Course 4
COURSE 4 · SIMULATION, AI & DATA PRACTICE
Contents
Every tutorial in the course, topic by topic. Within a topic the tutorials are listed alphabetically.
← Back to Course 4 — Simulation, AI & Data Practice
Topic 1 · Simulations
Monte Carlo, agent-based models, stochastic simulations, RL agents, tournaments. Topic page →
| Title | Description |
|---|---|
| Agent-based modeling for game theory in R | Build agent-based models where heterogeneous agents play games on networks, exploring emergent cooperation through update rules, spatial dynamics, and comparison with analytical equilibrium predictions. |
| Agent-based simulation of market dynamics — price discovery in a continuous double auction | Build an agent-based model of a continuous double auction with zero-intelligence, trend-following, and fundamentalist traders, simulate price discovery and convergence to competitive equilibrium, and visualize how agent composition affects market efficiency and volatility. |
| Cellular automata and spatial game theory | Implement spatial games on lattice grids where agents play the Prisoner’s Dilemma with neighbors, exploring strategy invasion dynamics, cluster formation, and the Nowak-May spatial chaos phenomenon. |
| Monte Carlo methods for equilibrium computation | Use Monte Carlo simulation in R to approximate Nash equilibria, estimate equilibrium payoffs under uncertainty, and validate analytical solutions through large-sample simulation experiments. |
| Replicator Dynamics Simulation for Three-Strategy Games | Large-scale replicator dynamics simulation for Rock-Paper-Scissors and other three-strategy games, comparing deterministic and stochastic dynamics across population sizes. |
| Spatial Prisoner’s Dilemma on a lattice — Nowak & May 1992 | Replicate the seminal Nowak & May (1992) spatial Prisoner’s Dilemma on a 2D lattice in R, showing how spatial structure sustains cooperation through cluster formation. |
Topic 2 · ML and Game Theory
AlphaZero, CFR, multi-agent RL, GANs, adversarial robustness. Topic page →
| Title | Description |
|---|---|
| Adversarial robustness as a game | Adversarial machine learning formulated as a two-player zero-sum game between a Defender (classifier) and an Attacker who perturbs inputs: minimax optimisation, adversarial training, and the effect on decision boundaries demonstrated with a 2D classification task. |
| Deep reinforcement learning for strategic games | Implement simple deep RL agents from scratch in R that learn game strategies through self-play, comparing learned policies with analytical Nash equilibria. |
| Fictitious play and convergence | Implement Brown’s fictitious play algorithm for 2x2 and 3x3 games, demonstrating convergence in zero-sum and potential games and the classic non-convergence cycling in Shapley’s counterexample. |
| GANs as a game-theoretic training process | Analyse generative adversarial networks through the lens of two-player zero-sum games, implementing a simple 1D GAN from scratch in R to study minimax formulation, training dynamics, mode collapse, and convergence conditions. |
| Multi-agent reinforcement learning in matrix games | Implement independent Q-learning agents that learn to play 2x2 matrix games, analyse convergence behaviour across game classes, and investigate when learning dynamics converge (or fail to converge) to Nash equilibria. |
| No-regret learning in games: from multiplicative weights to equilibrium | Implement the Multiplicative Weights Update algorithm and show that when all players use no-regret learning, the time-averaged strategy profile converges to a coarse correlated equilibrium. |
Topic 3 · AI/ML Foundations
From perceptrons to transformers — history, theory, and R implementations. Topic page →
| Title | Description |
|---|---|
| Automated Mechanism Design for Optimal Auctions | Training a data-driven auction mechanism using gradient-free optimisation in pure R — learning payment rules from bidder profiles, enforcing incentive compatibility and individual rationality, and benchmarking against Myerson’s optimal auction. |
| From Perceptron to deep learning — a historical R implementation | Trace the evolution of neural networks from Rosenblatt’s 1958 Perceptron through the Minsky-Papert critique to backpropagation and modern deep learning, implementing key milestones in R. |
| Generative adversarial networks as minimax games | Frame GANs as two-player zero-sum games in R, implement a simple GAN training loop for 1D data, visualise the generator-discriminator dynamics, and connect the equilibrium to Nash equilibrium theory. |
