changes to start to make atom style hybrid work

This commit is contained in:
Steve Plimpton
2019-11-26 18:34:51 -07:00
parent 3736af0aaf
commit 6c18e366d7
24 changed files with 586 additions and 1432 deletions

View File

@ -35,6 +35,15 @@ AtomVecHybrid::~AtomVecHybrid()
delete [] styles;
for (int k = 0; k < nstyles; k++) delete [] keywords[k];
delete [] keywords;
// these strings will be concatenated from sub-style strings
// fields_data_atom must start with fields common to all styles
fields_grow = fields_copy = fields_comm = fields_comm_vel = NULL;
fields_reverse = fields_border = fields_border_vel = NULL;
fields_exchange = fields_restart = fields_create = NULL;
fields_data_atom = (char *) "id type x";
fields_data_vel = NULL;
}
/* ----------------------------------------------------------------------
@ -81,51 +90,49 @@ void AtomVecHybrid::process_args(int narg, char **arg)
for (int i = 0; i < nallstyles; i++) delete [] allstyles[i];
delete [] allstyles;
// hybrid settings are MAX or MIN of sub-style settings
// hybrid sizes are minimal values plus extra values for each sub-style
// concatenate field strings from all sub-styles
molecular = 0;
comm_x_only = comm_f_only = 1;
concatenate_fields();
size_forward = 3;
size_reverse = 3;
size_border = 6;
size_data_atom = 5;
size_data_vel = 4;
xcol_data = 3;
maxexchange = 0;
// parent AtomVec will now operate on concatenated fields
setup_fields();
}
/* ---------------------------------------------------------------------- */
void AtomVecHybrid::concatenate_fields()
{
for (int k = 0; k < nstyles; k++) {
if ((styles[k]->molecular == 1 && molecular == 2) ||
(styles[k]->molecular == 2 && molecular == 1))
error->all(FLERR,"Cannot mix molecular and molecule template "
"atom styles");
molecular = MAX(molecular,styles[k]->molecular);
bonds_allow = MAX(bonds_allow,styles[k]->bonds_allow);
angles_allow = MAX(angles_allow,styles[k]->angles_allow);
dihedrals_allow = MAX(dihedrals_allow,styles[k]->dihedrals_allow);
impropers_allow = MAX(impropers_allow,styles[k]->impropers_allow);
mass_type = MAX(mass_type,styles[k]->mass_type);
dipole_type = MAX(dipole_type,styles[k]->dipole_type);
forceclearflag = MAX(forceclearflag,styles[k]->forceclearflag);
if (styles[k]->molecular == 2) onemols = styles[k]->onemols;
comm_x_only = MIN(comm_x_only,styles[k]->comm_x_only);
comm_f_only = MIN(comm_f_only,styles[k]->comm_f_only);
size_forward += styles[k]->size_forward - 3;
size_reverse += styles[k]->size_reverse - 3;
size_border += styles[k]->size_border - 6;
size_data_atom += styles[k]->size_data_atom - 5;
size_data_vel += styles[k]->size_data_vel - 4;
maxexchange += styles[k]->maxexchange;
concatenate(fields_grow,styles[k]->fields_grow);
concatenate(fields_copy,styles[k]->fields_copy);
concatenate(fields_comm,styles[k]->fields_comm);
concatenate(fields_comm_vel,styles[k]->fields_comm_vel);
concatenate(fields_reverse,styles[k]->fields_reverse);
concatenate(fields_border,styles[k]->fields_border);
concatenate(fields_border_vel,styles[k]->fields_border_vel);
concatenate(fields_exchange,styles[k]->fields_exchange);
concatenate(fields_restart,styles[k]->fields_restart);
concatenate(fields_create,styles[k]->fields_create);
