Optimal Reinforcement Learning for Gaussian Systems