Sarkas: A High-Performant Pure-Python Molecular Dynamics Code for Strongly Coupled Plasmas

Molecular dynamics (MD) simulations are a powerful technique to understand strongly coupled plasmas (SCPs) at the atomistic level, helping us understand the influence of micro-scale structural and temporal properties on the observed macroscopic properties of interest. Though there exists MD codes for such simulations, to our knowledge there isn't a pure-Python production-scale MD code with GPU implementation. This motivated us to develop Sarkas - a production-scale pure-Python open-source MD code for particles interacting through Coulomb and screened Coulomb potentials that are predominant particle interactions in SCPs. Sarkas is high-performant with execution speeds comparable to compiled languages (eg. C) due to the extensive use of Numpy arrays and a just-in-time compilation using Numba. Sarkas simulates 3D systems with the potential energy and forces computed using a highly efficient Particle-Particle Particle-Mesh algorithm. The systems of our interest are ultracold neutral plasmas, dusty plasmas, early stages of inertial confinement fusion and generic warm dense matter. The user-friendliness of Python combined with high performance enables Sarkas to serve as a useful design tool for experimentalists. (G. Dharuman’s present address is LLNL.)