src/gmxlib/nonbonded/nb_kernel_c/nb_kernel_ElecGB_VdwBham_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_VdwBham_GeomP1P1_VF_c
  35  * Electrostatics interaction: GeneralizedBorn
  36  * VdW interaction:            Buckingham
  37  * Geometry:                   Particle-Particle
  38  * Calculate force/pot:        PotentialAndForce
  39  */
  40 void
  41 nb_kernel_ElecGB_VdwBham_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      = 3*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
 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         = 3*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             cexp1_00         = vdwparam[vdwioffset0+vdwjidx0+1];
 175             cexp2_00         = vdwparam[vdwioffset0+vdwjidx0+2];
 176
 177             /* GENERALIZED BORN AND COULOMB ELECTROSTATICS */
 178             isaprod          = isai0*isaj0;
 179             gbqqfactor       = isaprod*(-qq00)*gbinvepsdiff;
 180             gbscale          = isaprod*gbtabscale;
 181             dvdaj            = dvda[jnr+0];
 182
 183             /* Calculate generalized born table index - this is a separate table from the normal one,
 184              * but we use the same procedure by multiplying r with scale and truncating to integer.
 185              */
 186             rt               = r00*gbscale;
 187             gbitab           = rt;
 188             gbeps            = rt-gbitab;
 189             gbitab           = 4*gbitab;
 190
 191             Y                = gbtab[gbitab];
 192             F                = gbtab[gbitab+1];
 193             Geps             = gbeps*gbtab[gbitab+2];
 194             Heps2            = gbeps*gbeps*gbtab[gbitab+3];
 195             Fp               = F+Geps+Heps2;
 196             VV               = Y+gbeps*Fp;
 197             vgb              = gbqqfactor*VV;
 198
 199             FF               = Fp+Geps+2.0*Heps2;
 200             fgb              = gbqqfactor*FF*gbscale;
 201             dvdatmp          = -0.5*(vgb+fgb*r00);
 202             dvdasum          = dvdasum + dvdatmp;
 203             dvda[jnr]        = dvdaj+dvdatmp*isaj0*isaj0;
 204             velec            = qq00*rinv00;
 205             felec            = (velec*rinv00-fgb)*rinv00;
 206
 207             /* BUCKINGHAM DISPERSION/REPULSION */
 208             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 209             vvdw6            = c6_00*rinvsix;
 210             br               = cexp2_00*r00;
 211             vvdwexp          = cexp1_00*exp(-br);
 212             vvdw             = vvdwexp - vvdw6*(1.0/6.0);
 213             fvdw             = (br*vvdwexp-vvdw6)*rinvsq00;
 214
 215             /* Update potential sums from outer loop */
 216             velecsum        += velec;
 217             vgbsum          += vgb;
 218             vvdwsum         += vvdw;
 219
 220             fscal            = felec+fvdw;
 221
 222             /* Calculate temporary vectorial force */
 223             tx               = fscal*dx00;
 224             ty               = fscal*dy00;
 225             tz               = fscal*dz00;
 226
 227             /* Update vectorial force */
 228             fix0            += tx;
 229             fiy0            += ty;
 230             fiz0            += tz;
 231             f[j_coord_offset+DIM*0+XX] -= tx;
 232             f[j_coord_offset+DIM*0+YY] -= ty;
 233             f[j_coord_offset+DIM*0+ZZ] -= tz;
 234
 235             /* Inner loop uses 97 flops */
 236         }
 237         /* End of innermost loop */
 238
 239         tx = ty = tz = 0;
 240         f[i_coord_offset+DIM*0+XX] += fix0;
 241         f[i_coord_offset+DIM*0+YY] += fiy0;
 242         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 243         tx                         += fix0;
 244         ty                         += fiy0;
 245         tz                         += fiz0;
 246         fshift[i_shift_offset+XX]  += tx;
 247         fshift[i_shift_offset+YY]  += ty;
 248         fshift[i_shift_offset+ZZ]  += tz;
 249
 250         ggid                        = gid[iidx];
 251         /* Update potential energies */
 252         kernel_data->energygrp_elec[ggid] += velecsum;
 253         kernel_data->energygrp_polarization[ggid] += vgbsum;
 254         kernel_data->energygrp_vdw[ggid] += vvdwsum;
 255         dvda[inr]                   = dvda[inr] + dvdasum*isai0*isai0;
 256
 257         /* Increment number of inner iterations */
 258         inneriter                  += j_index_end - j_index_start;
 259
 260         /* Outer loop uses 16 flops */
 261     }
 262
 263     /* Increment number of outer iterations */
 264     outeriter        += nri;
 265
 266     /* Update outer/inner flops */
 267
