haroldsultan/MCTS

What it solves

This project provides a Python implementation of the Monte Carlo Tree Search (MCTS) algorithm, allowing users to experiment with how the algorithm makes decisions in a state-based game environment.

How it works

The code implements MCTS to solve a custom game where a player must choose numbers from a specific range at each turn to keep a running total as close to zero as possible. The algorithm simulates multiple paths to evaluate the best move, accounting for the fact that early mistakes are more costly and moves do not commute.

Who it’s for

Developers and AI students who want a simple, concrete implementation of MCTS to study the algorithm's behavior and experimentation.

Highlights

• Python-based implementation for experimentation.
• Includes a toy game environment to demonstrate the algorithm's and search tree's effectiveness.
• Configurable simulation counts and decision levels via command-line arguments.