BEGIN:VCALENDAR
VERSION:2.0
CALSCALE:GREGORIAN
PRODID:UW-Madison-Physics-Events
BEGIN:VEVENT
SEQUENCE:0
UID:UW-Physics-Event-9842
DTSTART:20260918T140000Z
DTEND:20260918T160000Z
DTSTAMP:20260904T171423Z
LAST-MODIFIED:20260902T163052Z
LOCATION:5280 Chamberlin
SUMMARY:Intersecting AI and high energy theoretical physics\, Prelimin
 ary Exam\, Haotian Cao\, Physics PhD Graduate Student
DESCRIPTION:Artificial intelligence (AI) is increasingly advancing res
 earch in theoretical physics and pure mathematics. In this talk\, I'll
  present two uses of AI in the context of conformal field theory and s
 cattering amplitudes.<br>\n<br>\nWe first discuss an inverse problem e
 merging in two-dimensional conformal field theory (CFT): reconstructin
 g tensor products of CFTs from their low-lying spectra. We specificall
 y focus on rational conformal field theories (RCFTs) such as Wess-Zumi
 no-Witten models characterized by affine Kac-Moody algebras. The discr
 ete and structured nature of this problem for RCFTs motivates a sequen
 ce-to-sequence Transformer architecture that maps spectral information
  to the constituent affine algebras and central charges. We further de
 monstrate the model's ability to generalize to theories with larger ce
 ntral charges and to theories with unseen affine algebras during train
 ing.<br>\n<br>\nWe then turn to the amplitude bootstrap for N=4 super 
 Yang-Mills theory (SYM) in the planar limit. We use agentic AI workflo
 w to prove a mathematical conjecture related to the construction of a 
 good basis in this program. Rather than relying solely on informal mat
 hematical proofs\, we use a systematic and iterative agentic AI workfl
 ow that translates the conjecture into formal mathematical statements\
 , identifies the required intermediate lemmas\, and constructs and ver
 ifies their proofs.
URL:https://www.physics.wisc.edu/events/?id=9842
END:VEVENT
END:VCALENDAR