 268     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_VF,outeriter*16 + inneriter*97);
 269 }
 270 /*
 271  * Gromacs nonbonded kernel:   nb_kernel_ElecGB_VdwBham_GeomP1P1_F_c
 272  * Electrostatics interaction: GeneralizedBorn
 273  * VdW interaction:            Buckingham
 274  * Geometry:                   Particle-Particle
 275  * Calculate force/pot:        Force
 276  */
 277 void
 278 nb_kernel_ElecGB_VdwBham_GeomP1P1_F_c
 279                     (t_nblist * gmx_restrict                nlist,
 280                      rvec * gmx_restrict                    xx,
 281                      rvec * gmx_restrict                    ff,
 282                      t_forcerec * gmx_restrict              fr,
 283                      t_mdatoms * gmx_restrict               mdatoms,
 284                      nb_kernel_data_t * gmx_restrict        kernel_data,
 285                      t_nrnb * gmx_restrict                  nrnb)
 286 {
 287     int              i_shift_offset,i_coord_offset,j_coord_offset;
 288     int              j_index_start,j_index_end;
 289     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
 290     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
 291     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 292     real             *shiftvec,*fshift,*x,*f;
 293     int              vdwioffset0;
 294     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 295     int              vdwjidx0;
 296     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 297     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
 298     real             velec,felec,velecsum,facel,crf,krf,krf2;
 299     real             *charge;
 300     int              gbitab;
 301     real             vgb,fgb,vgbsum,dvdasum,gbscale,gbtabscale,isaprod,gbqqfactor,gbinvepsdiff,dvdaj,gbeps,dvdatmp;
 302     real             *invsqrta,*dvda,*gbtab;
 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              vfitab;
 308     real             rt,vfeps,vftabscale,Y,F,Geps,Heps2,Fp,VV,FF;
 309     real             *vftab;
 310
 311     x                = xx[0];
 312     f                = ff[0];
 313
 314     nri              = nlist->nri;
 315     iinr             = nlist->iinr;
 316     jindex           = nlist->jindex;
 317     jjnr             = nlist->jjnr;
 318     shiftidx         = nlist->shift;
 319     gid              = nlist->gid;
 320     shiftvec         = fr->shift_vec[0];
 321     fshift           = fr->fshift[0];
 322     facel            = fr->epsfac;
 323     charge           = mdatoms->chargeA;
 324     nvdwtype         = fr->ntype;
 325     vdwparam         = fr->nbfp;
 326     vdwtype          = mdatoms->typeA;
 327
 328     invsqrta         = fr->invsqrta;
 329     dvda             = fr->dvda;
 330     gbtabscale       = fr->gbtab.scale;
 331     gbtab            = fr->gbtab.data;
 332     gbinvepsdiff     = (1.0/fr->epsilon_r) - (1.0/fr->gb_epsilon_solvent);
 333
 334     outeriter        = 0;
 335     inneriter        = 0;
 336
 337     /* Start outer loop over neighborlists */
 338     for(iidx=0; iidx<nri; iidx++)
 339     {
 340         /* Load shift vector for this list */
 341         i_shift_offset   = DIM*shiftidx[iidx];
 342         shX              = shiftvec[i_shift_offset+XX];
 343         shY              = shiftvec[i_shift_offset+YY];
 344         shZ              = shiftvec[i_shift_offset+ZZ];
 345
 346         /* Load limits for loop over neighbors */
 347         j_index_start    = jindex[iidx];
 348         j_index_end      = jindex[iidx+1];
 349
 350         /* Get outer coordinate index */
 351         inr              = iinr[iidx];
 352         i_coord_offset   = DIM*inr;
 353
 354         /* Load i particle coords and add shift vector */
 355         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 356         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 357         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 358
 359         fix0             = 0.0;
 360         fiy0             = 0.0;
 361         fiz0             = 0.0;
 362
 363         /* Load parameters for i particles */
 364         iq0              = facel*charge[inr+0];
 365         isai0            = invsqrta[inr+0];
 366         vdwioffset0      = 3*nvdwtype*vdwtype[inr+0];
 367
 368         dvdasum          = 0.0;
 369
 370         /* Start inner kernel loop */
 371         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 372         {
 373             /* Get j neighbor index, and coordinate index */
 374             jnr              = jjnr[jidx];
 375             j_coord_offset   = DIM*jnr;
 376
 377             /* load j atom coordinates */
 378             jx0              = x[j_coord_offset+DIM*0+XX];
 379             jy0              = x[j_coord_offset+DIM*0+YY];
