Science of Predictive Modeling

This is a Winter 2026 course. Here is the syllabus

This is a one-of-a-kind** course, and I will be posting some supplementary materials here.

**Context: There are excellent classes in physics, mathematics and computational science. This class sits the intersection of all three and focuses on predictive modeling as a science of its own. As a result, we will not be too physicsy or mathy or computy, but will try to sufficiently manage depth.

Some notes:
Circular Scatterers & Two Hard Disks : A great problem to learn a lot of concepts
Complete derivation of Boltzmann Equation for Hard Spheres  :
The Operator Viewpoint : Forward & Backward Kolmogorov / Perron-Frobenius & Koopman

Some videos:
Bernoulli & Poisson distributions and application to particle collisions
Information Theory Concepts
Introduction to Meso scale modeling
Boltzmann Equation insights
The Operator Viewpoint
Prof Kaushik Bhattacharya's (Caltech) lecture on Effective Properties

Some homework submissions:
Liouville/Fokker Planck/Backward Kolmogorov Operators Submission Notebook1 Notebook2