Multiagent Reinforcement Learning: Theoretical Framework and an Algorithm