src/gromacs/gmxlib/nonbonded/nb_kernel_c/nb_kernel_ElecGB_VdwCSTab_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_VdwCSTab_GeomP1P1_VF_c
  35  * Electrostatics interaction: GeneralizedBorn
  36  * VdW interaction:            CubicSplineTable
  37  * Geometry:                   Particle-Particle
  38  * Calculate force/pot:        PotentialAndForce
  39  */
  40 void
  41 nb_kernel_ElecGB_VdwCSTab_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     vftab            = kernel_data->table_vdw->data;
  92     vftabscale       = kernel_data->table_vdw->scale;
  93
  94     invsqrta         = fr->invsqrta;
  95     dvda             = fr->dvda;
  96     gbtabscale       = fr->gbtab.scale;
  97     gbtab            = fr->gbtab.data;
  98     gbinvepsdiff     = (1.0/fr->epsilon_r) - (1.0/fr->gb_epsilon_solvent);
  99
 100     outeriter        = 0;
 101     inneriter        = 0;
 102
 103     /* Start outer loop over neighborlists */
 104     for(iidx=0; iidx<nri; iidx++)
 105     {
 106         /* Load shift vector for this list */
 107         i_shift_offset   = DIM*shiftidx[iidx];
 108         shX              = shiftvec[i_shift_offset+XX];
 109         shY              = shiftvec[i_shift_offset+YY];
 110         shZ              = shiftvec[i_shift_offset+ZZ];
 111
 112         /* Load limits for loop over neighbors */
 113         j_index_start    = jindex[iidx];
 114         j_index_end      = jindex[iidx+1];
 115
 116         /* Get outer coordinate index */
 117         inr              = iinr[iidx];
 118         i_coord_offset   = DIM*inr;
 119
 120         /* Load i particle coords and add shift vector */
 121         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 122         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 123         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 124
 125         fix0             = 0.0;
 126         fiy0             = 0.0;
 127         fiz0             = 0.0;
 128
 129         /* Load parameters for i particles */
 130         iq0              = facel*charge[inr+0];
 131         isai0            = invsqrta[inr+0];
 132         vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 133
 134         /* Reset potential sums */
 135         velecsum         = 0.0;
 136         vgbsum           = 0.0;
 137         vvdwsum          = 0.0;
 138         dvdasum          = 0.0;
 139
 140         /* Start inner kernel loop */
 141         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 142         {
 143             /* Get j neighbor index, and coordinate index */
 144             jnr              = jjnr[jidx];
 145             j_coord_offset   = DIM*jnr;
 146
 147             /* load j atom coordinates */
 148             jx0              = x[j_coord_offset+DIM*0+XX];
 149             jy0              = x[j_coord_offset+DIM*0+YY];
 150             jz0              = x[j_coord_offset+DIM*0+ZZ];
 151
 152             /* Calculate displacement vector */
 153             dx00             = ix0 - jx0;
 154             dy00             = iy0 - jy0;
 155             dz00             = iz0 - jz0;
 156
 157             /* Calculate squared distance and things based on it */
 158             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 159
 160             rinv00           = gmx_invsqrt(rsq00);
 161
 162             /* Load parameters for j particles */
 163             jq0              = charge[jnr+0];
 164             isaj0           = invsqrta[jnr+0];
 165             vdwjidx0         = 2*vdwtype[jnr+0];
 166
 167             /**************************
 168              * CALCULATE INTERACTIONS *
 169              **************************/
 170
 171             r00              = rsq00*rinv00;
 172
 173             qq00             = iq0*jq0;
 174             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 175             c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 176
 177             /* Calculate table index by multiplying r with table scale and truncate to integer */
 178             rt               = r00*vftabscale;
 179             vfitab           = rt;
 180             vfeps            = rt-vfitab;
 181             vfitab           = 2*4*vfitab;
 182
 183             /* GENERALIZED BORN AND COULOMB ELECTROSTATICS */
 184             isaprod          = isai0*isaj0;
 185             gbqqfactor       = isaprod*(-qq00)*gbinvepsdiff;
 186             gbscale          = isaprod*gbtabscale;
 187             dvdaj            = dvda[jnr+0];
 188
 189             /* Calculate generalized born table index - this is a separate table from the normal one,
 190              * but we use the same procedure by multiplying r with scale and truncating to integer.
