Problems in NPT run in PDLAMMPS

LAMMPS LAMMPS Mailing List Mirror
1

Dear lammps user
I have been working with PDLAMMPS since 9 month ago and ran into several problems.
I describe one of problems below.
I want to simulate a system with PDLAMMPS using NPT ensemble under external pressure and got #IND for all desired output data after 200 steps. I will really appreciate if anyone could help me to solve this problem.
Here are my input scripts and output results:

small Peridynamic cube under external pressure S0

clear
units si
boundary p p p
atom_style peri
atom_modify map array
neighbor 1.2e-8 bin
#------------- simulation box --------------

lattice sc 0.28e-8
region target block 0 0.5e-7 0 0.5e-7 0 0.5e-7 units box
create_box 1 target
create_atoms 1 region target
#------------- PD Potentials --------------
pair_style peri/lps
pair_coeff * * 3.14e9 1.2e9 0.9e-8 0.005 0.25
set group all density 1200
set group all volume 2.1982-26

velocity all create 0.1 4928459 mom yes rot yes dist uniform
#------------- NPT run under external pressure --------------

thermo 100
thermo_style custom step temp press vol lx ly lz
fix 1 all npt temp 0.1 0.1 100 iso -2.97e12 -2.97e12 1000
dump 2 all atom 10 NPT.xyz
timestep 1.0e-8
run 5000
output results

Step Temp Press Volume Lx Ly Lz
0 0.1 64.40458 1.25e-022 5e-008 5e-008 5e-008
100 0.099999995 64.404572 1.2500001e-022 5.0000001e-008 5.0000001e-008 5.0000001e-008
200 -1.#IND -1.#IND -1.#IND -1.#IND -1.#IND -1.#IND
300 -1.#IND -1.#IND -1.#IND -1.#IND -1.#IND -1.#IND
400 -1.#IND -1.#IND -1.#IND -1.#IND -1.#IND -1.#IND
500 -1.#IND -1.#IND -1.#IND -1.#IND -1.#IND -1.#IND
600 -1.#IND -1.#IND -1.#IND -1.#IND -1.#IND -1.#IND
700 -1.#IND -1.#IND -1.#IND -1.#IND -1.#IND -1.#IND
800 -1.#IND -1.#IND -1.#IND -1.#IND -1.#IND -1.#IND
900 -1.#IND -1.#IND -1.#IND -1.#IND -1.#IND -1.#IND
1000 -1.#IND -1.#IND -1.#IND -1.#IND -1.#IND -1.#IND
1100 -1.#IND -1.#IND -1.#IND -1.#IND -1.#IND -1.#IND
1200 -1.#IND -1.#IND -1.#IND -1.#IND -1.#IND -1.#IND
1300 -1.#IND -1.#IND -1.#IND -1.#IND -1.#IND -1.#IND
1400 -1.#IND -1.#IND -1.#IND -1.#IND -1.#IND -1.#IND
1500 -1.#IND -1.#IND -1.#IND -1.#IND -1.#IND -1.#IND
1600 -1.#IND -1.#IND -1.#IND -1.#IND -1.#IND -1.#IND
1700 -1.#IND -1.#IND -1.#IND -1.#IND -1.#IND -1.#IND
1800 -1.#IND -1.#IND -1.#IND -1.#IND -1.#IND -1.#IND
1900 -1.#IND -1.#IND -1.#IND -1.#IND -1.#IND -1.#IND
2000 -1.#IND -1.#IND -1.#IND -1.#IND -1.#IND -1.#IND
2100 -1.#IND -1.#IND -1.#IND -1.#IND -1.#IND -1.#IND
2200 -1.#IND -1.#IND -1.#IND -1.#IND -1.#IND -1.#IND
Regards
Faraz M. Honarvar
PhD candidate of Polymer Engineering
Polymer Engineering Faculty
Sahand University Of Technology (SUT)
Tabriz- IRAN
TEL: (+98) 9123194229

NPT run in PDLAMMPS.in (985 Bytes)

output results.txt (2.04 KB)

2

Dear lammps user
I have been working with PDLAMMPS since 9 month ago and ran into several
problems.
I describe one of problems below.
I want to simulate a system with PDLAMMPS using NPT ensemble under external
pressure and got #IND for all desired output data after 200 steps. I will
really appreciate if anyone could help me to solve this problem.

fix npt can be a very ticklish integrator for systems that are outside
their "comfort zone". if your settings are off, it can very easily
happen, that it won't run stable. i have very little experience in
peridynamics, but for atomistic simulations, it is usually a good idea
to first equilibrate the system properly *without* changing the volume
and then perhaps study what are reasonable settings for volume
adjustments with a simpler and less "temperamental" barostat, e.g. fix
press/berendsen and then later switch to fix npt. you may also need to
adjust time step and other settings to accommodate the fact that
volume changes result in changes of the neighborhood of atoms
throughout the entire volume.

axel.

3

I also don’t think hardly anyone does NPT for PD models.

Just NVE or possible NVT. I suggest you read the PD

literature, or email people who are PD experts who

have written papers using PD in LAMMPS. Several

of them have posted to the mail list in the past,

so you can probably find them by searching the mail list.

Steve

4

This does not look like a script that you have been working on for 9 months. You are running PD with a barostat set to -3e12, that means 3000 GPa of hydrostatic tension. Your pressure time constant is equal to 1.0e11 timesteps. Your temperature is 0.1 kelvins. Why are you surprised this simulation failed? I suggest you take some time to study the response of well-constructed PD examples. You can’t type in arbitrary numbers and expect everything to magically just work. Also, as Axel and Steve said, maybe you should not use the NPT barostat at all.

Aidan