Franz Pöschel

Machine-learned exchange-correlation functionals can now scale directly to condensed-phase matter in CP2K, matching diffusion Monte Carlo lattice energy accuracy across 13 ice phases while rivaling hybrid functionals in band-gap prediction.

Agentic AI could revolutionize software development, but it also introduces severe IT security risks that demand immediate attention.

Achieving a mean absolute deviation of just 0.020 kcal/mol from the Skala reference value, this implementation sets a new standard for accuracy in density-functional theory simulations.