| Multi-armed bandits and the exploration-exploitation trade-off | Implementation and analysis of bandit algorithms including epsilon-greedy, UCB1, and Thompson sampling, framed as single-agent games against nature with regret analysis, Bayesian updating, and interactive performance comparisons. |
| SHAP Values as Shapley Values — Cooperative Game Theory Meets Explainable AI | Demonstrate that SHAP (SHapley Additive exPlanations) is a direct application of the Shapley value from cooperative game theory, computing exact Shapley values for a 4-feature prediction model by enumerating all coalitions and verifying the efficiency, symmetry, dummy, and additivity axioms in R. |
Topic 4 · Public APIs and Datasets
World Bank, ACLED, NOAA, arXiv, Lichess — live data for game-theoretic analysis. Topic page →
| Title | Description |
|---|---|
| Development Economics as Game Theory with World Bank Data | Modelling international development as strategic interaction — aid allocation auctions, trade as repeated prisoner’s dilemma, and technology transfer as signalling games, using simulated indicators inspired by World Bank data. |
| FRED economic data for game-theoretic calibration | Access Federal Reserve FRED data via the fredr API or direct CSV download to calibrate game-theoretic models with real-world economic parameters such as interest rates, GDP growth, and manufacturing output. |
| Game Theory Environments for Reinforcement Learning in R | Build OpenAI Gym-inspired game theory environments in pure R using R6 classes with reset(), step(), and render() methods. Implement Q-learning agents that learn Nash equilibrium strategies in Prisoner’s Dilemma and Hawk-Dove games. |
| Kaggle competitions as strategic interactions | Analyze Kaggle competitions through a game-theoretic lens, modeling competitor strategies, Nash equilibria of submission timing, and strategic model selection under competition rules. |
| World Bank WDI — economic indicators for game-theoretic analysis | Access the World Bank’s World Development Indicators via the WDI R package, explore GDP, trade openness, and military expenditure data relevant to international game-theoretic models. |
| World trade data for strategic analysis | Analyse international trade flows as strategic interactions using simulated trade network data, gravity model estimation, trade war scenarios, tariff equilibria, and welfare effects of trade agreements. |
Topic 5 · Real-World Data Applications
Empirical game theory with public historical data — trade, conflict, climate, health. Topic page →
| Title | Description |
|---|---|
| Airbnb host pricing as a strategic game | Model Airbnb host pricing as a local competition game with spatial competition on a network, price undercutting dynamics, differentiation strategies, and equilibrium pricing with heterogeneous listings. |
| Penalty kicks as a minimax game — testing mixed-strategy equilibrium with football data | Model penalty kicks as a zero-sum game between kicker and goalkeeper, compute minimax mixed strategies from realistic scoring data, and test equilibrium predictions against observed professional football frequencies. |
| Spectrum Auction Analysis: Revenue, Efficiency, and the Exposure Problem | Simulate and analyse multi-round ascending spectrum auction data inspired by FCC designs, computing winning prices, revenue efficiency, VCG benchmarks, bidder surplus, and diagnosing the exposure problem with complementary valuations. |
| The Cuban Missile Crisis as a signaling game | Model the 1962 Cuban Missile Crisis as a game-theoretic signaling game in R, analyzing Kennedy’s and Khrushchev’s strategic choices through incomplete information, credibility, and escalation dynamics. |
| The arms race as iterated Prisoner’s Dilemma — a Cold War analysis | Model the US-Soviet nuclear arms race as an iterated Prisoner’s Dilemma in R, simulate strategy profiles including MAD, compare with stylised historical spending patterns, and connect to the folk theorem. |
| Uber surge pricing as a dynamic game | Model ride-sharing surge pricing as a multi-player game between platform, drivers, and riders, simulating demand-supply dynamics, driver repositioning strategies, price equilibria, and welfare analysis. |
Topic 6 · Visualization & Communication
Publication-ready ggplot2, plotly, game trees, network graphs, dashboards. Topic page →
| Title | Description |
|---|---|
| Advanced ggplot2 Annotation Techniques for Game Theory Figures | Master publication-quality game theory diagrams with Nash equilibrium labels, best-response arrows, payoff region shading, and mathematical annotations using ggplot2. |
