Reinforcement Learning Applied to Linear Quadratic Regulation