jianmu
/
lammps
forked from lijiext/lammps
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Commit new serial algorithms

This commit is contained in:
julient31 2017-11-06 15:40:05 -07:00
parent c6bb9586ef
commit 4cbda74df4
2 changed files with 64 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
in.co_magnetomech
View File

@ -68,7 +68,8 @@ velocity all create 200 4928459 rot yes dist gaussian
#pair_style eam/alloy
#pair_style hybrid/overlay eam/alloy pair/spin/soc 4.0
#pair_style hybrid/overlay eam/alloy pair/spin/exchange 4.0 pair/spin/soc/neel 4.0
pair_style hybrid/overlay eam/alloy pair/spin/exchange 6.5
#pair_style hybrid/overlay eam/alloy pair/spin/exchange 6.5
pair_style hybrid/overlay eam/alloy pair/spin/exchange 4.0
pair_coeff * * eam/alloy ../Co_PurjaPun_2012.eam.alloy Co
#pair_coeff * * ../Co_PurjaPun_2012.eam.alloy Co
@ -132,7 +133,7 @@ variable mag_force equal f_1
thermo 10
#thermo_style custom step time v_emag c_out_pe c_out_ke c_out_temp v_mag_force v_magnorm v_tmag etotal
#thermo_style custom step time v_emag c_out_pe c_out_ke etotal
thermo_style custom step time v_emag c_out_pe c_out_ke v_magnorm etotal
thermo_style custom step time v_magnorm etotal
thermo_modify format float %20.15g
#fix out_vals all ave/time 1 1 50 step v_emag file temp_lattice_VSRSV.dat format %20.16g
@ -141,5 +142,5 @@ dump 1 all custom 5000 dump_VSRSV.lammpstrj type x y z spx spy spz
#Running the simulations for N timesteps
run 10000
#run 1
#run 100

