Brandon L. Butler
University of Michigan, Department of Chemical Engineering

Vyas Ramasubramani
University of Michigan, Department of Chemical Engineering

Joshua A. Anderson
University of Michigan, Department of Chemical Engineering

Sharon C. Glotzer
University of Michigan, Department of Chemical Engineering
University of Michigan, Department of Material Science and Engineering
University of Michigan, Department of Physics
University of Michigan, Biointerfaces Institute

Video: https://youtu.be/fIFPYZsOVqI

Abstract

HOOMD-blue is a library for running molecular dynamics and hard particle Monte Carlo simulations that uses pybind11 to provide a Python interface to fast C++ internals. The package is designed to scale from a single CPU core to thousands of NVIDIA or AMD GPUs. In developing HOOMD-blue version 3.0, we significantly improve the application protocol interface (API) by making it more flexible, extensible, and Pythonic. We have also striven to provide simpler and more performant entry points to the internal C++ classes and data structures. With these updates, we show how HOOMD-blue users will be able to write completely custom Python classes which integrate directly into the simulation run loop and analyze previously inaccessible data. Throughout this paper, we focus on how these goals have been achieved and explain design decisions through examples of the newly developed API.

Keywords

molecular dynamics, molecular simulations, Monte Carlo simulations, object-oriented

DOI

10.25080/Majora-342d178e-004

Bibtex entry

Full text PDF

Copyright © 2020 Brandon L. Butler et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.