Files
lammps/src/GPU/pair_lj_cut_dipole_long_gpu.cpp
2024-02-06 00:57:46 -05:00

391 lines
14 KiB
C++

/* ----------------------------------------------------------------------
LAMMPS - Large-scale Atomic/Molecular Massively Parallel Simulator
https://www.lammps.org/, Sandia National Laboratories
LAMMPS development team: developers@lammps.org
Copyright (2003) Sandia Corporation. Under the terms of Contract
DE-AC04-94AL85000 with Sandia Corporation, the U.S. Government retains
certain rights in this software. This software is distributed under
the GNU General Public License.
See the README file in the top-level LAMMPS directory.
------------------------------------------------------------------------- */
/* ----------------------------------------------------------------------
Contributing author: Trung Dac Nguyen (Northwestern)
------------------------------------------------------------------------- */
#include "pair_lj_cut_dipole_long_gpu.h"
#include "atom.h"
#include "domain.h"
#include "error.h"
#include "ewald_const.h"
#include "force.h"
#include "gpu_extra.h"
#include "kspace.h"
#include "math_const.h"
#include "neigh_list.h"
#include "neighbor.h"
#include "suffix.h"
#include "update.h"
#include <cmath>
#include <cstring>
using namespace LAMMPS_NS;
using namespace MathConst;
using namespace EwaldConst;
// External functions from cuda library for atom decomposition
int dplj_gpu_init(const int ntypes, double **cutsq, double **host_lj1, double **host_lj2,
double **host_lj3, double **host_lj4, double **offset, double *special_lj,
const int nlocal, const int nall, const int max_nbors, const int maxspecial,
const double cell_size, int &gpu_mode, FILE *screen, double **host_cut_ljsq,
const double host_cut_coulsq, double *host_special_coul, const double qqrd2e,
const double g_ewald);
void dplj_gpu_clear();
int **dplj_gpu_compute_n(const int ago, const int inum, const int nall, double **host_x,
int *host_type, double *sublo, double *subhi, tagint *tag, int **nspecial,
tagint **special, const bool eflag, const bool vflag, const bool eatom,
const bool vatom, int &host_start, int **ilist, int **jnum,
const double cpu_time, bool &success, double *host_q, double **host_mu,
double *boxlo, double *prd);
void dplj_gpu_compute(const int ago, const int inum, const int nall, double **host_x,
int *host_type, int *ilist, int *numj, int **firstneigh, const bool eflag,
const bool vflag, const bool eatom, const bool vatom, int &host_start,
const double cpu_time, bool &success, double *host_q, double **host_mu,
const int nlocal, double *boxlo, double *prd);
double dplj_gpu_bytes();
/* ---------------------------------------------------------------------- */
PairLJCutDipoleLongGPU::PairLJCutDipoleLongGPU(LAMMPS *lmp) :
PairLJCutDipoleLong(lmp), gpu_mode(GPU_FORCE)
{
respa_enable = 0;
reinitflag = 0;
cpu_time = 0.0;
suffix_flag |= Suffix::GPU;
GPU_EXTRA::gpu_ready(lmp->modify, lmp->error);
}
/* ----------------------------------------------------------------------
free all arrays
