[lammps-users] how to relax the systerm?

dear lammps users:
i build the cu sample of 8092 atoms,and i want make it minimized and relaxed at 300K.At first,i wrote the in file below and i found that if timestep=1fs was used, the systerm volumn will became larger and larger,some atoms ran away from the sample.So i thought maybe the sample was not stable enough, i used the timestep=0.01 and
fix 1 all nvt 300.0 300.0 0.1 drag 0.2
write_restart restart0.01.*
then read_restart and change timestep into 0.05 and so on .but as long as timestep=0.1, the sample would explode again,i don’t know why.could someome explain how to relax the systerm explicitely?

thank you very much.

#testcu file

units real
atom_style atomic
boundary p p s
read_data data3t.cu
timestep 0.1

pair_style eam
pair_coeff * * …/potentials/Cu_u3.eam

neighbor 0.5 bin
neigh_modify every 100 delay 0 check no

dump 1 all xyz 1000 dump0.1testtt.xyz
thermo 1000
thermo_style custom step temp press pe ke etotal vol
thermo_modify lost warn

#energy minimization

fix 1 all nvt 300.0 300.0 0.1 drag 0.2

minimize 1.0e-8 1.0e-6 1000 10000
run 10000
minimize 1.0e-8 1.0e-6 1000 10000

In real units, a timestep of 0.1 is 100 fmsec, which
is about 50-100x too large.


Sorry, I confused metal and real units. You are running EAM for Cu,
so I assumed you were using metal units. A timestep of 0.1 in real
units is 0.1 fmsec, but you cannot use real units with EAM files.
All the parameters in the files are in metal units (e.g. eV). So you must
use metal units and then time is in psec, and a timestep of 0.1 would
be too big, as I said in the last message. So your real problem is
that you are using the wrong units with EAM. The pair eam doc
page discusses this.