Atmospheric Long-Lived Particle Search: Harnessing Cross-Disciplinary Techniques from Cosmic Ray and Deep Learning Physics

Bhattacharya et al. (2025) outlined how ultra-high energy cosmic rays could produce long-lived stau tracks in the atmosphere, yet current detection relies on methods adapted from neutrino searches. I propose developing new cross-disciplinary detection frameworks that merge atmospheric cosmic-ray modeling, advanced deep learning (e.g., graph neural networks for track reconstruction), and real-time anomaly detection to specifically enhance sensitivity to non-standard long-lived particle signatures in large-volume detectors like IceCube-Gen2. By exploiting directional, temporal, and energy correlations unique to atmospheric stau production, and leveraging ML to distinguish these from neutrino backgrounds, we could significantly increase discovery potential. This approach transcends traditional event selection and synthesizes techniques from astroparticle, collider, and computational physics.

References:

1. Beyond standard model particles in the atmospheric flux: a long-lived stau example. A. Bhattacharya, M. Reno, I. Sarcevic (2025). Journal of Cosmology and Astroparticle Physics.
2. Unravelling physics beyond the standard model with classical and quantum anomaly detection. Julian Schuhmacher, Laura Boggia, Vasilis Belis, E. Puljak, M. Grossi, M. Pierini, S. Vallecorsa, F. Tacchino, P. Barkoutsos, I. Tavernelli (2023). Machine Learning: Science and Technology.
3. Searches for beyond-standard-model physics with astroparticle physics instruments. M. Ackermann, K. Helbing (2023). Philosophical Transactions of the Royal Society A.