Research on parallel optimization of UCT algorithm for NoGo computer game
Zhang Zi-yang, Yuxia Sun · 2017
The theoretical system and evaluation methods of NoGo is not mature for it is been proposed in recent years. Recently UCT Algorithm has become a popular approach for NoGo game. But UCT Algorithm based on Monte-Carlo Tree Search only runs on a single thread, which does not make full use of multi-core CPU performance. In this paper, a multi-thread parallel optimization method of UCT algorithm is put forward. The experimental results show that the parallel optimized UCT algorithm enhances researching efficiency for NoGo, and makes the decisions of the best move for NoGo faster and more accurate. Meanwhile, it also increases the skill of NoGo game.