LAMMPS

Video about installing LAMMPS

https://www.youtube.com/watch?v=9NtUV18wK0I

LAMMPS input - NVE MD (HPCC)

#number of particles

variable natom equal 1000

units       lj

dimension   3

atom_style  atomic

boundary    p p p

# define 3D box region

region box block -100 100 -100 100 -100 100

#create the simulation box with one type of particle

create_box 1 box

#put particles with type 1 randomly into the box

create_atoms 1 random ${natom} 324523 box

#all particles has mass 1

mass  *  1

#Lennard-Jones potential between all particles

pair_style lj/cut 2.5

pair_coeff * * 1 1

#=============================================================

# Step 1: minimize energy first to avoid overlapping particles

#=============================================================

minimize 1e-7 1e-7 10000 10000

reset_timestep 0

#save snapshots

#shell "mkdir img"

#shell "rm img/*"

#dump img all image 1000 img/t*.jpg type type adiam 1.0 zoom 1.0

#dump_modify img backcolor white boxcolor black

#dump_modify img pad 6

# dump for OVITO to postprocess

shell "mkdir geo"

shell "rm geo/*"

dump 1 all custom 250  geo/dump.* id type x y z

#initialize velocity distribution corresponding to a temperature of 2

velocity all create 2 34234123 dist gaussian

# evolve system using Nose-Hoover themostat (an NVT simulation)

fix 1 all nve

#output time step, temperature, average kinetic and potential energy for plotting

thermo_style custom step temp ke pe

thermo       100

#time step of integrator

timestep    0.001

#start the simulation

run     100000

LAMMPS submission script (HPCC)

#!/bin/bash

#SBATCH --job-name=lammps

#SBATCH --time=1-00:00:00

#SBATCH --nodes=1

#SBATCH --ntasks-per-node=24

#SBATCH --cpus-per-task=1

module load lammps

srun lmp_hpcc -in input.lammps > output.out

LAMMPS input

If you are working on the PFGP project, we first need to correct the Lammps input file with the following function:

#scratch is the path of the folder containing the file

#The original file is scratch+'in.'+fil 

#here fil is a variable that contains the name

#home is a different folder 

import os

def RW(fil, scratch, home):

    with open(scratch+'in.'+fil, 'r') as f:

        w = open(home+'in.'+fil,'w')

        for line in f:

            w.write(line)

            if '500 steps CG Minimization' in line:

                w.write('fix 1 all box/relax iso 0.0 vmax 0.001\n')

            if 'undump' in line:

                w.close()

                break    

    os.rename(home+'in.'+fil,scratch+'in.'+fil)

After that we use the following script:

-------------------------------------------------

#!/bin/bash

#SBATCH --job-name='lammps' 

#SBATCH -n 1

#SBATCH -t 01:00:00

#SBATCH -p mendoza_q

#SBATCH -o 'output.out' 

#SBATCH -e 'error.e'

lmp < in.LammpsInputfFlleName 

Example - Parallel Job (RCC-FSU manual is wrong)

This is the correct example.  We will use the same in.lammps  example here.

Following submit script will use 4 processes to run this example.

#!/bin/bash

#SBATCH -J "in.lammps"

#SBATCH -N 1

#SBATCH --ntasks-per-node=4

#SBATCH -p mendoza_q

#SBATCH -t 200:10:00

module purge all

module load gnu-openmpi

mpirun -np 4 lmp < in.lammps

Suggestion from Prasad Maddumage

#!/bin/bash

#SBATCH -J "in.lammps"

#SBATCH -N 1

#SBATCH --ntasks-per-node=4

#SBATCH -p mendoza_q

#SBATCH -t 200:10:00

module purge all

module load gnu-openmpi

mpirun -np 4 lmp

Also, using srun instead of mpirun is a good practice (srun -n 4 instead of mpirun -np 4).

Output explanation:

ITEM: TIMESTEP 

0

ITEM: NUMBER OF ATOMS

196

ITEM: BOX BOUNDS pp pp pp (x_initial, x_final, y_initial, y_final, z_initial, z_final are all periodic)

0.0000000000000000e+00 2.1883970000000001e+01 (x initial, x final)

0.0000000000000000e+00 2.1883970000000001e+01 (y initial, y final)

0.0000000000000000e+00 1.7868350000000000e+01 (z initial, z final)

ITEM: ATOMS id type xs ys zs ix iy iz  (atom id, atom type, 3 scaled coordinates,  box images that the atom is in)

33 5 5.02651e-05 0.288646 0.161259 1 0 0

34 5 0.0314248 0.342989 0.294371 1 0 0

Scaled coordinates are numbers from 0 to 1, a value of .25 means the atom is at 1/4 of the distance from x_initial to x_final.

For periodic dimensions, ix, iy, iz specify which image of the simulation box the atom is considered to be in. An image of (0,0,0) means it is inside the box defined. A value of (-1,0,0) means substract 1 x-length to get the true value.