src/gmxlib/nonbonded/nb_kernel_c/nb_kernel_ElecGB_VdwLJ_GeomP1P1_c.c

   1 /*
   2  * This file is part of the GROMACS molecular simulation package.
   3  *
   4  * Copyright (c) 2012, by the GROMACS development team, led by
   5  * David van der Spoel, Berk Hess, Erik Lindahl, and including many
   6  * others, as listed in the AUTHORS file in the top-level source
   7  * 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 #ifdef HAVE_CONFIG_H
  39 #include <config.h>
  40 #endif
  41
  42 #include <math.h>
  43
  44 #include "../nb_kernel.h"
  45 #include "types/simple.h"
  46 #include "vec.h"
  47 #include "nrnb.h"
  48
  49 /*
  50  * GROMACS nonbonded kernel:   nb_kernel_ElecGB_VdwLJ_GeomP1P1_VF_c
  51  * Electrostatics interaction: GeneralizedBorn
  52  * VdW interaction:            LennardJones
  53  * Geometry:                   Particle-Particle
  54  * Calculate force/pot:        PotentialAndForce
  55  */
  56 void
  57 nb_kernel_ElecGB_VdwLJ_GeomP1P1_VF_c
  58                     (t_nblist * gmx_restrict                nlist,
  59                      rvec * gmx_restrict                    xx,
  60                      rvec * gmx_restrict                    ff,
  61                      t_forcerec * gmx_restrict              fr,
  62                      t_mdatoms * gmx_restrict               mdatoms,
  63                      nb_kernel_data_t * gmx_restrict        kernel_data,
  64                      t_nrnb * gmx_restrict                  nrnb)
  65 {
  66     int              i_shift_offset,i_coord_offset,j_coord_offset;
  67     int              j_index_start,j_index_end;
  68     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
  69     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
  70     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
  71     real             *shiftvec,*fshift,*x,*f;
  72     int              vdwioffset0;
  73     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
  74     int              vdwjidx0;
  75     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
  76     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
  77     real             velec,felec,velecsum,facel,crf,krf,krf2;
  78     real             *charge;
  79     int              gbitab;
  80     real             vgb,fgb,vgbsum,dvdasum,gbscale,gbtabscale,isaprod,gbqqfactor,gbinvepsdiff,dvdaj,gbeps,dvdatmp;
  81     real             *invsqrta,*dvda,*gbtab;
  82     int              nvdwtype;
  83     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
  84     int              *vdwtype;
  85     real             *vdwparam;
  86     int              vfitab;
  87     real             rt,vfeps,vftabscale,Y,F,Geps,Heps2,Fp,VV,FF;
  88     real             *vftab;
  89
  90     x                = xx[0];
  91     f                = ff[0];
  92
  93     nri              = nlist->nri;
  94     iinr             = nlist->iinr;
  95     jindex           = nlist->jindex;
  96     jjnr             = nlist->jjnr;
  97     shiftidx         = nlist->shift;
  98     gid              = nlist->gid;
  99     shiftvec         = fr->shift_vec[0];
 100     fshift           = fr->fshift[0];
 101     facel            = fr->epsfac;
 102     charge           = mdatoms->chargeA;
 103     nvdwtype         = fr->ntype;
 104     vdwparam         = fr->nbfp;
 105     vdwtype          = mdatoms->typeA;
 106
 107     invsqrta         = fr->invsqrta;
 108     dvda             = fr->dvda;
 109     gbtabscale       = fr->gbtab.scale;
 110     gbtab            = fr->gbtab.data;
 111     gbinvepsdiff     = (1.0/fr->epsilon_r) - (1.0/fr->gb_epsilon_solvent);
 112
 113     outeriter        = 0;