concatenate(fields_data_atom,styles[k]->fields_data_atom);
concatenate(fields_data_vel,styles[k]->fields_data_vel);
}
}
size_velocity = 3;
if (atom->omega_flag) size_velocity += 3;
if (atom->angmom_flag) size_velocity += 3;
/* ---------------------------------------------------------------------- */
void AtomVecHybrid::concatenate(char *&root, char *add)
{
/*
char **rootwords,**addwords;
int nroot = parse(root,rootwords);
int nadd = parse(add,addwords);
for (int iadd = 0; iadd < nadd; iadd++) {
if (check(addwords[iadd],nroot,rootwords)) continue;
addone(addwords[iadd],nroot,rootwords);
}
*/
}
/* ---------------------------------------------------------------------- */
@ -136,76 +143,6 @@ void AtomVecHybrid::init()
for (int k = 0; k < nstyles; k++) styles[k]->init();
}
/* ----------------------------------------------------------------------
grow atom arrays
n = 0 grows arrays by a chunk
n > 0 allocates arrays to size n
------------------------------------------------------------------------- */
void AtomVecHybrid::grow(int n)
{
if (n == 0) grow_nmax();
else nmax = n;
atom->nmax = nmax;
if (nmax < 0 || nmax > MAXSMALLINT)
error->one(FLERR,"Per-processor system is too big");
// sub-styles perform all reallocation
// turn off nextra_grow so hybrid can do that once below
int tmp = atom->nextra_grow;
atom->nextra_grow = 0;
for (int k = 0; k < nstyles; k++) styles[k]->grow(nmax);
atom->nextra_grow = tmp;
// insure hybrid local ptrs and sub-style ptrs are up to date
// for sub-styles, do this in case
// multiple sub-style reallocs of same array occurred
grow_reset();
if (atom->nextra_grow)
for (int iextra = 0; iextra < atom->nextra_grow; iextra++)
modify->fix[atom->extra_grow[iextra]]->grow_arrays(nmax);
}
/* ----------------------------------------------------------------------
reset local array ptrs
------------------------------------------------------------------------- */
void AtomVecHybrid::grow_reset()
{
tag = atom->tag; type = atom->type;
mask = atom->mask; image = atom->image;
x = atom->x; v = atom->v; f = atom->f;
omega = atom->omega; angmom = atom->angmom;
for (int k = 0; k < nstyles; k++) styles[k]->grow_reset();
}
/* ----------------------------------------------------------------------
copy atom I info to atom J for all sub-styles
------------------------------------------------------------------------- */
void AtomVecHybrid::copy(int i, int j, int delflag)
{
int tmp = atom->nextra_grow;
atom->nextra_grow = 0;
for (int k = 0; k < nstyles; k++) styles[k]->copy(i,j,delflag);
atom->nextra_grow = tmp;
if (atom->nextra_grow)
for (int iextra = 0; iextra < atom->nextra_grow; iextra++)
modify->fix[atom->extra_grow[iextra]]->copy_arrays(i,j,delflag);
}
/* ---------------------------------------------------------------------- */
void AtomVecHybrid::clear_bonus()
{
for (int k = 0; k < nstyles; k++) styles[k]->clear_bonus();
}
/* ---------------------------------------------------------------------- */
void AtomVecHybrid::force_clear(int n, size_t nbytes)
@ -214,799 +151,6 @@ void AtomVecHybrid::force_clear(int n, size_t nbytes)
if (styles[k]->forceclearflag) styles[k]->force_clear(n,nbytes);
}
