UCL COMP0123 Multi-agent Artificial Intelligence (2023/24)

This repository contains the first coursework I completed for my MSc module COMP0123 Multi-agent Artificial Intelligence.

The second coursework is a group research project on a self-selected multi-agent problem. Our group worked on the issue of prioritising agents in multi-agent path finding when conflicts occur. Checkout the mapf-priority-study repository for more details of the research project.

Tasks for coursework 1

1. Matrix Game (1)

   • Solved and analysed a matrix game by hand.

2. Matrix Game (2)

   • Solved a matrix game using NashPy.
   • Implemented the support enumeration algorithm to find Nash equilibria.

3. Repeated Game

   • Implemented several agents with the following strategy:
     • Grim Trigger
     • Tit for Tat
     • Limited Punishment
     • All Defect
   • Implemented a reinforcement learning agent using Q-learning with tabular value update.

4. Congestion Game

   • Solved and analysed a congestion game.

5. Bertrand Model of Oligopoly

   • Modelled market competition with 2 firms.
   • Simulated the Bertrand model and analysed results.

Python Environment

Requirement: python=3.11

pip install -r requirements.txt