------------------------------------------------------------------------- */
PairLJCutDipoleLongGPU::~PairLJCutDipoleLongGPU()
{
dplj_gpu_clear();
}
/* ---------------------------------------------------------------------- */
void PairLJCutDipoleLongGPU::compute(int eflag, int vflag)
{
ev_init(eflag, vflag);
int nall = atom->nlocal + atom->nghost;
int inum, host_start;
bool success = true;
int *ilist, *numneigh, **firstneigh;
if (gpu_mode != GPU_FORCE) {
double sublo[3], subhi[3];
if (domain->triclinic == 0) {
sublo[0] = domain->sublo[0];
sublo[1] = domain->sublo[1];
sublo[2] = domain->sublo[2];
subhi[0] = domain->subhi[0];
subhi[1] = domain->subhi[1];
subhi[2] = domain->subhi[2];
} else {
domain->bbox(domain->sublo_lamda, domain->subhi_lamda, sublo, subhi);
}
inum = atom->nlocal;
firstneigh = dplj_gpu_compute_n(neighbor->ago, inum, nall, atom->x, atom->type, sublo, subhi,
atom->tag, atom->nspecial, atom->special, eflag, vflag,
eflag_atom, vflag_atom, host_start, &ilist, &numneigh, cpu_time,
success, atom->q, atom->mu, domain->boxlo, domain->prd);
} else {
inum = list->inum;
ilist = list->ilist;
numneigh = list->numneigh;
firstneigh = list->firstneigh;
dplj_gpu_compute(neighbor->ago, inum, nall, atom->x, atom->type, ilist, numneigh, firstneigh,
eflag, vflag, eflag_atom, vflag_atom, host_start, cpu_time, success, atom->q,
atom->mu, atom->nlocal, domain->boxlo, domain->prd);
}
if (!success) error->one(FLERR, "Insufficient memory on accelerator");
if (atom->molecular != Atom::ATOMIC && neighbor->ago == 0)
neighbor->build_topology();
if (host_start < inum) {
cpu_time = platform::walltime();
cpu_compute(host_start, inum, eflag, vflag, ilist, numneigh, firstneigh);
cpu_time = platform::walltime() - cpu_time;
}
}
/* ----------------------------------------------------------------------
init specific to this pair style
------------------------------------------------------------------------- */
void PairLJCutDipoleLongGPU::init_style()
{
if (!atom->q_flag || !atom->mu_flag || !atom->torque_flag)
error->all(FLERR, "Pair dipole/cut/gpu requires atom attributes q, mu, torque");
if (strcmp(update->unit_style, "electron") == 0)
error->all(FLERR, "Cannot (yet) use 'electron' units with dipoles");
// Repeat cutsq calculation because done after call to init_style
double maxcut = -1.0;
double cut;
for (int i = 1; i <= atom->ntypes; i++) {
for (int j = i; j <= atom->ntypes; j++) {
if (setflag[i][j] != 0 || (setflag[i][i] != 0 && setflag[j][j] != 0)) {
cut = init_one(i, j);
cut *= cut;
if (cut > maxcut) maxcut = cut;
cutsq[i][j] = cutsq[j][i] = cut;
} else
cutsq[i][j] = cutsq[j][i] = 0.0;
}
}
double cell_size = sqrt(maxcut) + neighbor->skin;
cut_coulsq = cut_coul * cut_coul;
// ensure use of KSpace long-range solver, set g_ewald
if (force->kspace == nullptr) error->all(FLERR, "Pair style requires a KSpace style");
g_ewald = force->kspace->g_ewald;
// setup force tables
if (ncoultablebits) init_tables(cut_coul, nullptr);
int maxspecial = 0;
if (atom->molecular != Atom::ATOMIC) maxspecial = atom->maxspecial;
