[lammps-users] NVT reax


I am running a MD simulation using NVT ensemble. I use reax force field. Timpstep is 1fs, the simulation has ran for 10000 steps. I keep the temperature at 65 K,but it fluctuates to 400K, and not stable. Below is my input script
units real
dimension 3
boundary p p p
neigh_modify delay 0 every 1 check yes
atom_style charge

read_data data.psi

pair_style reax
pair_coeff * * ffield.reax 2 6

fix 1 all nvt temp 65.0 65.0 10.0 drag 0.2

thermo 10
thermo_style multi
timestep 1.0
restart 200000 restart.psi
run 1000000

dump 1 all custom 2500 dump.psi id type q x y z
run 100000

Before this simulation, I have ran another simulation of the same system.The only difference is that I set "drag 0.0’ in the fix command. I ran it for 1ns, the temperature is not stable,it even fluctuated to about 10000 K. Can anyone explain this and solve this problem?

Thank you!

Did you minimize the system before starting NVT?


Yes, I used minimized structure.

1.0 fmsec may be too large a timestep for a ReaxFF model. I wouldn;t
use drag. I would instead inititialize the system to 65K. If you can
run NVE in a stable manner, then you should be able to use NVT.


Hi, Steve, Vikas, Gilbert,

Thank you for all your replies. I set the timestep to 0.5 fs, and till now the temperature seems to be stable around 65 K.