Remove the rest of the device coordinates management from PME
[gromacs.git] / src / gromacs / mdlib / mdatoms.cpp
blobef4440198fa21e002195afded96aa98ed69c2d86
1 /*
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) 2012,2013,2014,2015,2016 by the GROMACS development team.
7 * Copyright (c) 2017,2018,2019,2020, by the GROMACS development team, led by
8 * Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl,
9 * and including many others, as listed in the AUTHORS file in the
10 * top-level source directory and at http://www.gromacs.org.
12 * GROMACS is free software; you can redistribute it and/or
13 * modify it under the terms of the GNU Lesser General Public License
14 * as published by the Free Software Foundation; either version 2.1
15 * of the License, or (at your option) any later version.
17 * GROMACS is distributed in the hope that it will be useful,
18 * but WITHOUT ANY WARRANTY; without even the implied warranty of
19 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
20 * Lesser General Public License for more details.
22 * You should have received a copy of the GNU Lesser General Public
23 * License along with GROMACS; if not, see
24 * http://www.gnu.org/licenses, or write to the Free Software Foundation,
25 * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
27 * If you want to redistribute modifications to GROMACS, please
28 * consider that scientific software is very special. Version
29 * control is crucial - bugs must be traceable. We will be happy to
30 * consider code for inclusion in the official distribution, but
31 * derived work must not be called official GROMACS. Details are found
32 * in the README & COPYING files - if they are missing, get the
33 * official version at http://www.gromacs.org.
35 * To help us fund GROMACS development, we humbly ask that you cite
36 * the research papers on the package. Check out http://www.gromacs.org.
38 #include "gmxpre.h"
40 #include "mdatoms.h"
42 #include <cmath>
44 #include <memory>
46 #include "gromacs/ewald/pme.h"
47 #include "gromacs/gpu_utils/hostallocator.h"
48 #include "gromacs/math/functions.h"
49 #include "gromacs/mdlib/gmx_omp_nthreads.h"
50 #include "gromacs/mdlib/qmmm.h"
51 #include "gromacs/mdtypes/inputrec.h"
52 #include "gromacs/mdtypes/md_enums.h"
53 #include "gromacs/topology/mtop_lookup.h"
54 #include "gromacs/topology/mtop_util.h"
55 #include "gromacs/topology/topology.h"
56 #include "gromacs/utility/exceptions.h"
57 #include "gromacs/utility/smalloc.h"
59 #define ALMOST_ZERO 1e-30
61 namespace gmx
64 MDAtoms::MDAtoms() : mdatoms_(nullptr) {}
66 MDAtoms::~MDAtoms()
68 if (mdatoms_ == nullptr)
70 return;
72 sfree(mdatoms_->massA);
73 sfree(mdatoms_->massB);
74 sfree(mdatoms_->massT);
75 gmx::AlignedAllocationPolicy::free(mdatoms_->invmass);
76 sfree(mdatoms_->invMassPerDim);
77 sfree(mdatoms_->typeA);
78 sfree(mdatoms_->chargeB);
79 sfree(mdatoms_->typeB);
80 sfree(mdatoms_->sqrt_c6A);
81 sfree(mdatoms_->sigmaA);
82 sfree(mdatoms_->sigma3A);
83 sfree(mdatoms_->sqrt_c6B);
84 sfree(mdatoms_->sigmaB);
85 sfree(mdatoms_->sigma3B);
86 sfree(mdatoms_->ptype);
87 sfree(mdatoms_->cTC);
88 sfree(mdatoms_->cENER);
89 sfree(mdatoms_->cACC);
90 sfree(mdatoms_->cFREEZE);
91 sfree(mdatoms_->cVCM);
92 sfree(mdatoms_->cORF);
93 sfree(mdatoms_->bPerturbed);
94 sfree(mdatoms_->cU1);
95 sfree(mdatoms_->cU2);
96 sfree(mdatoms_->bQM);
99 void MDAtoms::resize(int newSize)
101 chargeA_.resizeWithPadding(newSize);
102 mdatoms_->chargeA = chargeA_.data();
105 void MDAtoms::reserve(int newCapacity)
107 chargeA_.reserveWithPadding(newCapacity);
108 mdatoms_->chargeA = chargeA_.data();
111 std::unique_ptr<MDAtoms> makeMDAtoms(FILE* fp, const gmx_mtop_t& mtop, const t_inputrec& ir, const bool rankHasPmeGpuTask)
113 auto mdAtoms = std::make_unique<MDAtoms>();
114 // GPU transfers may want to use a suitable pinning mode.
