Problem with my infile

Dear Axel and Steve,
I got help from the previous email about when I ran the same infile and data file on different processors got a totally different results. Now I think there is something wrong in my infile. I need for your help to slove the different results on different processors.
This is my infile
units metal
atom_style atomic
boundary s s s
neighbor 2.0 bin
neigh_modify every 1
#create geometry
read_data 991414a.data
pair_style airebo 3.0
pair_coeff * * CH.airebo C C
#define groups
region 1 cylinder z 0 0 7 INF INF side out units box
region 2 cylinder z 0 0 7 INF INF side in units box
region 3 block INF INF INF INF INF -16 units box
region 4 block INF INF INF INF 16 INF units box
region 5 block INF INF INF INF -16 16 units box
region 6 intersect 2 1 3
region 7 intersect 2 1 4
region 8 intersect 2 1 5
region 9 union 2 6 7
region 10 union 2 2 8
group out region 1
group in region 2
group immobile region 9
group mobile region 10
#initial conditions
velocity mobile create 150 482748 units box dist gaussian
fix 1 mobile nvt temp 150 150 0.1
fix 2 immobile setforce 0.0 0.0 0.0
timestep 0.001
thermo 1000
compute intemp in temp
compute inKE in ke
thermo_style custom step c_intemp c_inKE

dump 1 all atom 1000 wan.lammpstrj
run 3000000

Thanks
Wan

Dear Axel and Steve,
    I got help from the previous email about when I ran the same infile and
data file on different processors got a totally different results. Now I
think there is something wrong in my infile. I need for your help to slove
the different results on different processors.

sorry, but that is your problem, not ours.

Thanks for your apply.
I think there was something wrong about my relaxation(using the nvt command) after consulting other experienced people. Is our relaxation method appropriate now?

Wan

Thanks for your apply.
I think there was something wrong about my relaxation(using the nvt command)
after consulting other experienced people. Is our relaxation method
appropriate now?

i repeat. to find out whether a simulation is doing correct science is
your concern and not ours.
what LAMMPS can do and how it works is written in the documentation
and the science behind that in the given references and text books on
molecular modeling, molecular dynamics, statistical mechanics and
statistical thermodynamics.

if you have a situation where you have proof that LAMMPS does not do
what it advertises or is in disagreement with the algorithms it claims
to use, then we will make an effort to look into it. but this is not a
training school for doing molecular dynamics simulations. to teach you
proper use of simulation programs, you have to consult with your
adviser or a person your adviser refers you to and the study the
corresponding text books.

think about it. if you have problems with your car, you don't go to
the manufacturer but to a local mechanic unless you can prove there is
a design fault or manufacturing fault. and even then you usually
consult with a local mechanic first to make sure you have sufficient
proof. right?

axel.