简单程序，很能说明DQN的运行方式，通过深度网络和Qlearning的结合，训练使得最后小球能移动到最左边(Simple procedures, it can explain the way of DQN, through the combination of deep network and Qlearning, so that the last ball can move to the left.)