115 if (rankHasPmeGpuTask)
117 changePinningPolicy(&mdAtoms->chargeA_, pme_get_pinning_policy());
119 t_mdatoms* md;
120 snew(md, 1);
121 mdAtoms->mdatoms_.reset(md);
123 md->nenergrp = mtop.groups.groups[SimulationAtomGroupType::EnergyOutput].size();
124 md->bVCMgrps = FALSE;
125 for (int i = 0; i < mtop.natoms; i++)
127 if (getGroupType(mtop.groups, SimulationAtomGroupType::MassCenterVelocityRemoval, i) > 0)
129 md->bVCMgrps = TRUE;
133 /* Determine the total system mass and perturbed atom counts */
134 double totalMassA = 0.0;
135 double totalMassB = 0.0;
137 md->haveVsites = FALSE;
138 gmx_mtop_atomloop_block_t aloop = gmx_mtop_atomloop_block_init(&mtop);
139 const t_atom* atom;
140 int nmol;
141 while (gmx_mtop_atomloop_block_next(aloop, &atom, &nmol))
143 totalMassA += nmol * atom->m;
144 totalMassB += nmol * atom->mB;
146 if (atom->ptype == eptVSite)
148 md->haveVsites = TRUE;
151 if (ir.efep != efepNO && PERTURBED(*atom))
153 md->nPerturbed++;
154 if (atom->mB != atom->m)
156 md->nMassPerturbed += nmol;
158 if (atom->qB != atom->q)
160 md->nChargePerturbed += nmol;
162 if (atom->typeB != atom->type)
164 md->nTypePerturbed += nmol;
169 md->tmassA = totalMassA;
170 md->tmassB = totalMassB;
172 if (ir.efep != efepNO && fp)
174 fprintf(fp, "There are %d atoms and %d charges for free energy perturbation\n",
175 md->nPerturbed, md->nChargePerturbed);
178 md->havePartiallyFrozenAtoms = FALSE;
179 for (int g = 0; g < ir.opts.ngfrz; g++)
181 for (int d = YY; d < DIM; d++)
183 if (ir.opts.nFreeze[g][d] != ir.opts.nFreeze[g][XX])
185 md->havePartiallyFrozenAtoms = TRUE;
190 md->bOrires = (gmx_mtop_ftype_count(&mtop, F_ORIRES) != 0);
192 return mdAtoms;
195 } // namespace gmx
197 void atoms2md(const gmx_mtop_t* mtop, const t_inputrec* ir, int nindex, const int* index, int homenr, gmx::MDAtoms* mdAtoms)
199 gmx_bool bLJPME;
200 const t_grpopts* opts;
201 int nthreads gmx_unused;
203 bLJPME = EVDW_PME(ir->vdwtype);
205 opts = &ir->opts;
207 const SimulationGroups& groups = mtop->groups;
209 auto md = mdAtoms->mdatoms();
210 /* nindex>=0 indicates DD where we use an index */
211 if (nindex >= 0)
213 md->nr = nindex;
215 else
217 md->nr = mtop->natoms;
220 if (md->nr > md->nalloc)
222 md->nalloc = over_alloc_dd(md->nr);
224 if (md->nMassPerturbed)
226 srenew(md->massA, md->nalloc);
227 srenew(md->massB, md->nalloc);
229 srenew(md->massT, md->nalloc);
230 /* The SIMD version of the integrator needs this aligned and padded.
231 * The padding needs to be with zeros, which we set later below.
233 gmx::AlignedAllocationPolicy::free(md->invmass);
234 md->invmass = new (gmx::AlignedAllocationPolicy::malloc(
235 (md->nalloc + GMX_REAL_MAX_SIMD_WIDTH) * sizeof(*md->invmass))) real;
236 srenew(md->invMassPerDim, md->nalloc);
237 // TODO eventually we will have vectors and just resize
238 // everything, but for now the semantics of md->nalloc being
239 // the capacity are preserved by keeping vectors within
240 // mdAtoms having the same properties as the other arrays.
241 mdAtoms->reserve(md->nalloc);
242 mdAtoms->resize(md->nr);
243 srenew(md->typeA, md->nalloc);
244 if (md->nPerturbed)
246 srenew(md->chargeB, md->nalloc);
247 srenew(md->typeB, md->nalloc);
249 if (bLJPME)
251 srenew(md->sqrt_c6A, md->nalloc);
252 srenew(md->sigmaA, md->nalloc);
253 srenew(md->sigma3A, md->nalloc);
254 if (md->nPerturbed)
256 srenew(md->sqrt_c6B, md->nalloc);
257 srenew(md->sigmaB, md->nalloc);
258 srenew(md->sigma3B, md->nalloc);
261 srenew(md->ptype, md->nalloc);
262 if (opts->ngtc > 1)
264 srenew(md->cTC, md->nalloc);
265 /* We always copy cTC with domain decomposition */
267 srenew(md->cENER, md->nalloc);
268 if (opts->ngacc > 1)
270 srenew(md->cACC, md->nalloc);
272 if (opts->nFreeze
273 && (opts->ngfrz > 1 || opts->nFreeze[0][XX] || opts->nFreeze[0][YY] || opts->nFreeze[0][ZZ]))
275 srenew(md->cFREEZE, md->nalloc);
277 if (md->bVCMgrps)
279 srenew(md->cVCM, md->nalloc);
281 if (md->bOrires)
283 srenew(md->cORF, md->nalloc);
285 if (md->nPerturbed)
287 srenew(md->bPerturbed, md->nalloc);
290 /* Note that these user t_mdatoms array pointers are NULL
291 * when there is only one group present.
