why some atoms run away?

Dear LAMMPS users and developers:
   I am using lammps to do my project.I found a problem simulation.
   Some atoms in the down group run away when I run the program.I search in your webside to find what happened.Someone says to reduce the timestep can solve the problem,but it doesn't change anymore when I try.Also I find the problem will not happened when I use a smaller size or lowwer temperature.Can you tell me how to sovle it?
  Thank you!
the following is my input script and two pictures about the model

dimension 3
units metal
boundary s s s
neighbor 2.0 nsq
neigh_modify every 1

atom_style atomic
read_data data.vmd
pair_style airebo 2.93 1 1
pair_coeff * * CH.airebo C

# define groups
region 1 block INF -0.5 INF INF INF INF units box
region 2 block 95.5 INF INF INF INF INF units box
region 3 block INF INF 49 INF INF INF units box
region 4 block INF INF INF 0.5 INF INF units box
group left region 1
group right region 2
group up region 3
group down region 4
group boundary union right left up down
group mobile subtract all boundary

min_style cg
minimize 1.0e-6 1.0e-6 1000 10000

velocity mobile create 300.0 4928459 dist gaussian
velocity boundary set 0.0 0.0 0.0 units box

fix 1 mobile nvt temp 300.0 300.0 0.05
fix 2 boundary nvt temp 300.0 300.0 0.05
fix 3 left setforce 0.0 NULL 0.0
fix 4 up setforce NULL 0.0 0.0
fix 5 down setforce NULL 0.0 0.0
fix 6 right setforce 0.0 NULL 0.0

timestep 0.0005
variable w equal bound(mobile,zmax) #the deflection of the graphene
thermo_style custom step temp etotal lx ly v_w
dump 1 all custom 1000 dump.out1000 id type x y z
thermo 500
run 10000

before running:

running:(some atoms in the down group run away )

2012-05-18

xuyumou

2.jpg

未命名1.jpg

Dear LAMMPS users and developers:
I am using lammps to do my project.I found a problem simulation.

you forgot to state which version of LAMMPS exactly.

Some atoms in the down group run away when I run the program.I search
in your webside to find what happened.Someone says to reduce the timestep
can solve the problem,but it doesn't change anymore when I try.Also I find
the problem will not happened when I use a smaller size or lowwer
temperature.Can you tell me how to sovle it?

no. but it may be due to using fix setforce on atoms that are integrated
with fix nvt. i don't believe that that is a valid option and that it will work
the way you expect it to work.

axel.