int mnf = 5e-2 * neighbor->oneatom;
int success =
dplj_gpu_init(atom->ntypes + 1, cutsq, lj1, lj2, lj3, lj4, offset, force->special_lj,
atom->nlocal, atom->nlocal + atom->nghost, mnf, maxspecial, cell_size, gpu_mode,
screen, cut_ljsq, cut_coulsq, force->special_coul, force->qqrd2e, g_ewald);
GPU_EXTRA::check_flag(success, error, world);
if (gpu_mode == GPU_FORCE) neighbor->add_request(this, NeighConst::REQ_FULL);
}
/* ---------------------------------------------------------------------- */
double PairLJCutDipoleLongGPU::memory_usage()
{
double bytes = Pair::memory_usage();
return bytes + dplj_gpu_bytes();
}
/* ---------------------------------------------------------------------- */
void PairLJCutDipoleLongGPU::cpu_compute(int start, int inum, int eflag, int vflag, int *ilist,
int *numneigh, int **firstneigh)
{
int i, j, ii, jj, jnum, itype, jtype;
double qtmp, xtmp, ytmp, ztmp, delx, dely, delz;
double rsq, r, rinv, r2inv, r6inv;
double forcecoulx, forcecouly, forcecoulz, fforce;
double tixcoul, tiycoul, tizcoul;
double fx, fy, fz, fdx, fdy, fdz, fax, fay, faz;
double pdotp, pidotr, pjdotr, pre1, pre2, pre3;
double grij, expm2, t, erfc;
double g0, g1, g2, b0, b1, b2, b3, d0, d1, d2, d3;
double zdix, zdiy, zdiz, zaix, zaiy, zaiz;
double g0b1_g1b2_g2b3, g0d1_g1d2_g2d3;
double forcelj, factor_coul, factor_lj, facm1;
double evdwl, ecoul;
int *jlist;
evdwl = ecoul = 0.0;
ev_init(eflag, vflag);
double **x = atom->x;
double **f = atom->f;
double *q = atom->q;
double **mu = atom->mu;
double **torque = atom->torque;
int *type = atom->type;
double *special_coul = force->special_coul;
double *special_lj = force->special_lj;
double qqrd2e = force->qqrd2e;
pre1 = 2.0 * g_ewald / MY_PIS;
pre2 = 4.0 * pow(g_ewald, 3.0) / MY_PIS;
pre3 = 8.0 * pow(g_ewald, 5.0) / MY_PIS;
// loop over neighbors of my atoms
for (ii = start; ii < inum; ii++) {
i = ilist[ii];
qtmp = q[i];
xtmp = x[i][0];
ytmp = x[i][1];
ztmp = x[i][2];
itype = type[i];
jlist = firstneigh[i];
jnum = numneigh[i];
for (jj = 0; jj < jnum; jj++) {
j = jlist[jj];
factor_lj = special_lj[sbmask(j)];
factor_coul = special_coul[sbmask(j)];
j &= NEIGHMASK;
delx = xtmp - x[j][0];
dely = ytmp - x[j][1];
delz = ztmp - x[j][2];
rsq = delx * delx + dely * dely + delz * delz;
jtype = type[j];
if (rsq < cutsq[itype][jtype]) {
r2inv = 1.0 / rsq;
rinv = sqrt(r2inv);
if (rsq < cut_coulsq) {
r = sqrt(rsq);
grij = g_ewald * r;
expm2 = exp(-grij * grij);
t = 1.0 / (1.0 + EWALD_P * grij);
erfc = t * (A1 + t * (A2 + t * (A3 + t * (A4 + t * A5)))) * expm2;
pdotp = mu[i][0] * mu[j][0] + mu[i][1] * mu[j][1] + mu[i][2] * mu[j][2];
pidotr = mu[i][0] * delx + mu[i][1] * dely + mu[i][2] * delz;
pjdotr = mu[j][0] * delx + mu[j][1] * dely + mu[j][2] * delz;
g0 = qtmp * q[j];
g1 = qtmp * pjdotr - q[j] * pidotr + pdotp;
g2 = -pidotr * pjdotr;
if (factor_coul > 0.0) {
b0 = erfc * rinv;
b1 = (b0 + pre1 * expm2) * r2inv;
b2 = (3.0 * b1 + pre2 * expm2) * r2inv;
b3 = (5.0 * b2 + pre3 * expm2) * r2inv;
g0b1_g1b2_g2b3 = g0 * b1 + g1 * b2 + g2 * b3;
