Project Overview

This project explores the implementation of Adversarial Search within a discrete, zero-sum environment. By comparing Stochastic (Random) move selection with Deterministic (Minimax) optimization, this agent demonstrates the power of state-space exploration in game theory.

Core Features

Matrix-Based State Management: The board is represented as a 3×3 nested list, ensuring O(1) access time for state updates using board[row][column].

Hybrid Interface: Features a standard CLI for traditional execution and an Event-Driven GUI using ipywidgets for interactive testing in Jupyter.

Dynamic Game Modes:

    Stochastic Mode: Computer utilizes randrange() for non-heuristic move selection.

    Minimax Mode: An unbeatable agent utilizing recursive backtracking.

    PvP Mode: Local Human-vs-Human state tracking.

Artificial Intelligence Logic

1. The Minimax Algorithm

The agent’s "intelligence" is derived from the Minimax Algorithm, a recursive search through the game's state space. The agent treats the game as a tree where each node represents a board configuration.

Maximizing Player (AI): Searches for the move sequence that leads to the highest possible terminal value.

Minimizing Player (Human): The algorithm assumes the opponent will play optimally to minimize the AI's utility.

Backtracking: The agent "simulates" moves, evaluates the result, and "undoes" the move to explore alternative branches, ensuring an exhaustive search of all 255,168 possible game positions.

Terminal State Utilities:

+1: AI Victory (Maximum Utility)

−1: Human Victory (Minimum Utility)

0: Draw (Neutral Utility)

2. Strategic Opening Heuristic

To optimize the search and adhere to laboratory constraints, the agent is programmed with a Center-Square Priority.

Logic: The center square (Index 5) is the most strategically significant cell, contributing to four possible winning vectors (Horizontal, Vertical, and two Diagonals). By claiming this "High-Ground" immediately, the agent reduces the human player's potential branching factor from the start.

Technical Stack & Usage

Prerequisites

Environment: Python 3.x, Jupyter Notebook / Lab

Libraries: ipywidgets (for GUI), random (for stochastic testing)

Execution

Initialize: Run all cells in the .ipynb file to load the backend logic and Minimax engine.

Interface: Scroll to the final cell to access the interactive GUI.

Selection: Use the dropdown menu to toggle between AI difficulty levels.