 380             jz0              = x[j_coord_offset+DIM*0+ZZ];
 381
 382             /* Calculate displacement vector */
 383             dx00             = ix0 - jx0;
 384             dy00             = iy0 - jy0;
 385             dz00             = iz0 - jz0;
 386
 387             /* Calculate squared distance and things based on it */
 388             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 389
 390             rinv00           = gmx_invsqrt(rsq00);
 391
 392             rinvsq00         = rinv00*rinv00;
 393
 394             /* Load parameters for j particles */
 395             jq0              = charge[jnr+0];
 396             isaj0           = invsqrta[jnr+0];
 397             vdwjidx0         = 3*vdwtype[jnr+0];
 398
 399             /**************************
 400              * CALCULATE INTERACTIONS *
 401              **************************/
 402
 403             r00              = rsq00*rinv00;
 404
 405             qq00             = iq0*jq0;
 406             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 407             cexp1_00         = vdwparam[vdwioffset0+vdwjidx0+1];
 408             cexp2_00         = vdwparam[vdwioffset0+vdwjidx0+2];
 409
 410             /* GENERALIZED BORN AND COULOMB ELECTROSTATICS */
 411             isaprod          = isai0*isaj0;
 412             gbqqfactor       = isaprod*(-qq00)*gbinvepsdiff;
 413             gbscale          = isaprod*gbtabscale;
 414             dvdaj            = dvda[jnr+0];
 415
 416             /* Calculate generalized born table index - this is a separate table from the normal one,
 417              * but we use the same procedure by multiplying r with scale and truncating to integer.
 418              */
 419             rt               = r00*gbscale;
 420             gbitab           = rt;
 421             gbeps            = rt-gbitab;
 422             gbitab           = 4*gbitab;
 423
 424             Y                = gbtab[gbitab];
 425             F                = gbtab[gbitab+1];
 426             Geps             = gbeps*gbtab[gbitab+2];
 427             Heps2            = gbeps*gbeps*gbtab[gbitab+3];
 428             Fp               = F+Geps+Heps2;
 429             VV               = Y+gbeps*Fp;
 430             vgb              = gbqqfactor*VV;
 431
 432             FF               = Fp+Geps+2.0*Heps2;
 433             fgb              = gbqqfactor*FF*gbscale;
 434             dvdatmp          = -0.5*(vgb+fgb*r00);
 435             dvdasum          = dvdasum + dvdatmp;
 436             dvda[jnr]        = dvdaj+dvdatmp*isaj0*isaj0;
 437             velec            = qq00*rinv00;
 438             felec            = (velec*rinv00-fgb)*rinv00;
 439
 440             /* BUCKINGHAM DISPERSION/REPULSION */
 441             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 442             vvdw6            = c6_00*rinvsix;
 443             br               = cexp2_00*r00;
 444             vvdwexp          = cexp1_00*exp(-br);
 445             fvdw             = (br*vvdwexp-vvdw6)*rinvsq00;
 446
 447             fscal            = felec+fvdw;
 448
 449             /* Calculate temporary vectorial force */
 450             tx               = fscal*dx00;
 451             ty               = fscal*dy00;
 452             tz               = fscal*dz00;
 453
 454             /* Update vectorial force */
 455             fix0            += tx;
 456             fiy0            += ty;
 457             fiz0            += tz;
 458             f[j_coord_offset+DIM*0+XX] -= tx;
 459             f[j_coord_offset+DIM*0+YY] -= ty;
 460             f[j_coord_offset+DIM*0+ZZ] -= tz;
 461
 462             /* Inner loop uses 92 flops */
 463         }
 464         /* End of innermost loop */
 465
 466         tx = ty = tz = 0;
 467         f[i_coord_offset+DIM*0+XX] += fix0;
 468         f[i_coord_offset+DIM*0+YY] += fiy0;
 469         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 470         tx                         += fix0;
 471         ty                         += fiy0;
 472         tz                         += fiz0;
 473         fshift[i_shift_offset+XX]  += tx;
 474         fshift[i_shift_offset+YY]  += ty;
 475         fshift[i_shift_offset+ZZ]  += tz;
 476
 477         dvda[inr]                   = dvda[inr] + dvdasum*isai0*isai0;
 478
 479         /* Increment number of inner iterations */
 480         inneriter                  += j_index_end - j_index_start;
 481
 482         /* Outer loop uses 13 flops */
 483     }
 484
 485     /* Increment number of outer iterations */
 486     outeriter        += nri;
 487
 488     /* Update outer/inner flops */
 489
 490     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_F,outeriter*13 + inneriter*92);
 491 }