292 * Therefore, when adding code, the user should use something like:
293 * gprnrU1 = (md->cU1==NULL ? 0 : md->cU1[localatindex])
295 if (!mtop->groups.groupNumbers[SimulationAtomGroupType::User1].empty())
297 srenew(md->cU1, md->nalloc);
299 if (!mtop->groups.groupNumbers[SimulationAtomGroupType::User2].empty())
301 srenew(md->cU2, md->nalloc);
304 if (ir->bQMMM)
306 srenew(md->bQM, md->nalloc);
310 int molb = 0;
312 nthreads = gmx_omp_nthreads_get(emntDefault);
313 #pragma omp parallel for num_threads(nthreads) schedule(static) firstprivate(molb)
314 for (int i = 0; i < md->nr; i++)
318 int g, ag;
319 real mA, mB, fac;
320 real c6, c12;
322 if (index == nullptr)
324 ag = i;
326 else
328 ag = index[i];
330 const t_atom& atom = mtopGetAtomParameters(mtop, ag, &molb);
332 if (md->cFREEZE)
334 md->cFREEZE[i] = getGroupType(groups, SimulationAtomGroupType::Freeze, ag);
336 if (EI_ENERGY_MINIMIZATION(ir->eI))
338 /* Displacement is proportional to F, masses used for constraints */
339 mA = 1.0;
340 mB = 1.0;
342 else if (ir->eI == eiBD)
344 /* With BD the physical masses are irrelevant.
345 * To keep the code simple we use most of the normal MD code path
346 * for BD. Thus for constraining the masses should be proportional
347 * to the friction coefficient. We set the absolute value such that
348 * m/2<(dx/dt)^2> = m/2*2kT/fric*dt = kT/2 => m=fric*dt/2
349 * Then if we set the (meaningless) velocity to v=dx/dt, we get the
350 * correct kinetic energy and temperature using the usual code path.
351 * Thus with BD v*dt will give the displacement and the reported
352 * temperature can signal bad integration (too large time step).
354 if (ir->bd_fric > 0)
356 mA = 0.5 * ir->bd_fric * ir->delta_t;
357 mB = 0.5 * ir->bd_fric * ir->delta_t;
359 else
361 /* The friction coefficient is mass/tau_t */
362 fac = ir->delta_t
363 / opts->tau_t[md->cTC ? groups.groupNumbers[SimulationAtomGroupType::TemperatureCoupling][ag] : 0];
364 mA = 0.5 * atom.m * fac;
365 mB = 0.5 * atom.mB * fac;
368 else
370 mA = atom.m;
371 mB = atom.mB;
373 if (md->nMassPerturbed)
375 md->massA[i] = mA;
376 md->massB[i] = mB;
378 md->massT[i] = mA;
380 if (mA == 0.0)
382 md->invmass[i] = 0;
383 md->invMassPerDim[i][XX] = 0;
384 md->invMassPerDim[i][YY] = 0;
385 md->invMassPerDim[i][ZZ] = 0;
387 else if (md->cFREEZE)
389 g = md->cFREEZE[i];
390 GMX_ASSERT(opts->nFreeze != nullptr,
391 "Must have freeze groups to initialize masses");
392 if (opts->nFreeze[g][XX] && opts->nFreeze[g][YY] && opts->nFreeze[g][ZZ])
394 /* Set the mass of completely frozen particles to ALMOST_ZERO
395 * iso 0 to avoid div by zero in lincs or shake.
397 md->invmass[i] = ALMOST_ZERO;
399 else
401 /* Note: Partially frozen particles use the normal invmass.
402 * If such particles are constrained, the frozen dimensions
403 * should not be updated with the constrained coordinates.
