Issue with NPH and Langevin fixes and minimize command

Dear Dr. Axel Kohlmeyer,

Hello Dr. Kohlmeyer. Once again, I deeply appreciate your many advices on LAMMPS.

I encountered a problem when running an MD simulation consisting of polymer chains (with bond, angle, dihedral, and improper under two fixes:

fix 1 all nph iso 1.0 1.0 2000
fix 2 all langevin 550 550 180 542305

Under these two fixes, I obtain an error stating “bond atom missing on processor” or my tabulated bonds exceed the specified distance. It is evident that the bong length is blowing up since my cutoff distance is correct.

On the contrary, if I include the minimization command such as the following:

min_style cg
minimize 1e-25 1e-25 500000 1000000

the MD run smoothly where the total potential energy reaches constant and the temperature hovers around 550 K, as expected.

Why is this happening considering that my initial configuration is already at equilibrium such that the initial bond length and bond angles are oriented around the minimum bond potential energy? (Minimize command should not be necessary.)

Moreover, the two fixes should relax the box size and the time integration should not result in atoms separating far apart since all pairwise bond, angle, dihedral, and improper potential parameters are set within reasonable range.

I have attached a very short script that reproduces this issue along with my external data file and my tabulated bond data.

Thank you so much for your time!

Sincerely,

Masato Koizumi

LJ96_Relaxation.in (1.7 KB)

10c10m_polymer.data (73.5 KB)

P2_P3_Bond_Table.table (69.7 KB)

If the bond atom missing error is happening quickly when you don’t
minimize first, then it probably means the system is not
as relaxed as you think.

Steve