By Dr. Kausik Ghosh, King’s College London.

Abstract: The conformal bootstrap provides a powerful nonperturbative framework for studying strongly coupled quantum field theories using symmetry, unitarity, and crossing symmetry. In this talk, I will review recent developments in analytic functional approaches to the numerical bootstrap and present new results in this direction. I will then discuss how modern machine-learning ideas can open new ways of navigating the space of consistent conformal field theories. In particular, I will describe the use of neural networks to construct crossing-symmetric correlators from limited physical input, and argue that their implicit preference for simple, smooth solutions may point toward new organizing principles underlying bootstrap constraints.