/PBNN

Code for training and running reactive molecular dynamics and QM/MM simulations

Primary LanguagePython

PBNN

doi:10.1063/5.0063187

PBNN is a reactive molecular dynamics code designed to simulate reactions in the condensed phase. The Hamiltonian is based on the EVB model and uses a combination of physics-based force field terms (called with OpenMM) and neural networks (called with PyTorch). The package can currently be used to run QM/MM as well, and has features available for training the PBNN Hamiltonian to QM/MM data (see the Tutorial directory).

Table of contents

Install

To install, there are a few packages that need to be loaded. Make sure you have Python >=3.9.0. Then run the following commands

pip install -r requirements.txt
conda install -c conda-forge openmm
conda install -c conda-forge ase
conda install -c conda-forge MDAnalysis
conda install -c conda-forge mdtraj

Next, intall SchNetPack from my own repo. Follow the instructions for installing Plumed Finally, install Psi4 to run QM/MM. You can find instructions for installing Psi4 in the docs directory I have included in this package.