Optimal Adaptive Control and Differential Games by Reinforcement Learning Principles