Deep Learning for Solving Dynamic Stochastic Model

Lecturer: Professor Dr Simon Scheidegger (University of Lausanne)
Date: May 22-24, 2024
Venue: Leipzig University, Institutsgebäude, Raum I 411, Grimmaische Straße 12, 04109 Leipzig
Registration: until May 7th, 2024 via email: sprenger@wifa.uni-leipzig.de.

The course is designed for at most 20 participants.

Time Schedule
Deep Learning for Solving Dynamic Stochastic Models (May 22 – 24, 2024)
Total hours: 18

Wednesday, May 22
9:30 a.m. – 11:00 a.m. (2 x 45 min)
Introduction to Machine Learning and Deep Learning for Dynamic Stochastic Economic Models
Break: 30 min
11:30 a.m. – 1:00 p.m. (2 x 45 min)
A hands-on session on Deep Learning
Break: 90 min
2:30 p.m. – 4:00 p.m. (2 x 45 min)
Deep Equilibrium Nets

Thursday, May 23
9:30 a.m. – 11:00 a.m. (2 x 45 min)
Hands-on: Deep Equilibrium Nets (I) – solving a simple DSGE model
Break: 30 min
11:30 a.m. – 1:00 p.m. (2 x 45 min)
Hands-on: Deep Equilibrium Nets (II) – solving a simple DSGE model with nonlinearities
Break: 90 min
2:30 p.m. – 4:00 p.m. (2 x 45 min)
Uncertainty Quantification and Estimation for nonlinear models: Surrogate models

Friday, May 24
9:00 a.m. – 10:30 a.m. (2 x 45 min)
Introduction to (stochastic) Integrated Assessment Models
Break: 30 min
11:00 a.m. – 12:30 p.m. (2 x 45 min)
Solving the (non-stationary) DICE model with Deep Equilibrium Nets
Break: 90 min
2:00 p.m. – 3:30 p.m. (2 x 45 min)
Deep Uncertainty Quantification for stochastic integrated assessment models