[lammps-users] fix momentum

hi:
I use NPT for my MD. I found my system drifting and rotating. And two atoms leave my system. I think I should zero the linear and anguar moment, so I use the comand as :
fix 1 all npt 80.0 80.0 10 aniso NULL NULL NULL NULL 0.0 0.0 10 drag 1.0
fix 2 all momentum 10 linear 1 1 1 angular
Adding the comand fix momentum, but I found system’s drifting and rotating still appear. And two atoms still leave my system.Why?
this is in script:

3d al-cu simulation

units metal
boundary s s p

atom_style atomic
lattice bcc 3.610
region box block 0 10.0 0 10.0 0 40.0
create_box 2 box

create_atoms 1 box basis 2 2

pair_style eam/alloy
pair_coeff * * AlCu.eam.alloy Al Cu

neighbor 2.0 bin
neigh_modify delay 5

velocity all create 80.0 5812775 rot yes
fix 1 all npt 80.0 80.0 10 aniso NULL NULL NULL NULL 0.0 0.0 10 drag 1.0
fix 2 all momentum 10 linear 1 1 1 angular
thermo_style custom step temp pzz etotal vol
thermo 10
dump 1 all atom 10000 1dump11.atom
restart 50000 1npt11.restart
timestep 0.001
run 50000

unfix 1
fix 1 all npt 80.0 80.0 100 aniso NULL NULL NULL NULL 0.0 0.0 100 drag 0.2
unfix 2
fix 2 all momentum 1000 linear 1 1 1 angular
thermo_style custom step temp pzz etotal vol
thermo 10
dump 2 all atom 50000 1dump12.atom
restart 100000 1npt12.restart
timestep 0.001
run 200000

unfix 1
fix 1 all npt 80.0 80.0 1000 aniso NULL NULL NULL NULL 0.0 0.0 10000
unfix 2
fix 2 all momentum 10000 linear 1 1 1 angular
thermo_style custom step temp pzz etotal vol
thermo 10
dump 3 all atom 10000 1dump13.atom
restart 100000 1npt13.restart
timestep 0.001
run 500000

Sorry, your question is too vague. There are lots of reasons
atoms can leave your system. Do you have free surfaces?
Do you shrink-wrap the boundaries (preferred) or make them
fixed (can lose atoms easily). Or are the atoms being
blown out of your box?

Can you describe something specific with your simulation
where you think LAMMPS is doing the wrong thing, despite
using the commands you list?

Steve