OthelloBot is a self-hosted Discord Bot used to play othello in discord text channels against other players or a bot. It includes graphical interface to see the othello board and a database with statistics for each player.

/challenge @user

Challenges another user to an othello game. Another player can accept the challenge with the /accept command.

/challengebot level

Challenges the bot to an othello game. The bot can be level 1-6, each level using a different depth (for the bot to work on level 6 you need very good hardware)

/accept @user

Accept a challenge from a user.

/forfeit

Forfeits the game currently being played.

/move move

Make a move on the current game. Move format is column-row.

/view

View the current board state the game the user is playing, and all available moves.

/analyze level

Performs an analysis on the current game. Displays the bot's heuristic ranking for each move.

/stats

Fetches the stats for the current user. Displays rating, win rate, wins, losses, and draws.

/leaderboard

Shows the top users with the highest elo in the entire database.

/simulate Run a game between two bots real time in a text channel.

The evaluation algorithm uses a standard implementation of the Minimax algorithm with an Alpha-Beta pruning optimization. The Minimax algorithm is the backbone of the algorithm responsible for performing the search and deciding which nodes should be pruned.

The algorithm's heuristic evaluation function is responsible for deciding how good a board state is. The primary heuristic function is implemented as a combination of the 5 following specified heuristic functions

• Parity Heuristic: Measures the number of captured discs. Useful for closing out games. The weightage increases gradually as the game goes on.

• Corner Heuristic: Measures the number of captured corners, highly weighted at all stages of the game.

• XCSquare Heuristic: Measures the number of captured X squares and C squares, lowly weighted at all stages of the game.

• Mobility Heuristic: Measures the number of available moves, the weightage decreases gradually as the game goes on.

• Stability Heuristic: Measures the number of moves that cannot be flipped, highly weighted at all stages of the game.

All heuristic components are normalized to a weight between 100 and 0, with 100 being the highest weight and 0 being the lowest.

The Othello Board implementation uses 2 longs (8 bytes) (128 bits) to store the bit board. "White" is represented by 10, "Black" is represented by 01, and "Empty" (no disc) is represented by 00. Since each tile needs 2 bits to represent and there are a total of 64 tiles, we can represent the board using 64 * 2 = 128 bits.

This optimization provides both significant memory and time performance increases.

The algorithm keeps track of previously evaluated boards with a transposition table. The transposition table is implemented using a cache with a Deep2 replacement scheme. Each cache line has 2 buckets, one which is only replaced if the new board is found at a greater depth, the other is replaced if the first bucket is not.