fdx = delx * g0b1_g1b2_g2b3 - b1 * (qtmp * mu[j][0] - q[j] * mu[i][0]) +
b2 * (pjdotr * mu[i][0] + pidotr * mu[j][0]);
fdy = dely * g0b1_g1b2_g2b3 - b1 * (qtmp * mu[j][1] - q[j] * mu[i][1]) +
b2 * (pjdotr * mu[i][1] + pidotr * mu[j][1]);
fdz = delz * g0b1_g1b2_g2b3 - b1 * (qtmp * mu[j][2] - q[j] * mu[i][2]) +
b2 * (pjdotr * mu[i][2] + pidotr * mu[j][2]);
zdix = delx * (q[j] * b1 + b2 * pjdotr) - b1 * mu[j][0];
zdiy = dely * (q[j] * b1 + b2 * pjdotr) - b1 * mu[j][1];
zdiz = delz * (q[j] * b1 + b2 * pjdotr) - b1 * mu[j][2];
if (factor_coul < 1.0) {
fdx *= factor_coul;
fdy *= factor_coul;
fdz *= factor_coul;
zdix *= factor_coul;
zdiy *= factor_coul;
zdiz *= factor_coul;
}
} else {
fdx = fdy = fdz = 0.0;
zdix = zdiy = zdiz = 0.0;
}
if (factor_coul < 1.0) {
d0 = (erfc - 1.0) * rinv;
d1 = (d0 + pre1 * expm2) * r2inv;
d2 = (3.0 * d1 + pre2 * expm2) * r2inv;
d3 = (5.0 * d2 + pre3 * expm2) * r2inv;
g0d1_g1d2_g2d3 = g0 * d1 + g1 * d2 + g2 * d3;
fax = delx * g0d1_g1d2_g2d3 - d1 * (qtmp * mu[j][0] - q[j] * mu[i][0]) +
d2 * (pjdotr * mu[i][0] + pidotr * mu[j][0]);
fay = dely * g0d1_g1d2_g2d3 - d1 * (qtmp * mu[j][1] - q[j] * mu[i][1]) +
d2 * (pjdotr * mu[i][1] + pidotr * mu[j][1]);
faz = delz * g0d1_g1d2_g2d3 - d1 * (qtmp * mu[j][2] - q[j] * mu[i][2]) +
d2 * (pjdotr * mu[i][2] + pidotr * mu[j][2]);
zaix = delx * (q[j] * d1 + d2 * pjdotr) - d1 * mu[j][0];
zaiy = dely * (q[j] * d1 + d2 * pjdotr) - d1 * mu[j][1];
zaiz = delz * (q[j] * d1 + d2 * pjdotr) - d1 * mu[j][2];
if (factor_coul > 0.0) {
facm1 = 1.0 - factor_coul;
fax *= facm1;
fay *= facm1;
faz *= facm1;
zaix *= facm1;
zaiy *= facm1;
zaiz *= facm1;
}
} else {
fax = fay = faz = 0.0;
zaix = zaiy = zaiz = 0.0;
}
forcecoulx = fdx + fax;
forcecouly = fdy + fay;
forcecoulz = fdz + faz;
tixcoul = mu[i][1] * (zdiz + zaiz) - mu[i][2] * (zdiy + zaiy);
tiycoul = mu[i][2] * (zdix + zaix) - mu[i][0] * (zdiz + zaiz);
tizcoul = mu[i][0] * (zdiy + zaiy) - mu[i][1] * (zdix + zaix);
} else {
forcecoulx = forcecouly = forcecoulz = 0.0;
tixcoul = tiycoul = tizcoul = 0.0;
}
// LJ interaction
if (rsq < cut_ljsq[itype][jtype]) {
r6inv = r2inv * r2inv * r2inv;
forcelj = r6inv * (lj1[itype][jtype] * r6inv - lj2[itype][jtype]);
fforce = factor_lj * forcelj * r2inv;
} else
fforce = 0.0;
// total force
fx = qqrd2e * forcecoulx + delx * fforce;
fy = qqrd2e * forcecouly + dely * fforce;
fz = qqrd2e * forcecoulz + delz * fforce;
// force & torque accumulation
f[i][0] += fx;
f[i][1] += fy;
f[i][2] += fz;
torque[i][0] += qqrd2e * tixcoul;
torque[i][1] += qqrd2e * tiycoul;
torque[i][2] += qqrd2e * tizcoul;
if (eflag) {
if (rsq < cut_coulsq && factor_coul > 0.0) {
ecoul = qqrd2e * (b0 * g0 + b1 * g1 + b2 * g2);
if (factor_coul < 1.0) {
ecoul *= factor_coul;
ecoul += (1 - factor_coul) * qqrd2e * (d0 * g0 + d1 * g1 + d2 * g2);
}
} else
ecoul = 0.0;
if (rsq < cut_ljsq[itype][jtype]) {
evdwl = r6inv * (lj3[itype][jtype] * r6inv - lj4[itype][jtype]) - offset[itype][jtype];
evdwl *= factor_lj;
} else
evdwl = 0.0;
}
if (evflag) ev_tally_xyz_full(i, evdwl, ecoul, fx, fy, fz, delx, dely, delz);
}
}
}
}