Neural Network Molecular Dynamics Kit

NNMDKit generates polymer topology from SMILES strings and automatically builds LAMMPS data and input files for molecular dynamics simulations

Example

Below is an example python script where we use NNMDKit to generate LAMMPS data and input files for Tg measurement with a list of SMILES strings.

import nnmdkit

smiles = ['*CC*', '*CC(*)C', '*CC(*)CC', '*CC(*)CCC', '*CC(*)CCCC','*CC(*)c1ccccc1']

for s in smiles:
    sys = nnmdkit.System(smiles=s, mw=10000, ntotal=3000, density=0.5)
    data = sys.write_data(output_dir=s)

    lmp = nnmdkit.Lammps(data, NN_POTENTIAL='~/p-rramprasad3-0/NNLMP/potential_saved')
    lmp.add_procedure('minimization', min_style='cg')
    lmp.add_procedure('equilibration', Tfinal=600, Pfinal=1, Tmax=800, Pmax=49346.163)
    lmp.add_procedure('Tg_measurement', Tinit=600, Tfinal=100, Tinterval=25, step=1000000)
    lmp.write_input(output_dir=s)
                       
    job = nnmdkit.Job(jobname=s, project='GT-rramprasad3-CODA20', nodes=2, ppn=24, 
                      walltime='48:00:00', LAMMPS_EXEC='~/p-rramprasad3-0/NNLMP/lmp')
    job.write_pbs(output_dir=s)

A tutorial on using NNMDKit to create simulations of hydrocarbon polymers can be found here.

Installation

pip install .

Requirements

• EMC or PSP
• RDKit

Note that, both EMC/PSP and RDKit are required to be installed manually. NNMDKit requires EMC or PSP to create polymer structures. To configure the integration with EMC, two environment variables are required to be addded to locate your EMC executable (emc_linux64 for Linux, emc_macos for MacOS, or emc_win32 for Windows) and setup tool (emc_setup.pl). Add the paths of the EMC executable and setup tool as environment variables "EMC_EXEC" and "EMC_SETUP", respectively.

For Ramprasad Group users on Tyrion2, simply type in the command line:

echo "export EMC_EXEC=/data/kevin/EMC/bin/emc_linux64" >> ~/.bashrc
echo "export EMC_SETUP=/data/kevin/EMC/scripts/emc_setup.pl" >> ~/.bashrc

Copyright

Ramprasad Group, Georgia Tech, USA