2026/08/03 by Luigi Favaro, Tilman Plehn, Huilin Qu +1
Physics and Astronomy · #hep-ph #hep-ex
32 pages, 18 figures, 6 tables
arxiv created 2026/08/03 · arxiv updated 2026/08/05
We study for the first time the benefit of Lorentz-equivariant transformers for large-size jet tagging and flavor tagging. To control their computing demands, we optimize all implementations for inference cost metrics. In our scaling studies, we find that Lorentz-equivariant networks outperform standard transformers, provided geometric features are relevant. This holds true in an idealized world as well as for limited resources. The conditional gain from Lorentz equivariance provides interesting input to the development of foundation models for LHC data.