This project is a competitive project in course Introduction to Artificial Intelligence.

The Gomoku AIs are based on the Gomocup standard and you can create your own AI.exe through pisqpipe, the materials and document are here.

• The rule is: Free Style

We have three amazing guys here, Shun Zhang, Donghao Li and Pingxuan Huang.

• The Genetic Algorithm is supported by Pingxuan.
• The VCX Algorithm is supported by Donghao.
• The MCTS Algorithm and Minimax Algorithm are supported by me.
• The report is here.

Note that all the algorithms listed below can be packed into a win32 exe (less than 20m with numpy version 1.13.1), which is tested by us. We think the pisqpipe platform is far from perfect and there are too many unknown or stupid errors. So we just provide the terminal APIs. You can play with our AIs through command $python AI-name.py(more details please refer to relating page). Still, one can easily pack the AIs into win32 exe with the 'help' of pisqpipe.

• Monte-Carlo Tree Search
• Genetic Algorithm
• Minimax Search
  • VCX
  • Alpha-beta pruning
  • Zobrist transition table
  • dynamic updating with pattern score

More detailed document please refer to relating page.

Also, there is a python version of Threat Space Search, which is designed by Xavier Wee. However, we found it too slow for a board as big as 20 x 20. Optimizations are required.