/* ---------------------------------------------------------------------- */
int AtomVecHybrid::pack_comm(int n, int *list, double *buf,
int pbc_flag, int *pbc)
{
int i,j,k,m;
double dx,dy,dz;
m = 0;
if (pbc_flag == 0) {
for (i = 0; i < n; i++) {
j = list[i];
buf[m++] = x[j][0];
buf[m++] = x[j][1];
buf[m++] = x[j][2];
}
} else {
if (domain->triclinic == 0) {
dx = pbc[0]*domain->xprd;
dy = pbc[1]*domain->yprd;
dz = pbc[2]*domain->zprd;
} else {
dx = pbc[0]*domain->xprd + pbc[5]*domain->xy + pbc[4]*domain->xz;
dy = pbc[1]*domain->yprd + pbc[3]*domain->yz;
dz = pbc[2]*domain->zprd;
}
for (i = 0; i < n; i++) {
j = list[i];
buf[m++] = x[j][0] + dx;
buf[m++] = x[j][1] + dy;
buf[m++] = x[j][2] + dz;
}
}
// pack sub-style contributions as contiguous chunks
for (k = 0; k < nstyles; k++)
m += styles[k]->pack_comm_hybrid(n,list,&buf[m]);
return m;
}
/* ---------------------------------------------------------------------- */
int AtomVecHybrid::pack_comm_vel(int n, int *list, double *buf,
int pbc_flag, int *pbc)
{
int i,j,k,m;
double dx,dy,dz,dvx,dvy,dvz;
int omega_flag = atom->omega_flag;
int angmom_flag = atom->angmom_flag;
m = 0;
if (pbc_flag == 0) {
for (i = 0; i < n; i++) {
j = list[i];
buf[m++] = x[j][0];
buf[m++] = x[j][1];
buf[m++] = x[j][2];
buf[m++] = v[j][0];
buf[m++] = v[j][1];
buf[m++] = v[j][2];
if (omega_flag) {
buf[m++] = omega[j][0];
buf[m++] = omega[j][1];
buf[m++] = omega[j][2];
}
if (angmom_flag) {
buf[m++] = angmom[j][0];
buf[m++] = angmom[j][1];
buf[m++] = angmom[j][2];
}
}
} else {
if (domain->triclinic == 0) {
dx = pbc[0]*domain->xprd;
dy = pbc[1]*domain->yprd;
dz = pbc[2]*domain->zprd;
} else {
dx = pbc[0]*domain->xprd + pbc[5]*domain->xy + pbc[4]*domain->xz;
dy = pbc[1]*domain->yprd + pbc[3]*domain->yz;
dz = pbc[2]*domain->zprd;
}
if (!deform_vremap) {
for (i = 0; i < n; i++) {
j = list[i];
buf[m++] = x[j][0] + dx;
buf[m++] = x[j][1] + dy;
buf[m++] = x[j][2] + dz;
buf[m++] = v[j][0];
buf[m++] = v[j][1];
buf[m++] = v[j][2];
if (omega_flag) {
buf[m++] = omega[j][0];
buf[m++] = omega[j][1];
buf[m++] = omega[j][2];
}
if (angmom_flag) {
buf[m++] = angmom[j][0];
buf[m++] = angmom[j][1];
buf[m++] = angmom[j][2];
}
}
} else {
dvx = pbc[0]*h_rate[0] + pbc[5]*h_rate[5] + pbc[4]*h_rate[4];
dvy = pbc[1]*h_rate[1] + pbc[3]*h_rate[3];
dvz = pbc[2]*h_rate[2];
for (i = 0; i < n; i++) {
j = list[i];
buf[m++] = x[j][0] + dx;
buf[m++] = x[j][1] + dy;
buf[m++] = x[j][2] + dz;
if (mask[i] & deform_groupbit) {
buf[m++] = v[j][0] + dvx;
buf[m++] = v[j][1] + dvy;
buf[m++] = v[j][2] + dvz;
} else {
buf[m++] = v[j][0];
buf[m++] = v[j][1];
buf[m++] = v[j][2];
}
if (omega_flag) {
buf[m++] = omega[j][0];
buf[m++] = omega[j][1];
buf[m++] = omega[j][2];
}
if (angmom_flag) {
buf[m++] = angmom[j][0];
buf[m++] = angmom[j][1];
buf[m++] = angmom[j][2];
}
}
}
}
// pack sub-style contributions as contiguous chunks
for (k = 0; k < nstyles; k++)
m += styles[k]->pack_comm_hybrid(n,list,&buf[m]);
return m;
}
/* ---------------------------------------------------------------------- */
void AtomVecHybrid::unpack_comm(int n, int first, double *buf)
{
int i,k,m,last;
m = 0;
last = first + n;
for (i = first; i < last; i++) {
x[i][0] = buf[m++];
x[i][1] = buf[m++];
