsdiff;
 179             gbscale          = isaprod*gbtabscale;
 180             dvdaj            = dvda[jnr+0];
 181
 182             /* Calculate generalized born table index - this is a separate table from the normal one,
 183              * but we use the same procedure by multiplying r with scale and truncating to integer.
 184              */
 185             rt               = r00*gbscale;
 186             gbitab           = rt;
 187             gbeps            = rt-gbitab;
 188             gbitab           = 4*gbitab;
 189
 190             Y                = gbtab[gbitab];
 191             F                = gbtab[gbitab+1];
 192             Geps             = gbeps*gbtab[gbitab+2];
 193             Heps2            = gbeps*gbeps*gbtab[gbitab+3];
 194             Fp               = F+Geps+Heps2;
 195             VV               = Y+gbeps*Fp;
 196             vgb              = gbqqfactor*VV;
 197
 198             FF               = Fp+Geps+2.0*Heps2;
 199             fgb              = gbqqfactor*FF*gbscale;
 200             printf("  jnr=%d  fgb=%g\n",jnr,fgb);
 201             dvdatmp          = -0.5*(vgb+fgb*r00);
 202             dvdasum          = dvdasum + dvdatmp;
 203             printf("  dvdatmp=%g\n",dvdatmp);
 204             dvda[jnr]        = dvdaj+dvdatmp*isaj0*isaj0;
 205             printf("  dvda, jcontrib=%g\n",dvdatmp*isaj0*isaj0);
 206             velec            = qq00*rinv00;
 207             felec            = (velec*rinv00-fgb)*rinv00;
 208
 209             /* LENNARD-JONES DISPERSION/REPULSION */
 210
 211             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 212             vvdw6            = c6_00*rinvsix;
 213             vvdw12           = c12_00*rinvsix*rinvsix;
 214             vvdw             = vvdw12*(1.0/12.0) - vvdw6*(1.0/6.0);
 215             fvdw             = (vvdw12-vvdw6)*rinvsq00;
 216
 217             /* Update potential sums from outer loop */
 218             velecsum        += velec;
 219             vgbsum          += vgb;
 220             vvdwsum         += vvdw;
 221
 222             fscal            = felec+fvdw;
 223
 224             /* Calculate temporary vectorial force */
 225             tx               = fscal*dx00;
 226             ty               = fscal*dy00;
 227             tz               = fscal*dz00;
 228
 229             /* Update vectorial force */
 230             fix0            += tx;
 231             fiy0            += ty;
 232             fiz0            += tz;
 233             f[j_coord_offset+DIM*0+XX] -= tx;
 234             f[j_coord_offset+DIM*0+YY] -= ty;
 235             f[j_coord_offset+DIM*0+ZZ] -= tz;
 236
 237             /* Inner loop uses 71 flops */
 238         }
 239         /* End of innermost loop */
 240
 241         tx = ty = tz = 0;
 242         f[i_coord_offset+DIM*0+XX] += fix0;
 243         f[i_coord_offset+DIM*0+YY] += fiy0;
 244         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 245         tx                         += fix0;
 246         ty                         += fiy0;
 247         tz                         += fiz0;
 248         fshift[i_shift_offset+XX]  += tx;
 249         fshift[i_shift_offset+YY]  += ty;
 250         fshift[i_shift_offset+ZZ]  += tz;
 251
 252         ggid                        = gid[iidx];
 253         /* Update potential energies */
 254         kernel_data->energygrp_elec[ggid] += velecsum;
 255         kernel_data->energygrp_polarization[ggid] += vgbsum;
 256         kernel_data->energygrp_vdw[ggid] += vvdwsum;
 257         dvda[inr]                   = dvda[inr] + dvdasum*isai0*isai0;
 258
 259         /* Increment number of inner iterations */
 260         inneriter                  += j_index_end - j_index_start;
 261
 262         /* Outer loop uses 16 flops */
 263     }
 264
 265     /* Increment number of outer iterations */
 266     outeriter        += nri;
 267
 268     /* Update outer/inner flops */
 269
 270     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_VF,outeriter*16 + inneriter*71);
 271 }
 272 /*
 273  * Gromacs nonbonded kernel:   nb_kernel_ElecGB_VdwLJ_GeomP1P1_F_c
 274  * Electrostatics interaction: GeneralizedBorn
 275  * VdW interaction:            LennardJones
 276  * Geometry:                   Particle-Particle
 277  * Calculate force/pot:        Force
 278  */
 279 void
 280 nb_kernel_ElecGB_VdwLJ_GeomP1P1_F_c
 281                     (t_nblist * gmx_restrict                nlist,
