Lorentz Local Canonicalization: How to Make Any Network Lorentz-Equivariant

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Spinner, Jonas, Favaro, Luigi, Lippmann, Peter, Pitz, Sebastian, Gerhartz, Gerrit, Plehn, Tilman, Hamprecht, Fred A.
Format:	Preprint
Published:	2025
Subjects:	Machine Learning High Energy Physics - Phenomenology
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

_version_	1866909866686152704
author	Spinner, Jonas Favaro, Luigi Lippmann, Peter Pitz, Sebastian Gerhartz, Gerrit Plehn, Tilman Hamprecht, Fred A.
author_facet	Spinner, Jonas Favaro, Luigi Lippmann, Peter Pitz, Sebastian Gerhartz, Gerrit Plehn, Tilman Hamprecht, Fred A.
contents	Lorentz-equivariant neural networks are becoming the leading architectures for high-energy physics. Current implementations rely on specialized layers, limiting architectural choices. We introduce Lorentz Local Canonicalization (LLoCa), a general framework that renders any backbone network exactly Lorentz-equivariant. Using equivariantly predicted local reference frames, we construct LLoCa-transformers and graph networks. We adapt a recent approach for geometric message passing to the non-compact Lorentz group, allowing propagation of space-time tensorial features. Data augmentation emerges from LLoCa as a special choice of reference frame. Our models achieve competitive and state-of-the-art accuracy on relevant particle physics tasks, while being $4\times$ faster and using $10\times$ fewer FLOPs.
format	Preprint
id	arxiv_https___arxiv_org_abs_2505_20280
institution	arXiv
publishDate	2025
record_format	arxiv
spellingShingle	Lorentz Local Canonicalization: How to Make Any Network Lorentz-Equivariant Spinner, Jonas Favaro, Luigi Lippmann, Peter Pitz, Sebastian Gerhartz, Gerrit Plehn, Tilman Hamprecht, Fred A. Machine Learning High Energy Physics - Phenomenology Lorentz-equivariant neural networks are becoming the leading architectures for high-energy physics. Current implementations rely on specialized layers, limiting architectural choices. We introduce Lorentz Local Canonicalization (LLoCa), a general framework that renders any backbone network exactly Lorentz-equivariant. Using equivariantly predicted local reference frames, we construct LLoCa-transformers and graph networks. We adapt a recent approach for geometric message passing to the non-compact Lorentz group, allowing propagation of space-time tensorial features. Data augmentation emerges from LLoCa as a special choice of reference frame. Our models achieve competitive and state-of-the-art accuracy on relevant particle physics tasks, while being $4\times$ faster and using $10\times$ fewer FLOPs.
title	Lorentz Local Canonicalization: How to Make Any Network Lorentz-Equivariant
topic	Machine Learning High Energy Physics - Phenomenology
url	https://arxiv.org/abs/2505.20280