x[i][2] = buf[m++];
}
// unpack sub-style contributions as contiguous chunks
for (k = 0; k < nstyles; k++)
m += styles[k]->unpack_comm_hybrid(n,first,&buf[m]);
}
/* ---------------------------------------------------------------------- */
void AtomVecHybrid::unpack_comm_vel(int n, int first, double *buf)
{
int i,k,m,last;
int omega_flag = atom->omega_flag;
int angmom_flag = atom->angmom_flag;
m = 0;
last = first + n;
for (i = first; i < last; i++) {
x[i][0] = buf[m++];
x[i][1] = buf[m++];
x[i][2] = buf[m++];
v[i][0] = buf[m++];
v[i][1] = buf[m++];
v[i][2] = buf[m++];
if (omega_flag) {
omega[i][0] = buf[m++];
omega[i][1] = buf[m++];
omega[i][2] = buf[m++];
}
if (angmom_flag) {
angmom[i][0] = buf[m++];
angmom[i][1] = buf[m++];
angmom[i][2] = buf[m++];
}
}
// unpack sub-style contributions as contiguous chunks
for (k = 0; k < nstyles; k++)
m += styles[k]->unpack_comm_hybrid(n,first,&buf[m]);
}
/* ---------------------------------------------------------------------- */
int AtomVecHybrid::pack_reverse(int n, int first, double *buf)
{
int i,k,m,last;
m = 0;
last = first + n;
for (i = first; i < last; i++) {
buf[m++] = f[i][0];
buf[m++] = f[i][1];
buf[m++] = f[i][2];
}
// pack sub-style contributions as contiguous chunks
for (k = 0; k < nstyles; k++)
m += styles[k]->pack_reverse_hybrid(n,first,&buf[m]);
return m;
}
/* ---------------------------------------------------------------------- */
void AtomVecHybrid::unpack_reverse(int n, int *list, double *buf)
{
int i,j,k,m;
m = 0;
for (i = 0; i < n; i++) {
j = list[i];
f[j][0] += buf[m++];
f[j][1] += buf[m++];
f[j][2] += buf[m++];
}
// unpack sub-style contributions as contiguous chunks
for (k = 0; k < nstyles; k++)
m += styles[k]->unpack_reverse_hybrid(n,list,&buf[m]);
}
/* ---------------------------------------------------------------------- */
int AtomVecHybrid::pack_border(int n, int *list, double *buf,
int pbc_flag, int *pbc)
{
int i,j,k,m;
double dx,dy,dz;
m = 0;
if (pbc_flag == 0) {
for (i = 0; i < n; i++) {
j = list[i];
buf[m++] = x[j][0];
buf[m++] = x[j][1];
buf[m++] = x[j][2];
buf[m++] = ubuf(tag[j]).d;
buf[m++] = ubuf(type[j]).d;
buf[m++] = ubuf(mask[j]).d;
}
} else {
if (domain->triclinic == 0) {
dx = pbc[0]*domain->xprd;
dy = pbc[1]*domain->yprd;
dz = pbc[2]*domain->zprd;
} else {
dx = pbc[0];
dy = pbc[1];
dz = pbc[2];
}
for (i = 0; i < n; i++) {
j = list[i];
buf[m++] = x[j][0] + dx;
buf[m++] = x[j][1] + dy;
buf[m++] = x[j][2] + dz;
buf[m++] = ubuf(tag[j]).d;
buf[m++] = ubuf(type[j]).d;
buf[m++] = ubuf(mask[j]).d;
}
}
// pack sub-style contributions as contiguous chunks
for (k = 0; k < nstyles; k++)
m += styles[k]->pack_border_hybrid(n,list,&buf[m]);
if (atom->nextra_border)
for (int iextra = 0; iextra < atom->nextra_border; iextra++)
m += modify->fix[atom->extra_border[iextra]]->pack_border(n,list,&buf[m]);
return m;
}
/* ---------------------------------------------------------------------- */
int AtomVecHybrid::pack_border_vel(int n, int *list, double *buf,
int pbc_flag, int *pbc)
{
int i,j,k,m;
double dx,dy,dz,dvx,dvy,dvz;
int omega_flag = atom->omega_flag;
int angmom_flag = atom->angmom_flag;
m = 0;
if (pbc_flag == 0) {
for (i = 0; i < n; i++) {
j = list[i];