 282                      rvec * gmx_restrict                    xx,
 283                      rvec * gmx_restrict                    ff,
 284                      t_forcerec * gmx_restrict              fr,
 285                      t_mdatoms * gmx_restrict               mdatoms,
 286                      nb_kernel_data_t * gmx_restrict        kernel_data,
 287                      t_nrnb * gmx_restrict                  nrnb)
 288 {
 289     int              i_shift_offset,i_coord_offset,j_coord_offset;
 290     int              j_index_start,j_index_end;
 291     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
 292     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
 293     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 294     real             *shiftvec,*fshift,*x,*f;
 295     int              vdwioffset0;
 296     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 297     int              vdwjidx0;
 298     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 299     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
 300     real             velec,felec,velecsum,facel,crf,krf,krf2;
 301     real             *charge;
 302     int              gbitab;
 303     real             vgb,fgb,vgbsum,dvdasum,gbscale,gbtabscale,isaprod,gbqqfactor,gbinvepsdiff,dvdaj,gbeps,dvdatmp;
 304     real             *invsqrta,*dvda,*gbtab;
 305     int              nvdwtype;
 306     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
 307     int              *vdwtype;
 308     real             *vdwparam;
 309     int              vfitab;
 310     real             rt,vfeps,vftabscale,Y,F,Geps,Heps2,Fp,VV,FF;
 311     real             *vftab;
 312
 313     x                = xx[0];
 314     f                = ff[0];
 315
 316     nri              = nlist->nri;
 317     iinr             = nlist->iinr;
 318     jindex           = nlist->jindex;
 319     jjnr             = nlist->jjnr;
 320     shiftidx         = nlist->shift;
 321     gid              = nlist->gid;
 322     shiftvec         = fr->shift_vec[0];
 323     fshift           = fr->fshift[0];
 324     facel            = fr->epsfac;
 325     charge           = mdatoms->chargeA;
 326     nvdwtype         = fr->ntype;
 327     vdwparam         = fr->nbfp;
 328     vdwtype          = mdatoms->typeA;
 329
 330     invsqrta         = fr->invsqrta;
 331     dvda             = fr->dvda;
 332     gbtabscale       = fr->gbtab.scale;
 333     gbtab            = fr->gbtab.data;
 334     gbinvepsdiff     = (1.0/fr->epsilon_r) - (1.0/fr->gb_epsilon_solvent);
 335
 336     outeriter        = 0;
 337     inneriter        = 0;
 338
 339     /* Start outer loop over neighborlists */
 340     for(iidx=0; iidx<nri; iidx++)
 341     {
 342         /* Load shift vector for this list */
 343         i_shift_offset   = DIM*shiftidx[iidx];
 344         shX              = shiftvec[i_shift_offset+XX];
 345         shY              = shiftvec[i_shift_offset+YY];
 346         shZ              = shiftvec[i_shift_offset+ZZ];
 347
 348         /* Load limits for loop over neighbors */
 349         j_index_start    = jindex[iidx];
 350         j_index_end      = jindex[iidx+1];
 351
 352         /* Get outer coordinate index */
 353         inr              = iinr[iidx];
 354         i_coord_offset   = DIM*inr;
 355
 356         /* Load i particle coords and add shift vector */
 357         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 358         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 359         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 360
 361         fix0             = 0.0;
 362         fiy0             = 0.0;
 363         fiz0             = 0.0;
 364
 365         /* Load parameters for i particles */
 366         iq0              = facel*charge[inr+0];
 367         isai0            = invsqrta[inr+0];
 368         vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 369
 370         dvdasum          = 0.0;
 371
 372         /* Start inner kernel loop */
 373         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 374         {
 375             /* Get j neighbor index, and coordinate index */
 376             jnr              = jjnr[jidx];
 377             j_coord_offset   = DIM*jnr;
 378
 379             /* load j atom coordinates */
 380             jx0              = x[j_coord_offset+DIM*0+XX];
 381             jy0              = x[j_coord_offset+DIM*0+YY];
 382             jz0              = x[j_coord_offset+DIM*0+ZZ];
 383
