Speaker
Costis Papageorgakis
(Queen Mary University London, UK)
Description
I will introduce CFTs at finite temperature, where the KMS condition replaces crossing symmetry as the central consistency constraint. Unlike at zero temperature, thermal OPE coefficients can have either sign, invalidating the linear functional approach. I will discuss neural networks as universal function approximators and explain how physics-informed neural networks (PINNs) can enforce the KMS condition directly. This sets the stage for the deep bootstrap framework presented in the main seminar.
