Ultra-robust machine-learning models capable of stable molecular simulations at extreme temperatures

Researchers have created a groundbreaking physics-informed machine-learning model that can run molecular simulations for unprecedented lengths of time, even at temperatures as high as 1000 Kelvin.