發送短信: Exploration and value function factorisation in single and multi-agent reinforcement learning