Semi-Empirical Shadow Molecular Dynamics: A PyTorch Implementation

• DOI: 10.1021/acs.jctc.3c00234
• Publication Date: 2023-05-11T00:50:33Z
• Authors: Maksim Kulichenko; Kipton Barros; Nicholas Lubbers; Nikita Fedik; Guoqing Zhou; Sergei Tretiak; Benjamin Nebgen; Anders M. N. Nikl...
• Keywords: Chemical Physics (physics.chem-ph); Physics - Chemical Physics; FOS: Physical sciences; Computational Physics (physics.comp-ph); Physics - Computational Physics; 01 natural sciences; 0104 chemical sciences

Extended Lagrangian Born-Oppenheimer molecular dynamics (XL-BOMD) in its most recent shadow potential energy version has been implemented the semiempirical PyTorch-based software PySeQM. The implementation includes finite electronic temperatures, canonical density matrix perturbation theory, and an adaptive Krylov subspace approximation for integration of equations motion within XL-BOMB approach (KSA-XL-BOMD). PyTorch leverages use GPU machine learning hardware accelerators simulations. new XL-BOMD formulation allows studying more challenging chemical systems with charge instabilities low gaps. current public release PySeQM continues our development modular architecture large-scale simulations employing semi-empirical quantum-mechanical treatment. Applied to dynamics, simulation 840 carbon atoms, one time step executes 4 s on a single Nvidia RTX A6000 GPU.