Operator World Models for Reinforcement Learning

However, it is not directly applicable to Reinforcement Learning (RL) due to the inaccessibility of explicit action-value functions.