 114     inneriter        = 0;
 115
 116     /* Start outer loop over neighborlists */
 117     for(iidx=0; iidx<nri; iidx++)
 118     {
 119         /* Load shift vector for this list */
 120         i_shift_offset   = DIM*shiftidx[iidx];
 121         shX              = shiftvec[i_shift_offset+XX];
 122         shY              = shiftvec[i_shift_offset+YY];
 123         shZ              = shiftvec[i_shift_offset+ZZ];
 124
 125         /* Load limits for loop over neighbors */
 126         j_index_start    = jindex[iidx];
 127         j_index_end      = jindex[iidx+1];
 128
 129         /* Get outer coordinate index */
 130         inr              = iinr[iidx];
 131         i_coord_offset   = DIM*inr;
 132
 133         /* Load i particle coords and add shift vector */
 134         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 135         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 136         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 137
 138         fix0             = 0.0;
 139         fiy0             = 0.0;
 140         fiz0             = 0.0;
 141
 142         /* Load parameters for i particles */
 143         iq0              = facel*charge[inr+0];
 144         isai0            = invsqrta[inr+0];
 145         vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 146
 147         /* Reset potential sums */
 148         velecsum         = 0.0;
 149         vgbsum           = 0.0;
 150         vvdwsum          = 0.0;
 151         dvdasum          = 0.0;
 152         printf("inr=%d\n",inr);
 153         /* Start inner kernel loop */
 154         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 155         {
 156             /* Get j neighbor index, and coordinate index */
 157             jnr              = jjnr[jidx];
 158             j_coord_offset   = DIM*jnr;
 159
 160             /* load j atom coordinates */
 161             jx0              = x[j_coord_offset+DIM*0+XX];
 162             jy0              = x[j_coord_offset+DIM*0+YY];
 163             jz0              = x[j_coord_offset+DIM*0+ZZ];
 164
 165             /* Calculate displacement vector */
 166             dx00             = ix0 - jx0;
 167             dy00             = iy0 - jy0;
 168             dz00             = iz0 - jz0;
 169
 170             /* Calculate squared distance and things based on it */
 171             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 172
 173             rinv00           = gmx_invsqrt(rsq00);
 174
 175             rinvsq00         = rinv00*rinv00;
 176
 177             /* Load parameters for j particles */
 178             jq0              = charge[jnr+0];
 179             isaj0           = invsqrta[jnr+0];
 180             vdwjidx0         = 2*vdwtype[jnr+0];
 181
 182             /**************************
 183              * CALCULATE INTERACTIONS *
 184              **************************/
 185
 186             r00              = rsq00*rinv00;
 187
 188             qq00             = iq0*jq0;
 189             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 190             c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 191
 192             /* GENERALIZED BORN AND COULOMB ELECTROSTATICS */
 193             isaprod          = isai0*isaj0;
 194             gbqqfactor       = isaprod*(-qq00)*gbinvepsdiff;
 195             gbscale          = isaprod*gbtabscale;
 196             dvdaj            = dvda[jnr+0];
 197
 198             /* Calculate generalized born table index - this is a separate table from the normal one,
 199              * but we use the same procedure by multiplying r with scale and truncating to integer.
