2 * This file is part of the GROMACS molecular simulation package.
4 * Copyright (c) 1991-2000, University of Groningen, The Netherlands.
5 * Copyright (c) 2001-2004, The GROMACS development team.
6 * Copyright (c) 2013,2014,2015,2016,2017,2018,2019, by the GROMACS development team, led by
7 * Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl,
8 * and including many others, as listed in the AUTHORS file in the
9 * top-level source directory and at http://www.gromacs.org.
11 * GROMACS is free software; you can redistribute it and/or
12 * modify it under the terms of the GNU Lesser General Public License
13 * as published by the Free Software Foundation; either version 2.1
14 * of the License, or (at your option) any later version.
16 * GROMACS is distributed in the hope that it will be useful,
17 * but WITHOUT ANY WARRANTY; without even the implied warranty of
18 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
19 * Lesser General Public License for more details.
21 * You should have received a copy of the GNU Lesser General Public
22 * License along with GROMACS; if not, see
23 * http://www.gnu.org/licenses, or write to the Free Software Foundation,
24 * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
26 * If you want to redistribute modifications to GROMACS, please
27 * consider that scientific software is very special. Version
28 * control is crucial - bugs must be traceable. We will be happy to
29 * consider code for inclusion in the official distribution, but
30 * derived work must not be called official GROMACS. Details are found
31 * in the README & COPYING files - if they are missing, get the
32 * official version at http://www.gromacs.org.
34 * To help us fund GROMACS development, we humbly ask that you cite
35 * the research papers on the package. Check out http://www.gromacs.org.
45 #include "gromacs/domdec/dlbtiming.h"
46 #include "gromacs/domdec/domdec.h"
47 #include "gromacs/domdec/domdec_struct.h"
48 #include "gromacs/ewald/ewald.h"
49 #include "gromacs/ewald/long_range_correction.h"
50 #include "gromacs/ewald/pme.h"
51 #include "gromacs/gmxlib/network.h"
52 #include "gromacs/gmxlib/nrnb.h"
53 #include "gromacs/listed_forces/listed_forces.h"
54 #include "gromacs/math/vec.h"
55 #include "gromacs/math/vecdump.h"
56 #include "gromacs/mdlib/forcerec_threading.h"
57 #include "gromacs/mdlib/qmmm.h"
58 #include "gromacs/mdlib/rf_util.h"
59 #include "gromacs/mdlib/wall.h"
60 #include "gromacs/mdtypes/commrec.h"
61 #include "gromacs/mdtypes/enerdata.h"
62 #include "gromacs/mdtypes/forceoutput.h"
63 #include "gromacs/mdtypes/forcerec.h"
64 #include "gromacs/mdtypes/inputrec.h"
65 #include "gromacs/mdtypes/md_enums.h"
66 #include "gromacs/mdtypes/mdatom.h"
67 #include "gromacs/mdtypes/simulation_workload.h"
68 #include "gromacs/pbcutil/ishift.h"
69 #include "gromacs/pbcutil/mshift.h"
70 #include "gromacs/pbcutil/pbc.h"
71 #include "gromacs/timing/wallcycle.h"
72 #include "gromacs/utility/exceptions.h"
73 #include "gromacs/utility/fatalerror.h"
74 #include "gromacs/utility/smalloc.h"
76 static void clearEwaldThreadOutput(ewald_corr_thread_t *ewc_t)
80 ewc_t->dvdl[efptCOUL] = 0;
81 ewc_t->dvdl[efptVDW] = 0;
82 clear_mat(ewc_t->vir_q);
83 clear_mat(ewc_t->vir_lj);
86 static void reduceEwaldThreadOuput(int nthreads, ewald_corr_thread_t *ewc_t)
88 ewald_corr_thread_t &dest = ewc_t[0];
90 for (int t = 1; t < nthreads; t++)
92 dest.Vcorr_q += ewc_t[t].Vcorr_q;
93 dest.Vcorr_lj += ewc_t[t].Vcorr_lj;
94 dest.dvdl[efptCOUL] += ewc_t[t].dvdl[efptCOUL];
95 dest.dvdl[efptVDW] += ewc_t[t].dvdl[efptVDW];
96 m_add(dest.vir_q, ewc_t[t].vir_q, dest.vir_q);
97 m_add(dest.vir_lj, ewc_t[t].vir_lj, dest.vir_lj);
102 do_force_lowlevel(t_forcerec *fr,
103 const t_inputrec *ir,
106 const gmx_multisim_t *ms,
108 gmx_wallcycle_t wcycle,
110 gmx::ArrayRefWithPadding<gmx::RVec> coordinates,
112 gmx::ForceOutputs *forceOutputs,
113 gmx_enerdata_t *enerd,
117 const t_graph *graph,
119 const gmx::StepWorkload &stepWork,
120 const DDBalanceRegionHandler &ddBalanceRegionHandler)
122 // TODO: Replace all uses of x by const coordinates
123 rvec *x = as_rvec_array(coordinates.paddedArrayRef().data());
125 auto &forceWithVirial = forceOutputs->forceWithVirial();
127 /* do QMMM first if requested */
130 enerd->term[F_EQM] = calculate_QMMM(cr, &forceOutputs->forceWithShiftForces(), fr);
133 /* Call the short range functions all in one go. */
137 /* foreign lambda component for walls */
138 real dvdl_walls = do_walls(*ir, *fr, box, *md, x,
139 &forceWithVirial, lambda[efptVDW],
140 enerd->grpp.ener[egLJSR].data(), nrnb);
141 enerd->dvdl_lin[efptVDW] += dvdl_walls;
144 /* Shift the coordinates. Must be done before listed forces and PPPM,
145 * but is also necessary for SHAKE and update, therefore it can NOT
146 * go when no listed forces have to be evaluated.
