1-2-3 Seminar: All you need is $\nabla \log u$

The usual approach to solving the heat equation and its various Fokker-Planck friends is SDE Monte Carlo simulation. In this talk we will see a (mostly) deterministic algorithm that relies on approximating the gradient-log pdf. The talk will feature a little bit of everything except AG: PDEs, optimal transport, probability, neural networks, and algorithms.

Zoom Link: https://washington.zoom.us/j/92849568892