Authors: Pratik Nag, Andrew Zammit-Mangion, Noel Cressie, Sumeetpal Singh

Presentation slides are available here

This talk presents the use of Fourier neural operators (FNOs) for constructing statistical dynamical spatio-temporal (DST) models for forecasting. An FNO is a flexible mapping of functions that approximates the solution operator of possibly unknown linear or nonlinear partial differential equations (PDEs) in a computationally efficient manner, so explicit knowledge of an underlying PDE is not required. Related paper: Technometrics (2026).