148 * The shifting and PBC code is deliberately not timed, since with
149 * the Verlet scheme it only takes non-zero time with triclinic
150 * boxes, and even then the time is around a factor of 100 less
151 * than the next smallest counter.
155 /* Here sometimes we would not need to shift with NBFonly,
156 * but we do so anyhow for consistency of the returned coordinates.
160 shift_self(graph, box, x);
163 inc_nrnb(nrnb, eNR_SHIFTX, 2*graph->nnodes);
167 inc_nrnb(nrnb, eNR_SHIFTX, graph->nnodes);
174 /* Check whether we need to take into account PBC in listed interactions. */
175 const auto needPbcForListedForces = fr->bMolPBC && stepWork.computeListedForces && haveCpuListedForces(*fr, *idef, *fcd);
176 if (needPbcForListedForces)
178 /* Since all atoms are in the rectangular or triclinic unit-cell,
179 * only single box vector shifts (2 in x) are required.
181 set_pbc_dd(&pbc, fr->ePBC, DOMAINDECOMP(cr) ? cr->dd->nc : nullptr,
185 do_force_listed(wcycle, box, ir->fepvals, cr, ms,
188 fr, &pbc, graph, enerd, nrnb, lambda, md, fcd,
189 DOMAINDECOMP(cr) ? cr->dd->globalAtomIndices.data() : nullptr,
193 const bool computePmeOnCpu =
194 (EEL_PME(fr->ic->eeltype) || EVDW_PME(fr->ic->vdwtype)) &&
195 thisRankHasDuty(cr, DUTY_PME) &&
196 (pme_run_mode(fr->pmedata) == PmeRunMode::CPU);
198 const bool haveEwaldSurfaceTerm = haveEwaldSurfaceContribution(*ir);
200 /* Do long-range electrostatics and/or LJ-PME
201 * and compute PME surface terms when necessary.
203 if (computePmeOnCpu ||
204 fr->ic->eeltype == eelEWALD ||
205 haveEwaldSurfaceTerm)
208 real Vlr_q = 0, Vlr_lj = 0;
210 /* We reduce all virial, dV/dlambda and energy contributions, except
211 * for the reciprocal energies (Vlr_q, Vlr_lj) into the same struct.
213 ewald_corr_thread_t &ewaldOutput = fr->ewc_t[0];
214 clearEwaldThreadOutput(&ewaldOutput);
216 if (EEL_PME_EWALD(fr->ic->eeltype) || EVDW_PME(fr->ic->vdwtype))
218 /* Calculate the Ewald surface force and energy contributions, when necessary */
219 if (haveEwaldSurfaceTerm)
221 wallcycle_sub_start(wcycle, ewcsEWALD_CORRECTION);
225 gmx_fatal(FARGS, "TPI with PME currently only works in a 3D geometry with tin-foil boundary conditions");
228 int nthreads = fr->nthread_ewc;
229 #pragma omp parallel for num_threads(nthreads) schedule(static)
230 for (int t = 0; t < nthreads; t++)
234 ewald_corr_thread_t &ewc_t = fr->ewc_t[t];
237 clearEwaldThreadOutput(&ewc_t);
240 /* Threading is only supported with the Verlet cut-off
241 * scheme and then only single particle forces (no
242 * exclusion forces) are calculated, so we can store
243 * the forces in the normal, single forceWithVirial->force_ array.
245 ewald_LRcorrection(md->homenr, cr, nthreads, t, *fr, *ir,
246 md->chargeA, md->chargeB,
247 (md->nChargePerturbed != 0),
249 as_rvec_array(forceWithVirial.force_.data()),
252 &ewc_t.dvdl[efptCOUL]);
254 GMX_CATCH_ALL_AND_EXIT_WITH_FATAL_ERROR;
258 reduceEwaldThreadOuput(nthreads, fr->ewc_t);
260 wallcycle_sub_stop(wcycle, ewcsEWALD_CORRECTION);
263 if (EEL_PME_EWALD(fr->ic->eeltype) && fr->n_tpi == 0)
265 /* This is not in a subcounter because it takes a
266 negligible and constant-sized amount of time */
267 ewaldOutput.Vcorr_q +=
268 ewald_charge_correction(cr, fr, lambda[efptCOUL], box,
269 &ewaldOutput.dvdl[efptCOUL],
275 /* Do reciprocal PME for Coulomb and/or LJ. */
276 assert(fr->n_tpi >= 0);
277 if (fr->n_tpi == 0 || stepWork.stateChanged)
279 int pme_flags = GMX_PME_SPREAD | GMX_PME_SOLVE;
281 if (stepWork.computeForces)
283 pme_flags |= GMX_PME_CALC_F;
285 if (stepWork.computeVirial)
287 pme_flags |= GMX_PME_CALC_ENER_VIR;
291 /* We don't calculate f, but we do want the potential */
292 pme_flags |= GMX_PME_CALC_POT;
295 /* With domain decomposition we close the CPU side load
296 * balancing region here, because PME does global
297 * communication that acts as a global barrier.