buf[m++] = x[j][0];
buf[m++] = x[j][1];
buf[m++] = x[j][2];
buf[m++] = ubuf(tag[j]).d;
buf[m++] = ubuf(type[j]).d;
buf[m++] = ubuf(mask[j]).d;
buf[m++] = v[j][0];
buf[m++] = v[j][1];
buf[m++] = v[j][2];
if (omega_flag) {
buf[m++] = omega[j][0];
buf[m++] = omega[j][1];
buf[m++] = omega[j][2];
}
if (angmom_flag) {
buf[m++] = angmom[j][0];
buf[m++] = angmom[j][1];
buf[m++] = angmom[j][2];
}
}
} else {
if (domain->triclinic == 0) {
dx = pbc[0]*domain->xprd;
dy = pbc[1]*domain->yprd;
dz = pbc[2]*domain->zprd;
} else {
dx = pbc[0];
dy = pbc[1];
dz = pbc[2];
}
if (!deform_vremap) {
for (i = 0; i < n; i++) {
j = list[i];
buf[m++] = x[j][0] + dx;
buf[m++] = x[j][1] + dy;
buf[m++] = x[j][2] + dz;
buf[m++] = ubuf(tag[j]).d;
buf[m++] = ubuf(type[j]).d;
buf[m++] = ubuf(mask[j]).d;
buf[m++] = v[j][0];
buf[m++] = v[j][1];
buf[m++] = v[j][2];
if (omega_flag) {
buf[m++] = omega[j][0];
buf[m++] = omega[j][1];
buf[m++] = omega[j][2];
}
if (angmom_flag) {
buf[m++] = angmom[j][0];
buf[m++] = angmom[j][1];
buf[m++] = angmom[j][2];
}
}
} else {
dvx = pbc[0]*h_rate[0] + pbc[5]*h_rate[5] + pbc[4]*h_rate[4];
dvy = pbc[1]*h_rate[1] + pbc[3]*h_rate[3];
dvz = pbc[2]*h_rate[2];
for (i = 0; i < n; i++) {
j = list[i];
buf[m++] = x[j][0] + dx;
buf[m++] = x[j][1] + dy;
buf[m++] = x[j][2] + dz;
buf[m++] = ubuf(tag[j]).d;
buf[m++] = ubuf(type[j]).d;
buf[m++] = ubuf(mask[j]).d;
if (mask[i] & deform_groupbit) {
buf[m++] = v[j][0] + dvx;
buf[m++] = v[j][1] + dvy;
buf[m++] = v[j][2] + dvz;
} else {
buf[m++] = v[j][0];
buf[m++] = v[j][1];
buf[m++] = v[j][2];
}
if (omega_flag) {
buf[m++] = omega[j][0];
buf[m++] = omega[j][1];
buf[m++] = omega[j][2];
}
if (angmom_flag) {
buf[m++] = angmom[j][0];
buf[m++] = angmom[j][1];
buf[m++] = angmom[j][2];
}
}
}
}
// pack sub-style contributions as contiguous chunks
for (k = 0; k < nstyles; k++)
m += styles[k]->pack_border_hybrid(n,list,&buf[m]);
if (atom->nextra_border)
for (int iextra = 0; iextra < atom->nextra_border; iextra++)
m += modify->fix[atom->extra_border[iextra]]->pack_border(n,list,&buf[m]);
return m;
}
/* ---------------------------------------------------------------------- */
void AtomVecHybrid::unpack_border(int n, int first, double *buf)
{
int i,k,m,last;
m = 0;
last = first + n;
for (i = first; i < last; i++) {
if (i == nmax) grow(0);
x[i][0] = buf[m++];
x[i][1] = buf[m++];
x[i][2] = buf[m++];
tag[i] = (tagint) ubuf(buf[m++]).i;
type[i] = (int) ubuf(buf[m++]).i;
mask[i] = (int) ubuf(buf[m++]).i;
}
// unpack sub-style contributions as contiguous chunks
for (k = 0; k < nstyles; k++)
m += styles[k]->unpack_border_hybrid(n,first,&buf[m]);
if (atom->nextra_border)
for (int iextra = 0; iextra < atom->nextra_border; iextra++)
m += modify->fix[atom->extra_border[iextra]]->
unpack_border(n,first,&buf[m]);
}
/* ---------------------------------------------------------------------- */
void AtomVecHybrid::unpack_border_vel(int n, int first, double *buf)
{
int i,k,m,last;
int omega_flag = atom->omega_flag;
int angmom_flag = atom->angmom_flag;
m = 0;
last = first + n;
for (i = first; i < last; i++) {
if (i == nmax) grow(0);
x[i][0] = buf[m++];
x[i][1] = buf[m++];
x[i][2] = buf[m++];
tag[i] = (tagint) ubuf(buf[m++]).i;
