Summarizing the Q-learning algorithm