2 * Note: this file was generated by the Gromacs c kernel generator.
4 * This source code is part of
8 * Copyright (c) 2001-2012, The GROMACS Development Team
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
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
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.
28 #include "../nb_kernel.h"
29 #include "types/simple.h"
34 * Gromacs nonbonded kernel: nb_kernel_ElecEwSw_VdwBhamSw_GeomP1P1_VF_c
35 * Electrostatics interaction: Ewald
36 * VdW interaction: Buckingham
37 * Geometry: Particle-Particle
38 * Calculate force/pot: PotentialAndForce
41 nb_kernel_ElecEwSw_VdwBhamSw_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)
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;
57 real ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
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;
64 real rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
68 real ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
70 real rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
77 jindex = nlist->jindex;
79 shiftidx = nlist->shift;
81 shiftvec = fr->shift_vec[0];
82 fshift = fr->fshift[0];
84 charge = mdatoms->chargeA;
87 vdwtype = mdatoms->typeA;
89 sh_ewald = fr->ic->sh_ewald;
90 ewtab = fr->ic->tabq_coul_FDV0;
91 ewtabscale = fr->ic->tabq_scale;
92 ewtabhalfspace = 0.5/ewtabscale;
94 /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
95 rcutoff = fr->rcoulomb;
96 rcutoff2 = rcutoff*rcutoff;
98 rswitch = fr->rcoulomb_switch;
99 /* Setup switch parameters */
101 swV3 = -10.0/(d*d*d);
102 swV4 = 15.0/(d*d*d*d);
103 swV5 = -6.0/(d*d*d*d*d);
104 swF2 = -30.0/(d*d*d);
105 swF3 = 60.0/(d*d*d*d);
106 swF4 = -30.0/(d*d*d*d*d);
111 /* Start outer loop over neighborlists */
112 for(iidx=0; iidx<nri; iidx++)
114 /* Load shift vector for this list */
115 i_shift_offset = DIM*shiftidx[iidx];
116 shX = shiftvec[i_shift_offset+XX];
117 shY = shiftvec[i_shift_offset+YY];
118 shZ = shiftvec[i_shift_offset+ZZ];
120 /* Load limits for loop over neighbors */
121 j_index_start = jindex[iidx];
122 j_index_end = jindex[iidx+1];
124 /* Get outer coordinate index */
126 i_coord_offset = DIM*inr;
128 /* Load i particle coords and add shift vector */
129 ix0 = shX + x[i_coord_offset+DIM*0+XX];
130 iy0 = shY + x[i_coord_offset+DIM*0+YY];
131 iz0 = shZ + x[i_coord_offset+DIM*0+ZZ];
137 /* Load parameters for i particles */
138 iq0 = facel*charge[inr+0];
139 vdwioffset0 = 3*nvdwtype*vdwtype[inr+0];
141 /* Reset potential sums */
145 /* Start inner kernel loop */
146 for(jidx=j_index_start; jidx<j_index_end; jidx++)
148 /* Get j neighbor index, and coordinate index */
150 j_coord_offset = DIM*jnr;
152 /* load j atom coordinates */
153 jx0 = x[j_coord_offset+DIM*0+XX];
154 jy0 = x[j_coord_offset+DIM*0+YY];
155 jz0 = x[j_coord_offset+DIM*0+ZZ];
157 /* Calculate displacement vector */
162 /* Calculate squared distance and things based on it */
163 rsq00 = dx00*dx00+dy00*dy00+dz00*dz00;
165 rinv00 = gmx_invsqrt(rsq00);
167 rinvsq00 = rinv00*rinv00;
169 /* Load parameters for j particles */
171 vdwjidx0 = 3*vdwtype[jnr+0];
173 /**************************
174 * CALCULATE INTERACTIONS *
175 **************************/
183 c6_00 = vdwparam[vdwioffset0+vdwjidx0];
184 cexp1_00 = vdwparam[vdwioffset0+vdwjidx0+1];
185 cexp2_00 = vdwparam[vdwioffset0+vdwjidx0+2];
187 /* EWALD ELECTROSTATICS */
189 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