66
src/SPIN/fix_integration_spin.cpp
View File

@ -220,6 +220,11 @@ void FixIntegrationSpin::initial_integrate(int vflag)
int *type = atom->type;
int *mask = atom->mask;
//#define MAG_TEST
#if defined MAG_TEST
tagint *tag = atom->tag;
#endif
// advance spin-lattice system, vsrsv
// update half v for all particles
for (int i = 0; i < nlocal; i++) {
@ -238,8 +243,14 @@ void FixIntegrationSpin::initial_integrate(int vflag)
int nseci;
for (int j = 0; j < nsectors; j++) { // advance quarter s for nlocal
comm->forward_comm();
#if defined MAG_TEST
if (j == 0) {
//for (int i = 0; i < nlocal; i++) {
printf("L1: test atom i=%d, tagi=%d \n",0,tag[0]);
//}
}
#endif
for (int i = 0; i < nlocal; i++) {
//comm->forward_comm();
xi[0] = x[i][0];
xi[1] = x[i][1];
xi[2] = x[i][2];
@ -252,7 +263,6 @@ void FixIntegrationSpin::initial_integrate(int vflag)
for (int j = nsectors-1; j >= 0; j--) { // advance quarter s for nlocal
comm->forward_comm();
for (int i = nlocal-1; i >= 0; i--) {
//comm->forward_comm();
xi[0] = x[i][0];
xi[1] = x[i][1];
xi[2] = x[i][2];
@ -263,6 +273,7 @@ void FixIntegrationSpin::initial_integrate(int vflag)
}
}
} else if (mpi_flag == 0) { // serial seq. update
comm->forward_comm(); // comm. positions of ghost atoms
for (int i = 0; i < nlocal-1; i++){ // advance quarter s for nlocal
ComputeInteractionsSpin(i);
AdvanceSingleSpin(i,dts,sp,fm);
@ -290,9 +301,15 @@ void FixIntegrationSpin::initial_integrate(int vflag)
if (mpi_flag == 1) { // mpi seq. update
int nseci;
for (int j = 0; j < nsectors; j++) { // advance quarter s for nlocal
comm->forward_comm();
comm->forward_comm();
#if defined MAG_TEST
if (j == 0) {
//for (int i = 0; i < nlocal; i++) {
printf("L2 test atom i=%d, tagi=%d \n",0,tag[0]);
//}
}
#endif
for (int i = 0; i < nlocal; i++) {
//comm->forward_comm();
xi[0] = x[i][0];
xi[1] = x[i][1];
xi[2] = x[i][2];
@ -305,7 +322,6 @@ void FixIntegrationSpin::initial_integrate(int vflag)
for (int j = nsectors-1; j >= 0; j--) { // advance quarter s for nlocal
comm->forward_comm();
for (int i = nlocal-1; i >= 0; i--) {
//comm->forward_comm();
xi[0] = x[i][0];
xi[1] = x[i][1];
xi[2] = x[i][2];
@ -316,6 +332,7 @@ void FixIntegrationSpin::initial_integrate(int vflag)
}
}
} else if (mpi_flag == 0) { // serial seq. update
comm->forward_comm(); // comm. positions of ghost atoms
for (int i = 0; i < nlocal-1; i++){ // advance quarter s for nlocal
ComputeInteractionsSpin(i);
AdvanceSingleSpin(i,dts,sp,fm);
@ -346,6 +363,12 @@ void FixIntegrationSpin::ComputeInteractionsSpin(int ii)
int *type = atom->type;
const int newton_pair = force->newton_pair;
//#define SERIAL2
#if defined SERIAL2
int num_j;
tagint *tag = atom->tag;
#endif
// add test here
if (exch_flag) {
inum = lockpairspinexchange->list->inum;
@ -363,9 +386,12 @@ void FixIntegrationSpin::ComputeInteractionsSpin(int ii)
int vflag = 0;
int pair_compute_flag = 1;
//#define SERIAL1
#if defined SERIAL1
if (mpi_flag == 0) {
comm->forward_comm();
}
#endif
// force computation for spin i
i = ilist[ii];
@ -384,11 +410,23 @@ void FixIntegrationSpin::ComputeInteractionsSpin(int ii)
// pair interaction
for (int jj = 0; jj < jnum; jj++) {
j = jlist[jj];
j &= NEIGHMASK;
itype = type[ii];
jtype = type[j];
#if defined SERIAL2
if (mpi_flag == 0) {
if (j >= nlocal) {
num_j = atom->map(tag[j]);
sp[j][0] = sp[num_j][0];
sp[j][1] = sp[num_j][1];
sp[j][2] = sp[num_j][2];
}
}
#endif
spj[0] = sp[j][0];
spj[1] = sp[j][1];
spj[2] = sp[j][2];
@ -521,6 +559,8 @@ int FixIntegrationSpin::coords2sector(double *xi)
void FixIntegrationSpin::AdvanceSingleSpin(int i, double dtl, double **sp, double **fm)
{
int j=0;
int *sametag = atom->sametag;
double dtfm,msq,scale,fm2,fmsq,sp2,spsq,energy,dts2;
double cp[3],g[3];
@ -550,13 +590,27 @@ void FixIntegrationSpin::AdvanceSingleSpin(int i, double dtl, double **sp, doubl
sp[i][0] = g[0];
sp[i][1] = g[1];
sp[i][2] = g[2];
// renormalization (may not be necessary)
msq = g[0]*g[0] + g[1]*g[1] + g[2]*g[2];
scale = 1.0/sqrt(msq);
sp[i][0] *= scale;
sp[i][1] *= scale;
sp[i][2] *= scale;
// comm. sp[i] to atoms with same tag (serial algo)
if (mpi_flag == 0) {
if (sametag[i] >= 0) {
j = sametag[i];
while (j >= 0) {
sp[j][0] = sp[i][0];
sp[j][1] = sp[i][1];
sp[j][2] = sp[i][2];
j = sametag[j];
}
}
}
}
/* ---------------------------------------------------------------------- */