[lammps-users] Lost atoms error

I am trying to perform NPT simulation of SPC/E water to simulate hydrothermal water. But the simulation never starts. I always get "ERROR: Lost atoms: original 5232 current 0" after step 0 itself. How can I lose all atoms in the first step itself?

somthing about your problem is not setup correctly - could be a
lot of reasons , post your input file


Dr. Plimpton,
I have attached my input files for hydrothermal SPC/E water. Your

help is >appreciated.

Thank you,

There are a couple problems with your input file. The damping parameters
in your fix npt are 0.01 when they should be something like 100.0. So it is
trying to equilibrate in 0.01 fmsec which will blow up your system. Also your
timestep is 0.01 fmsec which is tiny.

Also, your SHAKE stats at time 0 are showing:

SHAKE stats (type/ave/delta) on step 0
  1 1.09487 5.48436
  1 109.104 128.34

which means SHAKE is not able to constrain your bond lengths and angle.
The spread in bond lengths
is 5.4 Angs when you want them all to be 1.09 Angs. So some bonds are
6+ Angs. Ditto for angles - some are very far from 109 degrees.

So doing dynamics on this system is going to cause problems.

Running without SHAKE does better. But your pair potentials show no
interactions with H atoms. That seems wrong and probably what is causing
problems with SHAKE.