Optimal Execution Using Reinforcement Learning