405 md->invmass[i] = 1.0 / mA;
407 for (int d = 0; d < DIM; d++)
409 md->invMassPerDim[i][d] = (opts->nFreeze[g][d] ? 0 : 1.0 / mA);
412 else
414 md->invmass[i] = 1.0 / mA;
415 for (int d = 0; d < DIM; d++)
417 md->invMassPerDim[i][d] = 1.0 / mA;
421 md->chargeA[i] = atom.q;
422 md->typeA[i] = atom.type;
423 if (bLJPME)
425 c6 = mtop->ffparams.iparams[atom.type * (mtop->ffparams.atnr + 1)].lj.c6;
426 c12 = mtop->ffparams.iparams[atom.type * (mtop->ffparams.atnr + 1)].lj.c12;
427 md->sqrt_c6A[i] = sqrt(c6);
428 if (c6 == 0.0 || c12 == 0)
430 md->sigmaA[i] = 1.0;
432 else
434 md->sigmaA[i] = gmx::sixthroot(c12 / c6);
436 md->sigma3A[i] = 1 / (md->sigmaA[i] * md->sigmaA[i] * md->sigmaA[i]);
438 if (md->nPerturbed)
440 md->bPerturbed[i] = PERTURBED(atom);
441 md->chargeB[i] = atom.qB;
442 md->typeB[i] = atom.typeB;
443 if (bLJPME)
445 c6 = mtop->ffparams.iparams[atom.typeB * (mtop->ffparams.atnr + 1)].lj.c6;
446 c12 = mtop->ffparams.iparams[atom.typeB * (mtop->ffparams.atnr + 1)].lj.c12;
447 md->sqrt_c6B[i] = sqrt(c6);
448 if (c6 == 0.0 || c12 == 0)
450 md->sigmaB[i] = 1.0;
452 else
454 md->sigmaB[i] = gmx::sixthroot(c12 / c6);
456 md->sigma3B[i] = 1 / (md->sigmaB[i] * md->sigmaB[i] * md->sigmaB[i]);
459 md->ptype[i] = atom.ptype;
460 if (md->cTC)
462 md->cTC[i] = groups.groupNumbers[SimulationAtomGroupType::TemperatureCoupling][ag];
464 md->cENER[i] = getGroupType(groups, SimulationAtomGroupType::EnergyOutput, ag);
465 if (md->cACC)
467 md->cACC[i] = groups.groupNumbers[SimulationAtomGroupType::Acceleration][ag];
469 if (md->cVCM)
471 md->cVCM[i] = groups.groupNumbers[SimulationAtomGroupType::MassCenterVelocityRemoval][ag];
473 if (md->cORF)
475 md->cORF[i] = getGroupType(groups, SimulationAtomGroupType::OrientationRestraintsFit, ag);
478 if (md->cU1)
480 md->cU1[i] = groups.groupNumbers[SimulationAtomGroupType::User1][ag];
482 if (md->cU2)
484 md->cU2[i] = groups.groupNumbers[SimulationAtomGroupType::User2][ag];
487 if (ir->bQMMM)
489 if (groups.groupNumbers[SimulationAtomGroupType::QuantumMechanics].empty()
490 || groups.groupNumbers[SimulationAtomGroupType::QuantumMechanics][ag]
491 < groups.groups[SimulationAtomGroupType::QuantumMechanics].size() - 1)
493 md->bQM[i] = TRUE;
495 else
497 md->bQM[i] = FALSE;
501 GMX_CATCH_ALL_AND_EXIT_WITH_FATAL_ERROR
504 if (md->nr > 0)
506 /* Pad invmass with 0 so a SIMD MD update does not change v and x */
507 for (int i = md->nr; i < md->nr + GMX_REAL_MAX_SIMD_WIDTH; i++)
509 md->invmass[i] = 0;
513 md->homenr = homenr;
514 /* We set mass, invmass, invMassPerDim and tmass for lambda=0.
515 * For free-energy runs, these should be updated using update_mdatoms().
517 md->tmass = md->tmassA;
518 md->lambda = 0;
521 void update_mdatoms(t_mdatoms* md, real lambda)
523 if (md->nMassPerturbed && lambda != md->lambda)
525 real L1 = 1 - lambda;
527 /* Update masses of perturbed atoms for the change in lambda */
528 int gmx_unused nthreads = gmx_omp_nthreads_get(emntDefault);
529 #pragma omp parallel for num_threads(nthreads) schedule(static)
530 for (int i = 0; i < md->nr; i++)
532 if (md->bPerturbed[i])
534 md->massT[i] = L1 * md->massA[i] + lambda * md->massB[i];
535 /* Atoms with invmass 0 or ALMOST_ZERO are massless or frozen
536 * and their invmass does not depend on lambda.
538 if (md->invmass[i] > 1.1 * ALMOST_ZERO)
540 md->invmass[i] = 1.0 / md->massT[i];
541 for (int d = 0; d < DIM; d++)
543 if (md->invMassPerDim[i][d] > 1.1 * ALMOST_ZERO)
545 md->invMassPerDim[i][d] = md->invmass[i];
552 /* Update the system mass for the change in lambda */
553 md->tmass = L1 * md->tmassA + lambda * md->tmassB;
556 md->lambda = lambda;