| Building Shiny Apps for Interactive Game Theory Exploration | Complete Shiny app code for a 2x2 game solver, Cournot duopoly simulator, and evolutionary dynamics visualiser, with static ggplot2 recreations for the tutorial. |
| Interactive dashboards for game theory with flexdashboard | Build interactive dashboards to explore game-theoretic models using flexdashboard, combining reactive inputs for payoff parameters with real-time equilibrium computation and multiple coordinated visualizations. |
| Interactive game theory dashboards with plotly | Build advanced interactive visualisations for game theory using plotly in R: animated replicator dynamics, hover-enabled payoff matrices, linked strategy-payoff views, and 3D expected-payoff surfaces over the mixed-strategy simplex. |
| Network visualization for games with igraph | Visualize game-theoretic interaction structures using igraph in R, including best-response networks, Nash equilibrium connectivity, and centrality analysis of strategic players. |
| Publication-ready ggplot2 figures with theme_publication() | Build a complete publication-ready ggplot2 theme from scratch, demonstrating the Okabe-Ito colorblind-safe palette, dual PDF/PNG export, and consistent styling across all figure types used in this site. |
Topic 7 · Reproducibility & Open Science
Quarto, renv, targets, Docker, pre-registration, data sharing. Topic page →
| Title | Description |
|---|---|
| Docker and renv for Reproducible Game Theory Environments | Create fully reproducible computational environments for game theory research using Docker containers and renv package management, with reproducibility audit functions and environment comparison workflows. |
| Docker containers for reproducible game theory analysis | Use Docker to create reproducible computational environments for game theory research, covering Dockerfile construction, R package management with renv, reproducibility workflows, and sharing analysis containers. |
| Literate programming for game theory research | Best practices for reproducible game theory analysis using Quarto and R Markdown, covering document structure, parameterized reports, code-figure integration, and version control workflows for research projects. |
| Preregistration and Pre-Analysis Plans for Game Theory Experiments | How to create structured pre-analysis plans for game theory experiments, demonstrating that researcher degrees of freedom can flip conclusions about cooperation rates. |
| Quarto parameterized reports for game theory | Use Quarto’s params feature to create reusable game theory analyses that render multiple game variants — Prisoner’s Dilemma, Battle of the Sexes, Chicken — from a single parameterized template. |
| Reproducible game theory research — a Quarto + renv workflow | Set up a fully reproducible game theory research project using renv for dependency management, Quarto for literate programming, and version control best practices for simulation-heavy analyses. |
Topic 8 · R Package Development
Package structure, roxygen2, testthat, pkgdown, CRAN submission. Topic page →
| Title | Description |
|---|---|
| Accelerating Game Solvers with Rcpp and Optimised R | Profiling game-theoretic computations in R, implementing support enumeration and fictitious play, presenting Rcpp C++ code for acceleration, and benchmarking pure-R optimisation strategies for performance-critical solvers. |
| Building a game theory R package from scratch | Walk through every step of creating a minimal R package that includes a nash_2x2() solver for 2x2 normal-form games, complete with roxygen2 documentation, testthat unit tests, and a devtools-powered workflow. |
| Designing S4 Classes for Game Theory Objects in R | Build a formal S4 class hierarchy for game theory in R with NormalFormGame, Strategy, and Equilibrium classes, including validity checks, show/summary methods, and Nash equilibrium solving via support enumeration. |
| Testing game theory code with testthat | Best practices for testing game-theoretic computations using testthat, including property-based testing of Nash equilibria, dominance checking, payoff validation, and test-driven development for game solvers. |
| Unit testing game theory solvers in R | Build robust unit tests for game-theoretic computations using testthat, covering Nash equilibrium verification, numerical tolerance, degenerate games, and property-based testing. |
| Writing vignettes for game theory R packages | Best practices for creating package vignettes that demonstrate game-theoretic computations, from structure and examples to testing and CRAN submission. |