problem for using fix nve & fix aveforce

​Dear lammps developers,
In my simulation I need to use fix nve & fix aveforce for a group but this results in explosion of group and atom lost error.
Following is the concise form of the script showing the problem. ​
What is wrong with in the script?
Regards.
Rouhollah,

units metal

dimension 3
boundary p p s
atom_style atomic
variable runstep_relax equal 10000

variable TIMESTEP equal 0.002
variable lat_block equal 5.431 #Si
lattice diamond ${lat_block}

region s_box block 0 10 0 5 0 10
create_box 1 s_box
region UP block 0 10 0 5 3 8

create_atoms 1 region UP
pair_style tersoff

pair_coeff * * SiC.tersoff Si
mass 1 28.0855 ## Si
compute new all temp

velocity all create 300 577625 temp new

fix 1 all nve
fix 2 all aveforce 0.0 0.0 0.0

thermo 500
thermo_modify temp new
dump 10 all image 1000 image.*.jpg type type &
zoom 1.6 size 1024 768 adiam 1.5 axes yes 0.05 0.05 view 90 -90
timestep {TIMESTEP} run {runstep_relax}

​Dear lammps developers,
In my simulation I need to use fix nve & fix aveforce for a group but this
results in explosion of group and atom lost error.
Following is the concise form of the script showing the problem. ​
What is wrong with in the script?

​there is nothing wrong with the script. it is (syntactica​lly) correct and
the resulting trajectory is what you are supposed to be getting from that
combination of commands.

you set up your atoms to have random velocities (in magnitude and
direction) and then you change the forces on each atom to be the average
force on the whole group, which will be (close to) zero. so what should you
get when you do time integration with random velocities and (almost) no
force? look up newton's first law of motion, if you need a reference.

so if there is a problem, it would be located somewhere between your chair
and your computer.

axel.