Gun-Wall Game — Simulation Dashboard

MAS Course Final Project: Strategic Adaptation in the Gun-Wall Game under Partial Observability

Quick Start

# 1. Install dependencies
pip install flask numpy

# 2. Run the server
python app.py

# 3. Open in browser
# http://localhost:5000

Project Structure

project/
├── app.py                 # Flask entry point
├── config.py              # Game constants, payoff tables, solver params
├── engine/
│   ├── game.py            # Core rules: legal_actions, outcome, transitions
│   ├── belief.py          # Bayesian belief updates
│   ├── solver.py          # IBR solver with backward induction
│   ├── simulation.py      # Monte Carlo episode runner
│   └── personas.py        # Persona definitions (cautious/aggressive/balanced)
├── api/
│   └── routes.py          # REST API: /api/solve, /api/simulate, /api/personas
├── static/                # CSS + JS for the dashboard
└── templates/
    └── index.html         # Single-page dashboard (3 tabs)

Dashboard Tabs

1. Simulation — Pick P1/P2 personas, run N episodes, view outcome distributions
2. Game Replay — Step through a single episode round-by-round with belief evolution
3. Policy Viz — Plot action probabilities vs belief p for each round

API Endpoints

Method	Path	Purpose
GET	/api/personas	List available personas
POST	/api/solve	Run IBR solver for a persona pair
POST	/api/simulate	Run N episodes, return stats + episodes

Team

Name	Role
Tom Badash	Theory Lead
Rom Sheynis	Development Lead
Lioz Shor	Analysis and Research Lead