Title Information
Title
Learning the Geometry of Collider Events with Metric-Aware Deep Sets
Type of Resource (primo)
manuscripts
Type of Resource (primo)
manuscripts
Abstract
Optimal transport gives structured data a geometry, but exact evaluation is costly in large pairwise analyses that exploit relationships among distances. Learned surrogates are faster, but need not preserve this metric structure. We develop a Deep Sets surrogate for OT between variable-size weighted point clouds that enforces non-negativity, exchange symmetry, and zero self-distance, leaving the triangle inequality unconstrained. Applied to the Energy Mover’s Distance between collider events in a particle physics application, the Metric-Aware Particle Flow Network achieves percent-level mean absolute percentage error while significantly improving inference throughput over other exact and approximate methods surveyed. The architectural constraints are found to improve properties that are not explicitly enforced: across 10^6 held-out event triplets, triangle-inequality violations fall from 199 for a matched unconstrained network to 2, and the maximum from 149.5 to 5.8 GeV. These results demonstrate that targeted inductive biases can yield fast neural surrogates with substantially improved geometric fidelity.
Name
Name Part
Lauren Hay
Role
Role Term (marcrelator) (authorityURI="http://id.loc.gov/vocabulary/relators", valueURI="http://id.loc.gov/vocabulary/relators/aut")
Author
Name
Name Part
Rishabh Jain
Role
Role Term (marcrelator) (authorityURI="http://id.loc.gov/vocabulary/relators", valueURI="http://id.loc.gov/vocabulary/relators/aut")
Author
Name
Name Part
Matt LeBlanc
Role
Role Term (marcrelator) (authorityURI="http://id.loc.gov/vocabulary/relators", valueURI="http://id.loc.gov/vocabulary/relators/aut")
Author
Name
Name Part
Jennifer Roloff
Role
Role Term (marcrelator) (authorityURI="http://id.loc.gov/vocabulary/relators", valueURI="http://id.loc.gov/vocabulary/relators/aut")
Author
Origin Information
Date Created
2026-09-09
Subject (Local)
Topic
Optimal Transport
Subject (Local)
Topic
Large Hadron Collider
Subject (Local)
Topic
Particle Physics
Subject (Local)
Topic
AI/ML
Subject (Local)
Topic
metric spaces
Subject (Local)
Topic
deep sets
Subject (Local)
Topic
neural surrogates
Access Condition: use and reproduction (href="https://creativecommons.org/licenses/by/4.0/legalcode")
Attribution 4.0 International (CC BY 4.0)
Access Condition: logo (href="https://licensebuttons.net/l/by/4.0/88x31.png")
Access Condition: rights statement (href="http://rightsstatements.org/vocab/InC/1.0/")
In Copyright
Identifier: DOI
10.26300/1f5r-bs32