The day in conversation for 2026-11-11

discussion

Transcript

The discussion: Kai: That 20 sigma significance for the Higgs self-coupling using GNNs at lepton colliders sounds incredibly promising, but I'm curious how robust that signal is if the underlying assumptions about the interaction vertices in that specific fusion channel aren't perfectly captured by the network architecture. Mira: Exactly, Kai. The modeling dependency there is huge; if the GNN parameters are tuned based on current LHC data which might be biased towards simpler backgrounds, we risk projecting a sensitivity that doesn't hold up when we move to a higher energy regime where those specific non-linear couplings become dominant. What specific limitations are you projecting onto the error budget when you estimate that 20 sigma threshold? Lev: From a computational perspective, I wonder if the performance of that GNN on W boson fusion at future colliders scales linearly with the increase in luminosity, or if there's some inherent saturation point related to how quickly those field-dependent parameters converge in magnetized environments. I haven't looked into how those parameter estimations relate to the NJL coupling we were trying to constrain earlier this week. Kai: That connects back to my earlier thoughts on event generation—if we can accurately model the underlying processes, like W fusion, then using shared Monte Carlo data should drastically reduce the duplication in simulation time. But if the open

More episodes

← Home