299 ddBalanceRegionHandler.closeAfterForceComputationCpu();
301 wallcycle_start(wcycle, ewcPMEMESH);
302 status = gmx_pme_do(fr->pmedata,
303 gmx::constArrayRefFromArray(coordinates.unpaddedConstArrayRef().data(), md->homenr - fr->n_tpi),
304 forceWithVirial.force_,
305 md->chargeA, md->chargeB,
306 md->sqrt_c6A, md->sqrt_c6B,
307 md->sigmaA, md->sigmaB,
309 DOMAINDECOMP(cr) ? dd_pme_maxshift_x(cr->dd) : 0,
310 DOMAINDECOMP(cr) ? dd_pme_maxshift_y(cr->dd) : 0,
312 ewaldOutput.vir_q, ewaldOutput.vir_lj,
314 lambda[efptCOUL], lambda[efptVDW],
315 &ewaldOutput.dvdl[efptCOUL],
316 &ewaldOutput.dvdl[efptVDW],
318 wallcycle_stop(wcycle, ewcPMEMESH);
321 gmx_fatal(FARGS, "Error %d in reciprocal PME routine", status);
324 /* We should try to do as little computation after
325 * this as possible, because parallel PME synchronizes
326 * the nodes, so we want all load imbalance of the
327 * rest of the force calculation to be before the PME
328 * call. DD load balancing is done on the whole time
329 * of the force call (without PME).
334 if (EVDW_PME(ir->vdwtype))
337 gmx_fatal(FARGS, "Test particle insertion not implemented with LJ-PME");
339 /* Determine the PME grid energy of the test molecule
340 * with the PME grid potential of the other charges.
342 gmx_pme_calc_energy(fr->pmedata,
343 coordinates.unpaddedConstArrayRef().subArray(md->homenr - fr->n_tpi, fr->n_tpi),
344 gmx::arrayRefFromArray(md->chargeA + md->homenr - fr->n_tpi, fr->n_tpi),
350 if (fr->ic->eeltype == eelEWALD)
352 Vlr_q = do_ewald(ir, x, as_rvec_array(forceWithVirial.force_.data()),
353 md->chargeA, md->chargeB,
355 ewaldOutput.vir_q, fr->ic->ewaldcoeff_q,
356 lambda[efptCOUL], &ewaldOutput.dvdl[efptCOUL],
360 /* Note that with separate PME nodes we get the real energies later */
361 // TODO it would be simpler if we just accumulated a single
362 // long-range virial contribution.
363 forceWithVirial.addVirialContribution(ewaldOutput.vir_q);
364 forceWithVirial.addVirialContribution(ewaldOutput.vir_lj);
365 enerd->dvdl_lin[efptCOUL] += ewaldOutput.dvdl[efptCOUL];
366 enerd->dvdl_lin[efptVDW] += ewaldOutput.dvdl[efptVDW];
367 enerd->term[F_COUL_RECIP] = Vlr_q + ewaldOutput.Vcorr_q;
368 enerd->term[F_LJ_RECIP] = Vlr_lj + ewaldOutput.Vcorr_lj;
372 fprintf(debug, "Vlr_q = %g, Vcorr_q = %g, Vlr_corr_q = %g\n",
373 Vlr_q, ewaldOutput.Vcorr_q, enerd->term[F_COUL_RECIP]);
374 pr_rvecs(debug, 0, "vir_el_recip after corr", ewaldOutput.vir_q, DIM);
375 rvec *fshift = as_rvec_array(forceOutputs->forceWithShiftForces().shiftForces().data());
376 pr_rvecs(debug, 0, "fshift after LR Corrections", fshift, SHIFTS);
377 fprintf(debug, "Vlr_lj: %g, Vcorr_lj = %g, Vlr_corr_lj = %g\n",
378 Vlr_lj, ewaldOutput.Vcorr_lj, enerd->term[F_LJ_RECIP]);
379 pr_rvecs(debug, 0, "vir_lj_recip after corr", ewaldOutput.vir_lj, DIM);
385 print_nrnb(debug, nrnb);
390 rvec *fshift = as_rvec_array(forceOutputs->forceWithShiftForces().shiftForces().data());
391 pr_rvecs(debug, 0, "fshift after bondeds", fshift, SHIFTS);