type[i] = (int) ubuf(buf[m++]).i;
mask[i] = (int) ubuf(buf[m++]).i;
v[i][0] = buf[m++];
v[i][1] = buf[m++];
v[i][2] = buf[m++];
if (omega_flag) {
omega[i][0] = buf[m++];
omega[i][1] = buf[m++];
omega[i][2] = buf[m++];
}
if (angmom_flag) {
angmom[i][0] = buf[m++];
angmom[i][1] = buf[m++];
angmom[i][2] = buf[m++];
}
}
// unpack sub-style contributions as contiguous chunks
for (k = 0; k < nstyles; k++)
m += styles[k]->unpack_border_hybrid(n,first,&buf[m]);
if (atom->nextra_border)
for (int iextra = 0; iextra < atom->nextra_border; iextra++)
m += modify->fix[atom->extra_border[iextra]]->
unpack_border(n,first,&buf[m]);
}
/* ----------------------------------------------------------------------
pack data for atom I for sending to another proc
pack each sub-style one after the other
------------------------------------------------------------------------- */
int AtomVecHybrid::pack_exchange(int i, double *buf)
{
int k,m;
int tmp = atom->nextra_grow;
atom->nextra_grow = 0;
m = 0;
for (k = 0; k < nstyles; k++)
m += styles[k]->pack_exchange(i,&buf[m]);
atom->nextra_grow = tmp;
if (atom->nextra_grow)
for (int iextra = 0; iextra < atom->nextra_grow; iextra++)
m += modify->fix[atom->extra_grow[iextra]]->pack_exchange(i,&buf[m]);
buf[0] = m;
return m;
}
/* ----------------------------------------------------------------------
unpack data for single atom received from another proc
unpack each sub-style one after the other
grow() occurs here so arrays for all sub-styles are grown
------------------------------------------------------------------------- */
int AtomVecHybrid::unpack_exchange(double *buf)
{
int k,m;
int nlocal = atom->nlocal;
if (nlocal == nmax) grow(0);
int tmp = atom->nextra_grow;
atom->nextra_grow = 0;
m = 0;
for (k = 0; k < nstyles; k++) {
m += styles[k]->unpack_exchange(&buf[m]);
atom->nlocal--;
}
atom->nextra_grow = tmp;
if (atom->nextra_grow)
for (int iextra = 0; iextra < atom->nextra_grow; iextra++)
m += modify->fix[atom->extra_grow[iextra]]->
unpack_exchange(nlocal,&buf[m]);
atom->nlocal++;
return m;
}
/* ----------------------------------------------------------------------
size of restart data for all atoms owned by this proc
include extra data stored by fixes
------------------------------------------------------------------------- */
int AtomVecHybrid::size_restart()
{
int tmp = atom->nextra_restart;
atom->nextra_restart = 0;
int n = 0;
for (int k = 0; k < nstyles; k++)
n += styles[k]->size_restart();
atom->nextra_restart = tmp;
int nlocal = atom->nlocal;
if (atom->nextra_restart)
for (int iextra = 0; iextra < atom->nextra_restart; iextra++)
for (int i = 0; i < nlocal; i++)
n += modify->fix[atom->extra_restart[iextra]]->size_restart(i);
return n;
}
/* ----------------------------------------------------------------------
pack atom I's data for restart file including extra quantities
xyz must be 1st 3 values, so that read_restart can test on them
pack each sub-style one after the other
------------------------------------------------------------------------- */
int AtomVecHybrid::pack_restart(int i, double *buf)
{
int tmp = atom->nextra_restart;
atom->nextra_restart = 0;
int m = 0;
for (int k = 0; k < nstyles; k++)
