project-alphazero

Implementing an chess agent to play at an intermediate level.

This project aims to create a strong chess engine in python without the use of neural networks, just by using search algorithms and heuristics.

Currently the engine can search at speeds of 10-15k nodes per second, without the use of transposition tables.

Features implemented so far -

• Minimax Search
• Alpha Beta Pruning
• Move Ordering

Currently working on -

• Improving Iterative deepening search
• Adding zobrist hashing for transposition tables
• Implementing Quiescence search
• A simplified version of an opening book
• Piece square tables
• Monte Carlo Tree Search
• MVV-LVA (Most Valuable Victim - Least Valuable Aggressor)

To be implemented -

• Null move heuristic
• Killer heuristic
• Endgame table connections
• Search Statistics
• Tests (yes I have not done much to test my code)
• Static Exchange Evaluation (SEE)
• SPRT (Sequential probability ratio test) for evaluating the strength of my engine