2 * This file is part of the GROMACS molecular simulation package.
4 * Copyright (c) 2012,2013, by the GROMACS development team, led by
5 * Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl,
6 * and including many others, as listed in the AUTHORS file in the
7 * top-level source directory and at http://www.gromacs.org.
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.
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.
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.
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.
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.
36 /* When calculating RF or Ewald interactions we calculate the electrostatic
37 * forces and energies on excluded atom pairs here in the non-bonded loops.
39 #if defined CHECK_EXCLS && defined CALC_COULOMB
53 egp_cj = nbat->energrp[cj];
55 for (i = 0; i < UNROLLI; i++)
63 type_i_off = type[ai]*ntype2;
65 for (j = 0; j < UNROLLJ; j++)
72 real FrLJ6 = 0, FrLJ12 = 0, VLJ = 0;
88 /* A multiply mask used to zero an interaction
89 * when either the distance cutoff is exceeded, or
90 * (if appropriate) the i and j indices are
91 * unsuitable for this kind of inner loop. */
93 #ifdef VDW_CUTOFF_CHECK
97 /* A multiply mask used to zero an interaction
98 * when that interaction should be excluded
99 * (e.g. because of bonding). */
102 interact = ((l_cj[cjind].excl>>(i*UNROLLI + j)) & 1);
106 skipmask = !(cj == ci_sh && j <= i);
115 dx = xi[i*XI_STRIDE+XX] - x[aj*X_STRIDE+XX];
116 dy = xi[i*XI_STRIDE+YY] - x[aj*X_STRIDE+YY];
117 dz = xi[i*XI_STRIDE+ZZ] - x[aj*X_STRIDE+ZZ];
119 rsq = dx*dx + dy*dy + dz*dz;
121 /* Prepare to enforce the cut-off. */
122 skipmask = (rsq >= rcut2) ? 0 : skipmask;
123 /* 9 flops for r^2 + cut-off check */
126 /* Excluded atoms are allowed to be on top of each other.
127 * To avoid overflow of rinv, rinvsq and rinvsix
128 * we add a small number to rsq for excluded pairs only.
130 rsq += (1 - interact)*NBNXN_AVOID_SING_R2_INC;
137 rinv = gmx_invsqrt(rsq);
138 /* 5 flops for invsqrt */
140 /* Partially enforce the cut-off (and perhaps
141 * exclusions) to avoid possible overflow of
142 * rinvsix when computing LJ, and/or overflowing
143 * the Coulomb table during lookup. */
144 rinv = rinv * skipmask;
152 rinvsix = interact*rinvsq*rinvsq*rinvsq;
154 #ifdef VDW_CUTOFF_CHECK
155 skipmask_rvdw = (rsq < rvdw2);
156 rinvsix *= skipmask_rvdw;
159 c6 = nbfp[type_i_off+type[aj]*2 ];
160 c12 = nbfp[type_i_off+type[aj]*2+1];
162 FrLJ12 = c12*rinvsix*rinvsix;
163 /* 6 flops for r^-2 + LJ force */
165 VLJ = (FrLJ12 - c12*sh_invrc6*sh_invrc6)/12 -
166 (FrLJ6 - c6*sh_invrc6)/6;
167 /* Need to zero the interaction if r >= rcut
168 * or there should be exclusion. */
169 VLJ = VLJ * skipmask * interact;
170 /* 9 flops for LJ energy */
171 #ifdef VDW_CUTOFF_CHECK
172 VLJ *= skipmask_rvdw;
175 Vvdw[egp_sh_i[i]+((egp_cj>>(nbat->neg_2log*j)) & egp_mask)] += VLJ;
178 /* 1 flop for LJ energy addition */
184 /* Enforce the cut-off and perhaps exclusions. In
185 * those cases, rinv is zero because of skipmask,
186 * but fcoul and vcoul will later be non-zero (in
187 * both RF and table cases) because of the
188 * contributions that do not depend on rinv. These
189 * contributions cannot be allowed to accumulate
190 * to the force and potential, and the easiest way
191 * to do this is to zero the charges in
193 qq = skipmask * qi[i] * q[aj];
196 fcoul = qq*(interact*rinv*rinvsq - k_rf2);
197 /* 4 flops for RF force */
199 vcoul = qq*(interact*rinv + k_rf*rsq - c_rf);
200 /* 4 flops for RF energy */
205 rs = rsq*rinv*ic->tabq_scale;
209 /* fexcl = F_i + frac * (F_(i+1)-F_i) */
210 fexcl = tab_coul_FDV0[ri*4] + frac*tab_coul_FDV0[ri*4+1];
212 /* fexcl = (1-frac) * F_i + frac * F_(i+1) */
213 fexcl = (1 - frac)*tab_coul_F[ri] + frac*tab_coul_F[ri+1];
215 fcoul = interact*rinvsq - fexcl;
216 /* 7 flops for float 1/r-table force */
219 vcoul = qq*(interact*(rinv - ic->sh_ewald)
220 -(tab_coul_FDV0[ri*4+2]
221 -halfsp*frac*(tab_coul_FDV0[ri*4] + fexcl)));
222 /* 7 flops for float 1/r-table energy (8 with excls) */
224 vcoul = qq*(interact*(rinv - ic->sh_ewald)
226 -halfsp*frac*(tab_coul_F[ri] + fexcl)));
234 Vc[egp_sh_i[i]+((egp_cj>>(nbat->neg_2log*j)) & egp_mask)] += vcoul;
237 /* 1 flop for Coulomb energy addition */
247 fscal = (FrLJ12 - FrLJ6)*rinvsq + fcoul;
248 /* 3 flops for scalar LJ+Coulomb force */
257 fscal = (FrLJ12 - FrLJ6)*rinvsq;
263 /* Increment i-atom force */
264 fi[i*FI_STRIDE+XX] += fx;
265 fi[i*FI_STRIDE+YY] += fy;
266 fi[i*FI_STRIDE+ZZ] += fz;
267 /* Decrement j-atom force */
268 f[aj*F_STRIDE+XX] -= fx;
269 f[aj*F_STRIDE+YY] -= fy;
270 f[aj*F_STRIDE+ZZ] -= fz;
271 /* 9 flops for force addition */