Data for reinforcement learning at Periodic comes from its physical labs, which serve as the ultimate source of truth