190 ewrt = r00*ewtabscale;
194 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
195 velec = qq00*(rinv00-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
196 felec = qq00*rinv00*(rinvsq00-felec);
198 /* BUCKINGHAM DISPERSION/REPULSION */
199 rinvsix = rinvsq00*rinvsq00*rinvsq00;
200 vvdw6 = c6_00*rinvsix;
202 vvdwexp = cexp1_00*exp(-br);
203 vvdw = vvdwexp - vvdw6*(1.0/6.0);
204 fvdw = (br*vvdwexp-vvdw6)*rinvsq00;
207 d = (d>0.0) ? d : 0.0;
209 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
211 dsw = d2*(swF2+d*(swF3+d*swF4));
213 /* Evaluate switch function */
214 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
215 felec = felec*sw - rinv00*velec*dsw;
216 fvdw = fvdw*sw - rinv00*vvdw*dsw;
220 /* Update potential sums from outer loop */
226 /* Calculate temporary vectorial force */
231 /* Update vectorial force */
235 f[j_coord_offset+DIM*0+XX] -= tx;
236 f[j_coord_offset+DIM*0+YY] -= ty;
237 f[j_coord_offset+DIM*0+ZZ] -= tz;
241 /* Inner loop uses 101 flops */
243 /* End of innermost loop */
246 f[i_coord_offset+DIM*0+XX] += fix0;
247 f[i_coord_offset+DIM*0+YY] += fiy0;
248 f[i_coord_offset+DIM*0+ZZ] += fiz0;
252 fshift[i_shift_offset+XX] += tx;
253 fshift[i_shift_offset+YY] += ty;
254 fshift[i_shift_offset+ZZ] += tz;
257 /* Update potential energies */
258 kernel_data->energygrp_elec[ggid] += velecsum;
259 kernel_data->energygrp_vdw[ggid] += vvdwsum;
261 /* Increment number of inner iterations */
262 inneriter += j_index_end - j_index_start;
264 /* Outer loop uses 15 flops */
267 /* Increment number of outer iterations */
270 /* Update outer/inner flops */
272 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_VF,outeriter*15 + inneriter*101);
275 * Gromacs nonbonded kernel: nb_kernel_ElecEwSw_VdwBhamSw_GeomP1P1_F_c
276 * Electrostatics interaction: Ewald
277 * VdW interaction: Buckingham
278 * Geometry: Particle-Particle
279 * Calculate force/pot: Force
282 nb_kernel_ElecEwSw_VdwBhamSw_GeomP1P1_F_c
283 (t_nblist * gmx_restrict nlist,
284 rvec * gmx_restrict xx,
285 rvec * gmx_restrict ff,
286 t_forcerec * gmx_restrict fr,
287 t_mdatoms * gmx_restrict mdatoms,
288 nb_kernel_data_t * gmx_restrict kernel_data,
289 t_nrnb * gmx_restrict nrnb)
291 int i_shift_offset,i_coord_offset,j_coord_offset;
292 int j_index_start,j_index_end;
293 int nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
294 real shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
295 int *iinr,*jindex,*jjnr,*shiftidx,*gid;
296 real *shiftvec,*fshift,*x,*f;
298 real ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
300 real jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
301 real dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
302 real velec,felec,velecsum,facel,crf,krf,krf2;
305 real rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
309 real ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
311 real rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
318 jindex = nlist->jindex;
320 shiftidx = nlist->shift;
322 shiftvec = fr->shift_vec[0];
323 fshift = fr->fshift[0];
325 charge = mdatoms->chargeA;
326 nvdwtype = fr->ntype;
328 vdwtype = mdatoms->typeA;
330 sh_ewald = fr->ic->sh_ewald;
331 ewtab = fr->ic->tabq_coul_FDV0;
332 ewtabscale = fr->ic->tabq_scale;
333 ewtabhalfspace = 0.5/ewtabscale;
335 /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
336 rcutoff = fr->rcoulomb;
337 rcutoff2 = rcutoff*rcutoff;
