[lammps-users] Bad principal moments

LAMMPS (21 May 2008)
Lattice spacing in x,y,z = 35.2 35.2 35.2
Created orthogonal box = (0 0 -0.1) to (7500 750 0.1)
  8 by 1 by 1 processor grid
Created 4708 atoms
4708 atoms in group argon
Lattice spacing in x,y,z = 3.52 3.52 3.52
Deleted 185 atoms, new total = 4523
Created 18494 atoms
18494 atoms in group nickel
1 rigid bodies with 18494 atoms
ERROR: Bad principal moments

If I change lattice fir rigid body (for example lattice 5) all is work
That is that error mean and how deal with it?

units real
dimension 2
atom_style atomic
boundary p p p

lattice sq 35.2
region box block 0 7500 0 750 -0.1 0.1 units box
create_box 2 box
create_atoms 1 box

mass 1 49.0
group argon type 1
mass 2 50.0

lattice sq 3.52
region sphere1 sphere 3750 375 0 270 units box

delete_atoms region sphere1
create_atoms 2 region sphere1

velocity argon create 300.0 87287 dist gaussian units box
group nickel type 2
velocity nickel set 0.00325 0.0 0.0 sum yes

pair_style lj/cut 3.4
pair_coeff 1 1 0.238 3.4 3.4
pair_coeff 2 2 0.238 3.5 3.5
pair_coeff 1 2 0.238 6.4 6.4

neighbor 10.0 bin
neigh_modify delay 0

fix 1 argon nve
fix 2 all enforce2d
fix 3 nickel rigid single

dump id all atom 50 dump.body.2D
thermo 1000

timestep 1.5
run 400000

For some reason LAMMPS doesn't like the rigid body that
has been defined. You'll probably need to instrument the
error message and figure out what it didn't like. Are the 18K
atoms highly symmetric? That can cause problems sometimes.

Steve