 200              */
 201             rt               = r00*gbscale;
 202             gbitab           = rt;
 203             gbeps            = rt-gbitab;
 204             gbitab           = 4*gbitab;
 205
 206             Y                = gbtab[gbitab];
 207             F                = gbtab[gbitab+1];
 208             Geps             = gbeps*gbtab[gbitab+2];
 209             Heps2            = gbeps*gbeps*gbtab[gbitab+3];
 210             Fp               = F+Geps+Heps2;
 211             VV               = Y+gbeps*Fp;
 212             vgb              = gbqqfactor*VV;
 213
 214             FF               = Fp+Geps+2.0*Heps2;
 215             fgb              = gbqqfactor*FF*gbscale;
 216             printf("  jnr=%d  fgb=%g\n",jnr,fgb);
 217             dvdatmp          = -0.5*(vgb+fgb*r00);
 218             dvdasum          = dvdasum + dvdatmp;
 219             printf("  dvdatmp=%g\n",dvdatmp);
 220             dvda[jnr]        = dvdaj+dvdatmp*isaj0*isaj0;
 221             printf("  dvda, jcontrib=%g\n",dvdatmp*isaj0*isaj0);
 222             velec            = qq00*rinv00;
 223             felec            = (velec*rinv00-fgb)*rinv00;
 224
 225             /* LENNARD-JONES DISPERSION/REPULSION */
 226
 227             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 228             vvdw6            = c6_00*rinvsix;
 229             vvdw12           = c12_00*rinvsix*rinvsix;
 230             vvdw             = vvdw12*(1.0/12.0) - vvdw6*(1.0/6.0);
 231             fvdw             = (vvdw12-vvdw6)*rinvsq00;
 232
 233             /* Update potential sums from outer loop */
 234             velecsum        += velec;
 235             vgbsum          += vgb;
 236             vvdwsum         += vvdw;
 237
 238             fscal            = felec+fvdw;
 239
 240             /* Calculate temporary vectorial force */
 241             tx               = fscal*dx00;
 242             ty               = fscal*dy00;
 243             tz               = fscal*dz00;
 244
 245             /* Update vectorial force */
 246             fix0            += tx;
 247             fiy0            += ty;
 248             fiz0            += tz;
 249             f[j_coord_offset+DIM*0+XX] -= tx;
 250             f[j_coord_offset+DIM*0+YY] -= ty;
 251             f[j_coord_offset+DIM*0+ZZ] -= tz;
 252
 253             /* Inner loop uses 71 flops */
 254         }
 255         /* End of innermost loop */
 256
 257         tx = ty = tz = 0;
 258         f[i_coord_offset+DIM*0+XX] += fix0;
 259         f[i_coord_offset+DIM*0+YY] += fiy0;
 260         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 261         tx                         += fix0;
 262         ty                         += fiy0;
 263         tz                         += fiz0;
 264         fshift[i_shift_offset+XX]  += tx;
 265         fshift[i_shift_offset+YY]  += ty;
 266         fshift[i_shift_offset+ZZ]  += tz;
 267
 268         ggid                        = gid[iidx];
 269         /* Update potential energies */
 270         kernel_data->energygrp_elec[ggid] += velecsum;
 271         kernel_data->energygrp_polarization[ggid] += vgbsum;
 272         kernel_data->energygrp_vdw[ggid] += vvdwsum;
 273         dvda[inr]                   = dvda[inr] + dvdasum*isai0*isai0;
 274
 275         /* Increment number of inner iterations */
 276         inneriter                  += j_index_end - j_index_start;
 277
 278         /* Outer loop uses 16 flops */
 279     }
 280
 281     /* Increment number of outer iterations */
 282     outeriter        += nri;
 283
 284     /* Update outer/inner flops */
 285
 286     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_VF,outeriter*16 + inneriter*71);
 287 }
 288 /*
 289  * GROMACS nonbonded kernel:   nb_kernel_ElecGB_VdwLJ_GeomP1P1_F_c
 290  * Electrostatics interaction: GeneralizedBorn
 291  * VdW interaction:            LennardJones
 292  * Geometry:                   Particle-Particle
 293  * Calculate force/pot:        Force
 294  */
 295 void
 296 nb_kernel_ElecGB_VdwLJ_GeomP1P1_F_c
 297                     (t_nblist * gmx_restrict                nlist,
 298                      rvec * gmx_restrict                    xx,
 299                      rvec * gmx_restrict                    ff,
 300                      t_forcerec * gmx_restrict              fr,
 301                      t_mdatoms * gmx_restrict               mdatoms,
 302                      nb_kernel_data_t * gmx_restrict        kernel_data,
 303                      t_nrnb * gmx_restrict                  nrnb)
 304 {
 305     int              i_shift_offset,i_coord_offset,j_coord_offset;
 306     int              j_index_start,j_index_end;
 307     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
 308     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
 309     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 310     real             *shiftvec,*fshift,*x,*f;
 311     int              vdwioffset0;
 312     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 313     int              vdwjidx0;