 191              */
 192             rt               = r00*gbscale;
 193             gbitab           = rt;
 194             gbeps            = rt-gbitab;
 195             gbitab           = 4*gbitab;
 196
 197             Y                = gbtab[gbitab];
 198             F                = gbtab[gbitab+1];
 199             Geps             = gbeps*gbtab[gbitab+2];
 200             Heps2            = gbeps*gbeps*gbtab[gbitab+3];
 201             Fp               = F+Geps+Heps2;
 202             VV               = Y+gbeps*Fp;
 203             vgb              = gbqqfactor*VV;
 204
 205             FF               = Fp+Geps+2.0*Heps2;
 206             fgb              = gbqqfactor*FF*gbscale;
 207             dvdatmp          = -0.5*(vgb+fgb*r00);
 208             dvdasum          = dvdasum + dvdatmp;
 209             dvda[jnr]        = dvdaj+dvdatmp*isaj0*isaj0;
 210             velec            = qq00*rinv00;
 211             felec            = (velec*rinv00-fgb)*rinv00;
 212
 213             /* CUBIC SPLINE TABLE DISPERSION */
 214             vfitab          += 0;
 215             Y                = vftab[vfitab];
 216             F                = vftab[vfitab+1];
 217             Geps             = vfeps*vftab[vfitab+2];
 218             Heps2            = vfeps*vfeps*vftab[vfitab+3];
 219             Fp               = F+Geps+Heps2;
 220             VV               = Y+vfeps*Fp;
 221             vvdw6            = c6_00*VV;
 222             FF               = Fp+Geps+2.0*Heps2;
 223             fvdw6            = c6_00*FF;
 224
 225             /* CUBIC SPLINE TABLE REPULSION */
 226             Y                = vftab[vfitab+4];
 227             F                = vftab[vfitab+5];
 228             Geps             = vfeps*vftab[vfitab+6];
 229             Heps2            = vfeps*vfeps*vftab[vfitab+7];
 230             Fp               = F+Geps+Heps2;
 231             VV               = Y+vfeps*Fp;
 232             vvdw12           = c12_00*VV;
 233             FF               = Fp+Geps+2.0*Heps2;
 234             fvdw12           = c12_00*FF;
 235             vvdw             = vvdw12+vvdw6;
 236             fvdw             = -(fvdw6+fvdw12)*vftabscale*rinv00;
 237
 238             /* Update potential sums from outer loop */
 239             velecsum        += velec;
 240             vgbsum          += vgb;
 241             vvdwsum         += vvdw;
 242
 243             fscal            = felec+fvdw;
 244
 245             /* Calculate temporary vectorial force */
 246             tx               = fscal*dx00;
 247             ty               = fscal*dy00;
 248             tz               = fscal*dz00;
 249
 250             /* Update vectorial force */
 251             fix0            += tx;
 252             fiy0            += ty;
 253             fiz0            += tz;
 254             f[j_coord_offset+DIM*0+XX] -= tx;
 255             f[j_coord_offset+DIM*0+YY] -= ty;
 256             f[j_coord_offset+DIM*0+ZZ] -= tz;
 257
 258             /* Inner loop uses 91 flops */
 259         }
 260         /* End of innermost loop */
 261
 262         tx = ty = tz = 0;
 263         f[i_coord_offset+DIM*0+XX] += fix0;
 264         f[i_coord_offset+DIM*0+YY] += fiy0;
 265         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 266         tx                         += fix0;
 267         ty                         += fiy0;
 268         tz                         += fiz0;
 269         fshift[i_shift_offset+XX]  += tx;
 270         fshift[i_shift_offset+YY]  += ty;
 271         fshift[i_shift_offset+ZZ]  += tz;
 272
 273         ggid                        = gid[iidx];
 274         /* Update potential energies */
 275         kernel_data->energygrp_elec[ggid] += velecsum;
 276         kernel_data->energygrp_polarization[ggid] += vgbsum;
 277         kernel_data->energygrp_vdw[ggid] += vvdwsum;
 278         dvda[inr]                   = dvda[inr] + dvdasum*isai0*isai0;
 279
 280         /* Increment number of inner iterations */
 281         inneriter                  += j_index_end - j_index_start;
 282
 283         /* Outer loop uses 16 flops */
 284     }
 285
 286     /* Increment number of outer iterations */
 287     outeriter        += nri;
 288
 289     /* Update outer/inner flops */
 290
 291     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_VF,outeriter*16 + inneriter*91);
 292 }
 293 /*
 294  * Gromacs nonbonded kernel:   nb_kernel_ElecGB_VdwCSTab_GeomP1P1_F_c
 295  * Electrostatics interaction: GeneralizedBorn
 296  * VdW interaction:            CubicSplineTable
 297  * Geometry:                   Particle-Particle
 298  * Calculate force/pot:        Force
 299  */
 300 void
 301 nb_kernel_ElecGB_VdwCSTab_GeomP1P1_F_c
 302                     (t_nblist * gmx_restrict                nlist,
 303                      rvec * gmx_restrict                    xx,
 304                      rvec * gmx_restrict                    ff,
 305                      t_forcerec * gmx_restrict              fr,
 306                      t_mdatoms * gmx_restrict               mdatoms,
 307                      nb_kernel_data_t * gmx_restrict        kernel_data,
 308                      t_nrnb * gmx_restrict                  nrnb)
 309 {
 310     int              i_shift_offset,i_coord_offset,j_coord_offset;
 311     int              j_index_start,j_index_end;
 312     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
 313     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
 314     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 315     real             *shiftvec,*fshift,*x,*f;
