Accurate Large-Scale Atomistic Simulations of Dynamic Compression

For many decades, atomistic simulation has provided unique insight into the physics of dynamic compression of condensed matter. Simple physics-based interatomic potentials have provided fundamental insights into phenomena such as plasticity, spall, and phase change kinetics, albeit at the qualitative level. Now, in the era of machine learning potentials trained on abundant quantum electronic structure simulations, it is possible to make quantitative predictions about specific materials and under specific conditions. In this talk I will present results for recent very accurate, very large-scale simulations of plasticity and phase change kinetics running on leadership computing platforms. I will also present a new analysis of shockwave structure for the Lennard-Jones fluid running on my laptop.