 314     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 315     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
 316     real             velec,felec,velecsum,facel,crf,krf,krf2;
 317     real             *charge;
 318     int              gbitab;
 319     real             vgb,fgb,vgbsum,dvdasum,gbscale,gbtabscale,isaprod,gbqqfactor,gbinvepsdiff,dvdaj,gbeps,dvdatmp;
 320     real             *invsqrta,*dvda,*gbtab;
 321     int              nvdwtype;
 322     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
 323     int              *vdwtype;
 324     real             *vdwparam;
 325     int              vfitab;
 326     real             rt,vfeps,vftabscale,Y,F,Geps,Heps2,Fp,VV,FF;
 327     real             *vftab;
 328
 329     x                = xx[0];
 330     f                = ff[0];
 331
 332     nri              = nlist->nri;
 333     iinr             = nlist->iinr;
 334     jindex           = nlist->jindex;
 335     jjnr             = nlist->jjnr;
 336     shiftidx         = nlist->shift;
 337     gid              = nlist->gid;
 338     shiftvec         = fr->shift_vec[0];
 339     fshift           = fr->fshift[0];
 340     facel            = fr->epsfac;
 341     charge           = mdatoms->chargeA;
 342     nvdwtype         = fr->ntype;
 343     vdwparam         = fr->nbfp;
 344     vdwtype          = mdatoms->typeA;
 345
 346     invsqrta         = fr->invsqrta;
 347     dvda             = fr->dvda;
 348     gbtabscale       = fr->gbtab.scale;
 349     gbtab            = fr->gbtab.data;
 350     gbinvepsdiff     = (1.0/fr->epsilon_r) - (1.0/fr->gb_epsilon_solvent);
 351
 352     outeriter        = 0;
 353     inneriter        = 0;
 354
 355     /* Start outer loop over neighborlists */
 356     for(iidx=0; iidx<nri; iidx++)
 357     {
 358         /* Load shift vector for this list */
 359         i_shift_offset   = DIM*shiftidx[iidx];
 360         shX              = shiftvec[i_shift_offset+XX];
 361         shY              = shiftvec[i_shift_offset+YY];
 362         shZ              = shiftvec[i_shift_offset+ZZ];
 363
 364         /* Load limits for loop over neighbors */
 365         j_index_start    = jindex[iidx];
 366         j_index_end      = jindex[iidx+1];
 367
 368         /* Get outer coordinate index */
 369         inr              = iinr[iidx];
 370         i_coord_offset   = DIM*inr;
 371
 372         /* Load i particle coords and add shift vector */
 373         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 374         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 375         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 376
 377         fix0             = 0.0;
 378         fiy0             = 0.0;
 379         fiz0             = 0.0;
 380
 381         /* Load parameters for i particles */
 382         iq0              = facel*charge[inr+0];
 383         isai0            = invsqrta[inr+0];
 384         vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 385
 386         dvdasum          = 0.0;
 387
 388         /* Start inner kernel loop */
 389         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 390         {
 391             /* Get j neighbor index, and coordinate index */
 392             jnr              = jjnr[jidx];
 393             j_coord_offset   = DIM*jnr;
 394
 395             /* load j atom coordinates */
 396             jx0              = x[j_coord_offset+DIM*0+XX];
 397             jy0              = x[j_coord_offset+DIM*0+YY];
 398             jz0              = x[j_coord_offset+DIM*0+ZZ];
 399
 400             /* Calculate displacement vector */
 401             dx00             = ix0 - jx0;
 402             dy00             = iy0 - jy0;
 403             dz00             = iz0 - jz0;
 404
 405             /* Calculate squared distance and things based on it */
 406             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 407
 408             rinv00           = gmx_invsqrt(rsq00);
 409
 410             rinvsq00         = rinv00*rinv00;
 411
 412             /* Load parameters for j particles */
 413             jq0              = charge[jnr+0];
 414             isaj0           = invsqrta[jnr+0];
 415             vdwjidx0         = 2*vdwtype[jnr+0];
 416
 417             /**************************
 418              * CALCULATE INTERACTIONS *
 419              **************************/
 420
 421             r00              = rsq00*rinv00;
 422
 423             qq00             = iq0*jq0;
 424             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 425             c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 426
 427             /* GENERALIZED BORN AND COULOMB ELECTROSTATICS */
 428             isaprod          = isai0*isaj0;
 429             gbqqfactor       = isaprod*(-qq00)*gbinvepsdiff;
 430             gbscale          = isaprod*gbtabscale;
 431             dvdaj            = dvda[jnr+0];
 432
 433             /* Calculate generalized born table index - this is a separate table from the normal one,
 434              * but we use the same procedure by multiplying r with scale and truncating to integer.