 316     int              vdwioffset0;
 317     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 318     int              vdwjidx0;
 319     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 320     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
 321     real             velec,felec,velecsum,facel,crf,krf,krf2;
 322     real             *charge;
 323     int              gbitab;
 324     real             vgb,fgb,vgbsum,dvdasum,gbscale,gbtabscale,isaprod,gbqqfactor,gbinvepsdiff,dvdaj,gbeps,dvdatmp;
 325     real             *invsqrta,*dvda,*gbtab;
 326     int              nvdwtype;
 327     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
 328     int              *vdwtype;
 329     real             *vdwparam;
 330     int              vfitab;
 331     real             rt,vfeps,vftabscale,Y,F,Geps,Heps2,Fp,VV,FF;
 332     real             *vftab;
 333
 334     x                = xx[0];
 335     f                = ff[0];
 336
 337     nri              = nlist->nri;
 338     iinr             = nlist->iinr;
 339     jindex           = nlist->jindex;
 340     jjnr             = nlist->jjnr;
 341     shiftidx         = nlist->shift;
 342     gid              = nlist->gid;
 343     shiftvec         = fr->shift_vec[0];
 344     fshift           = fr->fshift[0];
 345     facel            = fr->epsfac;
 346     charge           = mdatoms->chargeA;
 347     nvdwtype         = fr->ntype;
 348     vdwparam         = fr->nbfp;
 349     vdwtype          = mdatoms->typeA;
 350
 351     vftab            = kernel_data->table_vdw->data;
 352     vftabscale       = kernel_data->table_vdw->scale;
 353
 354     invsqrta         = fr->invsqrta;
 355     dvda             = fr->dvda;
 356     gbtabscale       = fr->gbtab.scale;
 357     gbtab            = fr->gbtab.data;
 358     gbinvepsdiff     = (1.0/fr->epsilon_r) - (1.0/fr->gb_epsilon_solvent);
 359
 360     outeriter        = 0;
 361     inneriter        = 0;
 362
 363     /* Start outer loop over neighborlists */
 364     for(iidx=0; iidx<nri; iidx++)
 365     {
 366         /* Load shift vector for this list */
 367         i_shift_offset   = DIM*shiftidx[iidx];
 368         shX              = shiftvec[i_shift_offset+XX];
 369         shY              = shiftvec[i_shift_offset+YY];
 370         shZ              = shiftvec[i_shift_offset+ZZ];
 371
 372         /* Load limits for loop over neighbors */
 373         j_index_start    = jindex[iidx];
 374         j_index_end      = jindex[iidx+1];
 375
 376         /* Get outer coordinate index */
 377         inr              = iinr[iidx];
 378         i_coord_offset   = DIM*inr;
 379
 380         /* Load i particle coords and add shift vector */
 381         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 382         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 383         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 384
 385         fix0             = 0.0;
 386         fiy0             = 0.0;
 387         fiz0             = 0.0;
 388
 389         /* Load parameters for i particles */
 390         iq0              = facel*charge[inr+0];
 391         isai0            = invsqrta[inr+0];
 392         vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 393
 394         dvdasum          = 0.0;
 395
 396         /* Start inner kernel loop */
 397         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 398         {
 399             /* Get j neighbor index, and coordinate index */
 400             jnr              = jjnr[jidx];
 401             j_coord_offset   = DIM*jnr;
 402
 403             /* load j atom coordinates */
 404             jx0              = x[j_coord_offset+DIM*0+XX];
 405             jy0              = x[j_coord_offset+DIM*0+YY];
 406             jz0              = x[j_coord_offset+DIM*0+ZZ];
 407
 408             /* Calculate displacement vector */
 409             dx00             = ix0 - jx0;
 410             dy00             = iy0 - jy0;
 411             dz00             = iz0 - jz0;
 412
 413             /* Calculate squared distance and things based on it */
 414             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 415
 416             rinv00           = gmx_invsqrt(rsq00);
 417
 418             /* Load parameters for j particles */
 419             jq0              = charge[jnr+0];
 420             isaj0           = invsqrta[jnr+0];
 421             vdwjidx0         = 2*vdwtype[jnr+0];
 422
 423             /**************************
 424              * CALCULATE INTERACTIONS *
 425              **************************/
 426
 427             r00              = rsq00*rinv00;
 428
 429             qq00             = iq0*jq0;
 430             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 431             c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 432
 433             /* Calculate table index by multiplying r with table scale and truncate to integer */
 434             rt               = r00*vftabscale;
 435             vfitab           = rt;
 436             vfeps            = rt-vfitab;
 437             vfitab           = 2*4*vfitab;
 438
 439             /* GENERALIZED BORN AND COULOMB ELECTROSTATICS */
 440             isaprod          = isai0*isaj0;
 441             gbqqfactor       = isaprod*(-qq00)*gbinvepsdiff;
 442             gbscale          = isaprod*gbtabscale;
 443             dvdaj            = dvda[jnr+0];
 444
 445             /* Calculate generalized born table index - this is a separate table from the normal one,
 446              * but we use the same procedure by multiplying r with scale and truncating to integer.