m += styles[k]->pack_restart(i,&buf[m]);
atom->nextra_restart = tmp;
if (atom->nextra_restart)
for (int iextra = 0; iextra < atom->nextra_restart; iextra++)
m += modify->fix[atom->extra_restart[iextra]]->pack_restart(i,&buf[m]);
buf[0] = m;
return m;
}
/* ----------------------------------------------------------------------
unpack data for one atom from restart file including extra quantities
unpack each sub-style one after the other
grow() occurs here so arrays for all sub-styles are grown
------------------------------------------------------------------------- */
int AtomVecHybrid::unpack_restart(double *buf)
{
int nlocal = atom->nlocal;
if (nlocal == nmax) {
grow(0);
if (atom->nextra_store)
memory->grow(atom->extra,nmax,atom->nextra_store,"atom:extra");
}
int tmp = atom->nextra_store;
atom->nextra_store = 0;
int m = 0;
for (int k = 0; k < nstyles; k++) {
m += styles[k]->unpack_restart(&buf[m]);
atom->nlocal--;
}
atom->nextra_store = tmp;
double **extra = atom->extra;
if (atom->nextra_store) {
int size = static_cast<int> (buf[0]) - m;
for (int i = 0; i < size; i++) extra[nlocal][i] = buf[m++];
}
atom->nlocal++;
return m;
}
/* ----------------------------------------------------------------------
create one atom of itype at coord
create each sub-style one after the other
grow() occurs here so arrays for all sub-styles are grown
------------------------------------------------------------------------- */
void AtomVecHybrid::create_atom(int itype, double *coord)
{
int nlocal = atom->nlocal;
if (nlocal == nmax) grow(0);
for (int k = 0; k < nstyles; k++) {
styles[k]->create_atom(itype,coord);
atom->nlocal--;
}
atom->nlocal++;
}
/* ----------------------------------------------------------------------
unpack one line from Atoms section of data file
grow() occurs here so arrays for all sub-styles are grown
------------------------------------------------------------------------- */
void AtomVecHybrid::data_atom(double *coord, imageint imagetmp, char **values)
{
int nlocal = atom->nlocal;
if (nlocal == nmax) grow(0);
tag[nlocal] = utils::tnumeric(FLERR,values[0],true,lmp);
type[nlocal] = utils::inumeric(FLERR,values[1],true,lmp);
if (type[nlocal] <= 0 || type[nlocal] > atom->ntypes)
error->one(FLERR,"Invalid atom type in Atoms section of data file");
x[nlocal][0] = coord[0];
x[nlocal][1] = coord[1];
x[nlocal][2] = coord[2];
image[nlocal] = imagetmp;
mask[nlocal] = 1;
v[nlocal][0] = 0.0;
v[nlocal][1] = 0.0;
v[nlocal][2] = 0.0;
if (atom->omega_flag) {
omega[nlocal][0] = 0.0;
omega[nlocal][1] = 0.0;
omega[nlocal][2] = 0.0;
}
if (atom->angmom_flag) {
angmom[nlocal][0] = 0.0;
angmom[nlocal][1] = 0.0;
angmom[nlocal][2] = 0.0;
}
// each sub-style parses sub-style specific values
int m = 5;
for (int k = 0; k < nstyles; k++)
m += styles[k]->data_atom_hybrid(nlocal,&values[m]);
atom->nlocal++;
}
/* ----------------------------------------------------------------------
unpack one line from Velocities section of data file
------------------------------------------------------------------------- */
void AtomVecHybrid::data_vel(int m, char **values)
{
v[m][0] = utils::numeric(FLERR,values[0],true,lmp);
v[m][1] = utils::numeric(FLERR,values[1],true,lmp);
