https://medium.com/@monocosmo77/how-bellman-operators-work-part2-advanced-reinforcement-learning-d6843b72d631