 447              */
 448             rt               = r00*gbscale;
 449             gbitab           = rt;
 450             gbeps            = rt-gbitab;
 451             gbitab           = 4*gbitab;
 452
 453             Y                = gbtab[gbitab];
 454             F                = gbtab[gbitab+1];
 455             Geps             = gbeps*gbtab[gbitab+2];
 456             Heps2            = gbeps*gbeps*gbtab[gbitab+3];
 457             Fp               = F+Geps+Heps2;
 458             VV               = Y+gbeps*Fp;
 459             vgb              = gbqqfactor*VV;
 460
 461             FF               = Fp+Geps+2.0*Heps2;
 462             fgb              = gbqqfactor*FF*gbscale;
 463             dvdatmp          = -0.5*(vgb+fgb*r00);
 464             dvdasum          = dvdasum + dvdatmp;
 465             dvda[jnr]        = dvdaj+dvdatmp*isaj0*isaj0;
 466             velec            = qq00*rinv00;
 467             felec            = (velec*rinv00-fgb)*rinv00;
 468
 469             /* CUBIC SPLINE TABLE DISPERSION */
 470             vfitab          += 0;
 471             F                = vftab[vfitab+1];
 472             Geps             = vfeps*vftab[vfitab+2];
 473             Heps2            = vfeps*vfeps*vftab[vfitab+3];
 474             Fp               = F+Geps+Heps2;
 475             FF               = Fp+Geps+2.0*Heps2;
 476             fvdw6            = c6_00*FF;
 477
 478             /* CUBIC SPLINE TABLE REPULSION */
 479             F                = vftab[vfitab+5];
 480             Geps             = vfeps*vftab[vfitab+6];
 481             Heps2            = vfeps*vfeps*vftab[vfitab+7];
 482             Fp               = F+Geps+Heps2;
 483             FF               = Fp+Geps+2.0*Heps2;
 484             fvdw12           = c12_00*FF;
 485             fvdw             = -(fvdw6+fvdw12)*vftabscale*rinv00;
 486
 487             fscal            = felec+fvdw;
 488
 489             /* Calculate temporary vectorial force */
 490             tx               = fscal*dx00;
 491             ty               = fscal*dy00;
 492             tz               = fscal*dz00;
 493
 494             /* Update vectorial force */
 495             fix0            += tx;
 496             fiy0            += ty;
 497             fiz0            += tz;
 498             f[j_coord_offset+DIM*0+XX] -= tx;
 499             f[j_coord_offset+DIM*0+YY] -= ty;
 500             f[j_coord_offset+DIM*0+ZZ] -= tz;
 501
 502             /* Inner loop uses 81 flops */
 503         }
 504         /* End of innermost loop */
 505
 506         tx = ty = tz = 0;
 507         f[i_coord_offset+DIM*0+XX] += fix0;
 508         f[i_coord_offset+DIM*0+YY] += fiy0;
 509         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 510         tx                         += fix0;
 511         ty                         += fiy0;
 512         tz                         += fiz0;
 513         fshift[i_shift_offset+XX]  += tx;
 514         fshift[i_shift_offset+YY]  += ty;
 515         fshift[i_shift_offset+ZZ]  += tz;
 516
 517         dvda[inr]                   = dvda[inr] + dvdasum*isai0*isai0;
 518
 519         /* Increment number of inner iterations */
 520         inneriter                  += j_index_end - j_index_start;
 521
 522         /* Outer loop uses 13 flops */
 523     }
 524
 525     /* Increment number of outer iterations */
 526     outeriter        += nri;
 527
 528     /* Update outer/inner flops */
 529
 530     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_F,outeriter*13 + inneriter*81);
 531 }