v[m][2] = utils::numeric(FLERR,values[2],true,lmp);
// each sub-style parses sub-style specific values
int n = 3;
for (int k = 0; k < nstyles; k++)
n += styles[k]->data_vel_hybrid(m,&values[n]);
}
/* ----------------------------------------------------------------------
pack atom info for data file including 3 image flags
------------------------------------------------------------------------- */
void AtomVecHybrid::pack_data(double **buf)
{
int k,m;
int nlocal = atom->nlocal;
for (int i = 0; i < nlocal; i++) {
buf[i][0] = ubuf(tag[i]).d;
buf[i][1] = ubuf(type[i]).d;
buf[i][2] = x[i][0];
buf[i][3] = x[i][1];
buf[i][4] = x[i][2];
m = 5;
for (k = 0; k < nstyles; k++)
m += styles[k]->pack_data_hybrid(i,&buf[i][m]);
buf[i][m] = ubuf((image[i] & IMGMASK) - IMGMAX).d;
buf[i][m+1] = ubuf((image[i] >> IMGBITS & IMGMASK) - IMGMAX).d;
buf[i][m+2] = ubuf((image[i] >> IMG2BITS) - IMGMAX).d;
}
}
/* ----------------------------------------------------------------------
write atom info to data file including 3 image flags
------------------------------------------------------------------------- */
void AtomVecHybrid::write_data(FILE *fp, int n, double **buf)
{
int k,m;
for (int i = 0; i < n; i++) {
fprintf(fp,TAGINT_FORMAT " %d %-1.16e %-1.16e %-1.16e",
(tagint) ubuf(buf[i][0]).i,(int) ubuf(buf[i][1]).i,
buf[i][2],buf[i][3],buf[i][4]);
m = 5;
for (k = 0; k < nstyles; k++)
m += styles[k]->write_data_hybrid(fp,&buf[i][m]);
fprintf(fp," %d %d %d\n",
(int) ubuf(buf[i][m]).i,(int) ubuf(buf[i][m+1]).i,
(int) ubuf(buf[i][m+2]).i);
}
}
/* ----------------------------------------------------------------------
pack velocity info for data file
------------------------------------------------------------------------- */
void AtomVecHybrid::pack_vel(double **buf)
{
int k,m;
int nlocal = atom->nlocal;
for (int i = 0; i < nlocal; i++) {
buf[i][0] = ubuf(tag[i]).d;
buf[i][1] = v[i][0];
buf[i][2] = v[i][1];
buf[i][3] = v[i][2];
m = 4;
for (k = 0; k < nstyles; k++)
m += styles[k]->pack_vel_hybrid(i,&buf[i][m]);
}
}
/* ----------------------------------------------------------------------
write velocity info to data file
------------------------------------------------------------------------- */
void AtomVecHybrid::write_vel(FILE *fp, int n, double **buf)
{
int k,m;
for (int i = 0; i < n; i++) {
fprintf(fp,TAGINT_FORMAT " %g %g %g",
(tagint) ubuf(buf[i][0]).i,buf[i][1],buf[i][2],buf[i][3]);
m = 4;
for (k = 0; k < nstyles; k++)
m += styles[k]->write_vel_hybrid(fp,&buf[i][m]);
fprintf(fp,"\n");
}
}
/* ----------------------------------------------------------------------
assign an index to named atom property and return index
returned value encodes which sub-style and index returned by sub-style
@ -1073,14 +217,3 @@ int AtomVecHybrid::known_style(char *str)
if (strcmp(str,allstyles[i]) == 0) return 1;
return 0;
}
/* ----------------------------------------------------------------------
return # of bytes of allocated memory
------------------------------------------------------------------------- */
bigint AtomVecHybrid::memory_usage()
{
bigint bytes = 0;
for (int k = 0; k < nstyles; k++) bytes += styles[k]->memory_usage();
return bytes;
}