arm algorithm reinforcement-learning simulation monte-carlo rank thompson-sampling reinforcement-learning-algorithms ucb reward multi-armed-bandit montecarlo-simulation contextual-bandits ranking-algorithm mab ranked-mab