Each 75-minute block is three to five short decks rather than one long one, so each can be
read in a sitting. The syllabus says which decks belong to which
session; the labs page has the matching notebooks.
Every number on these slides came from code that ran. The computational
sessions are generated from a measurement ledger — nothing is quoted from a textbook without
being reproduced, and where a source turned out to be wrong, the slide says so.
Session 1 — Introduction and the growth model Available
2.A2 — Aiyagari: Production and the Capital MarketPDF · 16 pp.
2.A3 — The Stationary Distribution — and Where the Model Goes NextPDF · 14 pp.
Block B
2.B1 — Krusell–Smith: The Distribution as a StatePDF · 16 pp.
2.B2 — Overlapping Generations: Finite LivesPDF · 16 pp.
2.B3 — Optimal Policy: Ramsey and Time ConsistencyPDF · 15 pp.
Decks are posted session by session, a few days before each class.
Later sessions will appear here as the course reaches them.
Session 3 — Computation and programming Available
October 8, 2026 · 9 decks · what it means to solve a model; VFI and time iteration; the Python, NumPy and PyTorch toolkit; the data workflow; a hands-on · Discuss
Block A
3.A1 — What Does It Mean to Solve a Model?PDF · 20 pp.
3.A2 — Value Function Iteration: The First Complete CrossingPDF · 17 pp.
3.A3 — Euler Equations, Time Iteration, and What Is Still MissingPDF · 17 pp.
3.A4 — Worked Example: The RBC Model, Fully DiscretizedPDF · 15 pp.
Block B
3.B1 — Your Toolkit: Python, Environments, and Version ControlPDF · 22 pp.
3.B3 — PyTorch from the Ground Up, and Coding with AIPDF · 16 pp.
3.B4 — The Data Workflow: Acquire, Clean, Persist, ExportPDF · 23 pp.
3.B5 — Hands-On: From a Closed Form to a Calibrated Economy (the steps; the solution follows in class)PDF · 13 pp.
Session 4 — Numerical methods I Available
October 15, 2026 · 6 decks · root-finding and optimization; where derivatives come from; approximation and interpolation; expectations and the Markov chain; how each speed-up works; the worked example · Discuss
Block A
4.A1 — Root-Finding and Optimization on an IntervalPDF · 18 pp.
4.A2 — Optimization in Many Dimensions, and Where Derivatives Come FromPDF · 31 pp.
Block B
4.B1 — Function Approximation and InterpolationPDF · 32 pp.
4.B2 — Expectations: Quadrature, and Discretizing the ShockPDF · 22 pp.
4.B3 — Fast and Accurate: How Each Improvement WorksPDF · 18 pp.
4.B4 — Worked Example: The Stochastic RBC Model, ProperlyPDF · 15 pp.
Session 5 — Numerical methods II In preparation
October 22, 2026 · perturbation and projection; one model, four methods · Discuss
Session 6 — Heterogeneous agents I and II In preparation
October 29, 2026 · Block A: Aiyagari as a nested fixed point, Krusell–Smith, OLG, optimal policy; Block B: when heterogeneity aggregates — Negishi and transitional dynamics · Discuss
Session 7 — Deep learning for macroeconomics In preparation
November 5, 2026 · machine learning essentials; neural networks as approximators; deep learning for dynamic models · Discuss
Sessions 8–10 In preparation
Reinforcement learning and heterogeneous agents, language models, and agentic AI.
Posted as each session is built. Discuss:
Session 8 ·
Session 9 ·
Session 10.