Abstract:
Hadronically decaying τ-leptons (τ's) are a key experimental signature at the LHC, contributing to a wide range of Standard Model (SM) measurements and Beyond the Standard Model (BSM) searches. Although τ's offer enhanced sensitivity in scenarios with mass-proportional couplings, their reconstruction and identification ("tagging") is challenging due to complex decay signatures and substantial QCD contamination - requiring dedicated treatment. In parallel, growing interest in low-mass BSM scenarios - where conventional τ-tagging methods are often limited - motivates complementary strategies that can operate in collimated di-τ topologies. Addressing these issues is essential for extending sensitivity to SM observables and to otherwise hard-to-probe regions of BSM parameter spaces in final states with τ's.
This talk discusses recent developments in hadronic τ-tagging and their application to low-mass BSM searches with the ATLAS detector. I will present a Graph Neural Network-based τ-identification algorithm (GNTau) for Run-3, describe the tagging of collimated di-τ systems for low-mass searches along with their associated identification efficiency measurement in SM Z+γ events using Run-2 data, and highlight the method's application in two Run-2 analyses targeting low-mass scenarios - both yielding the first ATLAS limits in their respective final states. Time permitting, I will also briefly show additional comparative machine-learning studies and ongoing validation results.
Biography:
Nadav is completing a PhD in experimental high-energy particle physics as part of the ATLAS collaboration at Tel-Aviv University. His research focuses on τ-tagging and searches for BSM physics. Nadav has worked on analyses targeting low-mass (pseudo)scalar resonances, led development efforts to improve the τ-tagging capabilities and the synergy of hadronic identification tasks in ATLAS, and conducted research on advanced ML-based classification techniques. Apart from BSM and jet physics, his broader interests include Higgs-sector measurements, SM-EFT interpretations, and the applications of ML in high-energy physics.
Zoom direct link:
https://cern.zoom.us/j/4042031778?pwd=N0RlZ0l3dGtEYk9zT25OWmlwVklSZz09
Alice Ling Lin