339 rswitch = fr->rcoulomb_switch;
340 /* Setup switch parameters */
342 swV3 = -10.0/(d*d*d);
343 swV4 = 15.0/(d*d*d*d);
344 swV5 = -6.0/(d*d*d*d*d);
345 swF2 = -30.0/(d*d*d);
346 swF3 = 60.0/(d*d*d*d);
347 swF4 = -30.0/(d*d*d*d*d);
352 /* Start outer loop over neighborlists */
353 for(iidx=0; iidx<nri; iidx++)
355 /* Load shift vector for this list */
356 i_shift_offset = DIM*shiftidx[iidx];
357 shX = shiftvec[i_shift_offset+XX];
358 shY = shiftvec[i_shift_offset+YY];
359 shZ = shiftvec[i_shift_offset+ZZ];
361 /* Load limits for loop over neighbors */
362 j_index_start = jindex[iidx];
363 j_index_end = jindex[iidx+1];
365 /* Get outer coordinate index */
367 i_coord_offset = DIM*inr;
369 /* Load i particle coords and add shift vector */
370 ix0 = shX + x[i_coord_offset+DIM*0+XX];
371 iy0 = shY + x[i_coord_offset+DIM*0+YY];
372 iz0 = shZ + x[i_coord_offset+DIM*0+ZZ];
378 /* Load parameters for i particles */
379 iq0 = facel*charge[inr+0];
380 vdwioffset0 = 3*nvdwtype*vdwtype[inr+0];
382 /* Start inner kernel loop */
383 for(jidx=j_index_start; jidx<j_index_end; jidx++)
385 /* Get j neighbor index, and coordinate index */
387 j_coord_offset = DIM*jnr;
389 /* load j atom coordinates */
390 jx0 = x[j_coord_offset+DIM*0+XX];
391 jy0 = x[j_coord_offset+DIM*0+YY];
392 jz0 = x[j_coord_offset+DIM*0+ZZ];
394 /* Calculate displacement vector */
399 /* Calculate squared distance and things based on it */
400 rsq00 = dx00*dx00+dy00*dy00+dz00*dz00;
402 rinv00 = gmx_invsqrt(rsq00);
404 rinvsq00 = rinv00*rinv00;
406 /* Load parameters for j particles */
408 vdwjidx0 = 3*vdwtype[jnr+0];
410 /**************************
411 * CALCULATE INTERACTIONS *
412 **************************/
420 c6_00 = vdwparam[vdwioffset0+vdwjidx0];
421 cexp1_00 = vdwparam[vdwioffset0+vdwjidx0+1];
422 cexp2_00 = vdwparam[vdwioffset0+vdwjidx0+2];
424 /* EWALD ELECTROSTATICS */
426 /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
427 ewrt = r00*ewtabscale;
431 felec = ewtab[ewitab]+eweps*ewtab[ewitab+1];
432 velec = qq00*(rinv00-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
433 felec = qq00*rinv00*(rinvsq00-felec);
435 /* BUCKINGHAM DISPERSION/REPULSION */
436 rinvsix = rinvsq00*rinvsq00*rinvsq00;
437 vvdw6 = c6_00*rinvsix;
439 vvdwexp = cexp1_00*exp(-br);
440 vvdw = vvdwexp - vvdw6*(1.0/6.0);
441 fvdw = (br*vvdwexp-vvdw6)*rinvsq00;
444 d = (d>0.0) ? d : 0.0;
446 sw = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
448 dsw = d2*(swF2+d*(swF3+d*swF4));
450 /* Evaluate switch function */
451 /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
452 felec = felec*sw - rinv00*velec*dsw;
453 fvdw = fvdw*sw - rinv00*vvdw*dsw;
457 /* Calculate temporary vectorial force */
462 /* Update vectorial force */
466 f[j_coord_offset+DIM*0+XX] -= tx;
467 f[j_coord_offset+DIM*0+YY] -= ty;
468 f[j_coord_offset+DIM*0+ZZ] -= tz;
472 /* Inner loop uses 97 flops */
474 /* End of innermost loop */
477 f[i_coord_offset+DIM*0+XX] += fix0;
478 f[i_coord_offset+DIM*0+YY] += fiy0;
479 f[i_coord_offset+DIM*0+ZZ] += fiz0;
483 fshift[i_shift_offset+XX] += tx;
484 fshift[i_shift_offset+YY] += ty;
485 fshift[i_shift_offset+ZZ] += tz;
487 /* Increment number of inner iterations */
488 inneriter += j_index_end - j_index_start;
490 /* Outer loop uses 13 flops */
493 /* Increment number of outer iterations */
496 /* Update outer/inner flops */
498 inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_F,outeriter*13 + inneriter*97);