[lammps-users] Bond atoms # # missing on proc 0 at step #

LAMMPS LAMMPS Mailing List Mirror
1

Hello All,

I am using the LAMMPS 27 May 2021 version. I am currently testing my H2O model. My data file is tally water data file. I set my bond and forcefield as the SPCE model. I first used fix shake to make my model rigid and then use fix nve to test it. However, I always met with the problem of the bond atom missing. I learned that a possible problem is the cutoff distance is too short, but I think my cutoff is long enough.

Below is my input file, could anyone give me some ideas on a possible problem?

Great thanks

Input file for H2O heat and quench to disorder

------------------------ INITIALIZATION ----------------------------

units real
atom_style full
bond_style harmonic
angle_style harmonic

----------------------- ATOM DEFINITION ----------------------------

read_data H2O.data

------------------------ FORCE FIELDS ------------------------------

pair_style lj/cut/coul/cut 9.0 9.0
pair_coeff 1 1 0.1553 3.166
pair_coeff 2 1*2 0 0
info coeffs

#------------------------- Bond and Angle Setting ------------------------------
bond_coeff 1 80 1.0
angle_coeff 1 80 109.47

------------------------- SETTINGS ---------------------------------

compute peratom all pe/atom
compute temp all temp

2

The problem is cause by this block of input:

reset_timestep 0
fix 1 all shake 0.0001 100 10 t 1 2 b 1 a 1
run 100
unfix 1
fix 1 all nve

First you run 100 steps with fix shake, but no time integration (so all atoms will remain where they are).
Then you turn off fix shake and instead enable fix nve. That means now your atoms can move, but the model is not constrained.
Without fix shake the time integration will diverge using a timestep of 1fs. Thus you will need both to keep your water molecules rigid while also moving them.

One more thing. Never EVER add “thermo_modify lost ignore” to your input unless you have an open system and explicitly want atoms to leave it (it won’t work for molecules like this anyway). Lost atoms or bonds are an indication of a serious issue with your input and you won’t correct that by ignoring the error.

Axel.