KalScope is an open-source Freestyle Gomoku AI, written during Jan 2014 ~ Mar 2014. KS was an exercise project for training my program optimize techniques initially, but reached pretty good competitiveness.

Kai Sun have linked this project in his blog thus I would keep this link valid.

You can get source and compiled x64 binary at GitHub. The source code is kind of messy and buggy, I wrote several posts to explain how KS works, at my old blog site. I am trying to migrate my old posts to this new site.

KS is built upon classic alpha-beta search framework. The core innovation of KalScope is called “incremental evaluation”. Beside the widely-used zobrist hash keys, KS incrementally maintains the board presents and evaluations, cutting the time cost of evaluation to negligible.