 435              */
 436             rt               = r00*gbscale;
 437             gbitab           = rt;
 438             gbeps            = rt-gbitab;
 439             gbitab           = 4*gbitab;
 440
 441             Y                = gbtab[gbitab];
 442             F                = gbtab[gbitab+1];
 443             Geps             = gbeps*gbtab[gbitab+2];
 444             Heps2            = gbeps*gbeps*gbtab[gbitab+3];
 445             Fp               = F+Geps+Heps2;
 446             VV               = Y+gbeps*Fp;
 447             vgb              = gbqqfactor*VV;
 448
 449             FF               = Fp+Geps+2.0*Heps2;
 450             fgb              = gbqqfactor*FF*gbscale;
 451             dvdatmp          = -0.5*(vgb+fgb*r00);
 452             dvdasum          = dvdasum + dvdatmp;
 453             dvda[jnr]        = dvdaj+dvdatmp*isaj0*isaj0;
 454             velec            = qq00*rinv00;
 455             felec            = (velec*rinv00-fgb)*rinv00;
 456
 457             /* LENNARD-JONES DISPERSION/REPULSION */
 458
 459             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 460             fvdw             = (c12_00*rinvsix-c6_00)*rinvsix*rinvsq00;
 461
 462             fscal            = felec+fvdw;
 463
 464             /* Calculate temporary vectorial force */
 465             tx               = fscal*dx00;
 466             ty               = fscal*dy00;
 467             tz               = fscal*dz00;
 468
 469             /* Update vectorial force */
 470             fix0            += tx;
 471             fiy0            += ty;
 472             fiz0            += tz;
 473             f[j_coord_offset+DIM*0+XX] -= tx;
 474             f[j_coord_offset+DIM*0+YY] -= ty;
 475             f[j_coord_offset+DIM*0+ZZ] -= tz;
 476
 477             /* Inner loop uses 64 flops */
 478         }
 479         /* End of innermost loop */
 480
 481         tx = ty = tz = 0;
 482         f[i_coord_offset+DIM*0+XX] += fix0;
 483         f[i_coord_offset+DIM*0+YY] += fiy0;
 484         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 485         tx                         += fix0;
 486         ty                         += fiy0;
 487         tz                         += fiz0;
 488         fshift[i_shift_offset+XX]  += tx;
 489         fshift[i_shift_offset+YY]  += ty;
 490         fshift[i_shift_offset+ZZ]  += tz;
 491
 492         dvda[inr]                   = dvda[inr] + dvdasum*isai0*isai0;
 493
 494         /* Increment number of inner iterations */
 495         inneriter                  += j_index_end - j_index_start;
 496
 497         /* Outer loop uses 13 flops */
 498     }
 499
 500     /* Increment number of outer iterations */
 501     outeriter        += nri;
 502
 503     /* Update outer/inner flops */
 504
 505     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_F,outeriter*13 + inneriter*64);
 506 }