A chess engine made entirely in C++ utilizing negamax, iterative deepening, quiescence search, transposition tables and move ordering using MVV-LVA (Most valuable victim-Least valuable aggressor).
It was able to achieve 1819 ELO on rapid matches on lichess which says that it's better than 82.7% of rapid players on lichess.
Here's the Ai's lichess account: https://lichess.org/@/TheShadowEngine
It was uploaded to render's cloud services as a docker container and played 3,456 matches before being suspended.
Copy the repository to your local machine:
git clone https://github.com/ISMAILGAMAL/Chess-engine.gitCompiling the code would give you the ai as a console application that supports uci you would still need a gui to run it.
The engine is capable of generating both pseudo-legal and legal moves for all pieces in chess including special moves like en-passants, castling and promotions but for promotions it assumes that pawns only promote to queens for simplicity
A chess bot is composed of two parts first the search and then the evaluation as the chess bot works by simulating moves up to a certain search depth. Then, evaluating the resuling positions. After that, it chooses the most promising moves based on the evaluation heuristic.
Source: https://chrisbutner.github.io/ChessCoach/high-level-explanation.html
The Minimax algorithm is a graph search algorithm that works by recursively exploring the game tree to a certain depth, evaluating each possible leaf or terminal node using some heuristic. It alternates between maximizing and minimizing players, hence the name "minimax."
My implementation uses alpha-beta pruning which is a tree pruning technique that reduces the number of nodes the algorithm needs to explore without affecting the final outcome improving the efficiency of the search.
It also uses iterative deepening which allows the bot to stop at any time by doing a search up to a depth of one then to a depth of two and so on until a certain time limit the bot calculated based on the remaining time the gui gives it returning the best move from the last successful seacrh.
Quiescence search is implemented to make sure the evaluation is reliable and transposition tables are used to store the evaluations found in previous parts of the search and removing redundant searches further optimizing the search.
The evaluation uses a simplified evaluation function proposed by Tomasz Michniewski, which consists of two parts: material piece values and piece square tables. I only used the material values from here. which works by giving each piece a value ex: white rook -> 500, black queen -900 and then adds up all the material values on the board
I then used the piece square tables from PeSTO's Evaluation Function.
The ai runs lichess-bot found here: https://github.com/lichess-bot-devs/lichess-bot
Which basically does all the matchmaking and interfacing with the website,