 384             /* Calculate displacement vector */
 385             dx00             = ix0 - jx0;
 386             dy00             = iy0 - jy0;
 387             dz00             = iz0 - jz0;
 388
 389             /* Calculate squared distance and things based on it */
 390             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 391
 392             rinv00           = gmx_invsqrt(rsq00);
 393
 394             rinvsq00         = rinv00*rinv00;
 395
 396             /* Load parameters for j particles */
 397             jq0              = charge[jnr+0];
 398             isaj0           = invsqrta[jnr+0];
 399             vdwjidx0         = 2*vdwtype[jnr+0];
 400
 401             /**************************
 402              * CALCULATE INTERACTIONS *
 403              **************************/
 404
 405             r00              = rsq00*rinv00;
 406
 407             qq00             = iq0*jq0;
 408             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 409             c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 410
 411             /* GENERALIZED BORN AND COULOMB ELECTROSTATICS */
 412             isaprod          = isai0*isaj0;
 413             gbqqfactor       = isaprod*(-qq00)*gbinvepsdiff;
 414             gbscale          = isaprod*gbtabscale;
 415             dvdaj            = dvda[jnr+0];
 416
 417             /* Calculate generalized born table index - this is a separate table from the normal one,
 418              * but we use the same procedure by multiplying r with scale and truncating to integer.
 419              */
 420             rt               = r00*gbscale;
 421             gbitab           = rt;
 422             gbeps            = rt-gbitab;
 423             gbitab           = 4*gbitab;
 424
 425             Y                = gbtab[gbitab];
 426             F                = gbtab[gbitab+1];
 427             Geps             = gbeps*gbtab[gbitab+2];
 428             Heps2            = gbeps*gbeps*gbtab[gbitab+3];
 429             Fp               = F+Geps+Heps2;
 430             VV               = Y+gbeps*Fp;
 431             vgb              = gbqqfactor*VV;
 432
 433             FF               = Fp+Geps+2.0*Heps2;
 434             fgb              = gbqqfactor*FF*gbscale;
 435             dvdatmp          = -0.5*(vgb+fgb*r00);
 436             dvdasum          = dvdasum + dvdatmp;
 437             dvda[jnr]        = dvdaj+dvdatmp*isaj0*isaj0;
 438             velec            = qq00*rinv00;
 439             felec            = (velec*rinv00-fgb)*rinv00;
 440
 441             /* LENNARD-JONES DISPERSION/REPULSION */
 442
 443             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 444             fvdw             = (c12_00*rinvsix-c6_00)*rinvsix*rinvsq00;
 445
 446             fscal            = felec+fvdw;
 447
 448             /* Calculate temporary vectorial force */
 449             tx               = fscal*dx00;
 450             ty               = fscal*dy00;
 451             tz               = fscal*dz00;
 452
 453             /* Update vectorial force */
 454             fix0            += tx;
 455             fiy0            += ty;
 456             fiz0            += tz;
 457             f[j_coord_offset+DIM*0+XX] -= tx;
 458             f[j_coord_offset+DIM*0+YY] -= ty;
 459             f[j_coord_offset+DIM*0+ZZ] -= tz;
 460
 461             /* Inner loop uses 64 flops */
 462         }
 463         /* End of innermost loop */
 464
 465         tx = ty = tz = 0;
 466         f[i_coord_offset+DIM*0+XX] += fix0;
 467         f[i_coord_offset+DIM*0+YY] += fiy0;
 468         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 469         tx                         += fix0;
 470         ty                         += fiy0;
 471         tz                         += fiz0;
 472         fshift[i_shift_offset+XX]  += tx;
 473         fshift[i_shift_offset+YY]  += ty;
 474         fshift[i_shift_offset+ZZ]  += tz;
 475
 476         dvda[inr]                   = dvda[inr] + dvdasum*isai0*isai0;
 477
 478         /* Increment number of inner iterations */
 479         inneriter                  += j_index_end - j_index_start;
 480
 481         /* Outer loop uses 13 flops */
 482     }
 483
 484     /* Increment number of outer iterations */
 485     outeriter        += nri;
 486
 487     /* Update outer/inner flops */
 488
 489     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_F,outeriter*13 + inneriter*64);
 490 }