M1: Foundations
We begin with the classical vocabulary—normal-form games, game trees, Nash and correlated equilibria, and no-regret learning—that every game-playing agent is built on. You will see how minimax and alpha-beta pruning power chess engines, and how no-regret learning connects to the correlated equilibria that multi-agent LLM systems naturally converge toward. In PA1, you will put language models into normal-form games and empirically estimate their payoff matrices.
Module preview
We begin with the classical vocabulary—normal-form games, game trees, Nash and correlated equilibria, and no-regret learning—that every game-playing agent is built on. You will see how minimax and alpha-beta pruning power chess engines, and how no-regret learning connects to the correlated equilibria that multi-agent LLM systems naturally converge toward. In PA1, you will put language models into normal-form games and empirically estimate their payoff matrices.
Lectures and materials
L1: Introduction to Game Theory
Normal-form games, utilities, and strategic reasoning.
L2: Chess and Game Trees
Minimax, alpha-beta pruning, and game-tree search.
L3: Nash Equilibria
Best responses, mixed strategies, and equilibrium existence.
L4: Correlated Equilibria
Mediators, no-regret learning, and multi-agent convergence.
Demo Notebooks
These notebooks support the Module 1 demonstrations. Download them to run locally after configuring your TAMU AI API access.
Programming assignment
PA1: Empirical Payoff Estimation and Equilibrium Analysis · Due September 20 at 11:59 PM CDT