HOOMD-blue v3.0: Creating a modern and extensible API for molecular dynamics and Monte Carlo particle simulations

HOOMD-blue is a general-purpose toolkit that performs molecular dynamics and Monte Carlo simulations of particles on CPUs and GPUs [1]. In this talk, we describe the design and implementation of the next major release of HOOMD-blue, v3.0. This release includes a new object-oriented Python interface, seamless integration with Python packages like NumPy, increased user extensibility within Python, and a flexible system for accessing and storing computed quantities. For instance, with HOOMD-blue’s v3.0 MoSDeF integration, a simulation’s initial conditions, atom-typing, and force field can be generated through MoSDeF and the simulation run through HOOMD within a single Python script. Version 3.0 changes enable a variety of new simulation methods such as hybrid MC/MD schemes, machine learned force fields, and advanced sampling schemes such as umbrella sampling. Furthermore, custom actions, which allow Python code to be injected into HOOMD’s run loop, streamline the prototyping and developing new simulation techniques and methods for HOOMD users. HOOMD-blue v3.0 is designed with the future of molecular simulations in mind with extensions of all kinds anticipated and facilitated through our Python and C++ APIs.

[1] https://doi.org/10.1016/j.commatsci.2019.109363