LAMMPS error lost atoms thermo.cpp:395

Dear LAMMPS user,

I have introduced in LAMMPS, the tabulated potentials for clathrate material, Ge46,.
My input potential file is such as:

Pair potential for atom types 1 1 i,r,energy,force

Ge-Ge
N 1000 R 0.1 6
1 0.100000 6.752174 -2.43226859917
2 0.105906 1.433602 -1.3536564063

1000 0.1054506 2.435258 -9.3536545853

In my param file, I used this command bellow to introduce in my simulation script the potential from the an input file see previously.

Commandes in the param file:
pair_style table spline 1000
pair_coeff 1 1 Ge46_potentials_eopp_LAMMPS.pot Ge-Ge 6

The rcutoff = 6 Angstrum for potential.
There is 1 type of atoms ‘Ge’. There are 46 Ge atoms in the unit cell (cubic 10 Angstrum).

I have this error:
ERROR: Lost atoms: original 46 current 36 (thermo.cpp:395)

I observe a relation between this error and fix command:
fix NVT all nvt temp $T $T 10 drag 0.2

Do you know the reason for this error problem ?
Why the simulation break, when I write the fixed command ?

I suspect 3 three different causes of this error.
1/ time step is so long. But in my script dt = 1 fs. I remove this cause.
2/ Structure is wrong. It is not the case, I can see perfect atomic arrangement with vmd.
3/ Problem with implemented potential. I don’t known the problem in this case.

Regards,
Pierre-François.

The error section of the doc page has a paragraph that explains
how lost atoms typically occur, It is almost always a bad model]
or initial conditions. You should monitor thermo output every tilmestep
and you should viz what is going on.

Steve