README.md

Backgammon AI Simulator

This project allows you to create AI players that can play Backgammon against each other using the game engine directly.

Architecture

IBackgammonAI Interface

All AI players must implement the IBackgammonAI interface:

public interface IBackgammonAI
{
    string Name { get; }
    List<Move> ChooseMoves(GameEngine engine);
    bool ShouldAcceptDouble(GameEngine engine);
    bool ShouldOfferDouble(GameEngine engine);
}

The AI has full access to the GameEngine to inspect:

• Board state (engine.Board)
• Current player (engine.CurrentPlayer)
• Available moves (engine.GetValidMoves())
• Remaining dice (engine.RemainingMoves)
• Opponent information (engine.GetOpponent())

Included AI Implementations

RandomAI - Makes random valid moves. Useful as a baseline for testing more sophisticated AIs.

Usage

Running Simulations

cd Backgammon.AI
dotnet run

The program will:

1. Ask how many games to simulate
2. Run the simulation between two AIs
3. Display win statistics
4. Show one example game with verbose output

Creating Your Own AI

Create a new class implementing IBackgammonAI:

public class MySmartAI : IBackgammonAI
{
    public string Name => "MySmartAI";

    public List<Move> ChooseMoves(GameEngine engine)
    {
        var chosenMoves = new List<Move>();
        
        while (engine.RemainingMoves.Count > 0)
        {
            var validMoves = engine.GetValidMoves();
            if (validMoves.Count == 0) break;

            // Your strategy here
            var bestMove = PickBestMove(validMoves, engine);
            
            if (engine.ExecuteMove(bestMove))
                chosenMoves.Add(bestMove);
            else
                break;
        }
        
        return chosenMoves;
    }

    public bool ShouldAcceptDouble(GameEngine engine)
    {
        // Your doubling strategy
        return true;
    }

    public bool ShouldOfferDouble(GameEngine engine)
    {
        return false;
    }

    private Move PickBestMove(List<Move> moves, GameEngine engine)
    {
        // Implement your move selection logic
        return moves[0];
    }
}

Using AISimulator Programmatically

// Create AIs
var ai1 = new RandomAI("AI-1");
var ai2 = new MySmartAI();

// Create simulator
var simulator = new AISimulator(ai1, ai2, verbose: false);

// Run a single game
var result = simulator.RunGame();
Console.WriteLine($"Winner: {result.Winner}, Points: {result.Points}");

// Run multiple games
var stats = simulator.RunSimulation(1000);
stats.PrintSummary();

Strategy Ideas

Some ideas for creating more sophisticated AIs:

1. Defensive AI - Avoid leaving blots, prioritize making points
2. Aggressive AI - Hit opponent's blots when possible
3. Racing AI - Focus on pip count, optimize bearing off
4. Positional AI - Evaluate board positions with heuristics
5. Monte Carlo AI - Simulate future positions to evaluate moves
6. Neural Network AI - Train on game data

Game Engine Access

The AI has access to these key methods:

• engine.GetValidMoves() - Get all legal moves for current state
• engine.ExecuteMove(move) - Execute a move
• engine.Board.GetPoint(position) - Inspect any point on the board
• engine.CurrentPlayer - Current player info
• engine.GetOpponent() - Opponent player info
• engine.RemainingMoves - Available dice to use

The AI doesn't need to:

• Handle dice rolling (simulator does this)
• Manage turn switching (simulator does this)
• Validate moves manually (use GetValidMoves())