What the Primary Lund Plane Omits: A Model-Stitching Diagnosis of a Jet-Tagger Representation Gap
Abstract
Lund-plane and Particle Transformer (ParT) jet taggers can differ in performance because one representation is weaker, or because a weaker downstream computation reads an adequate one. Standalone performance cannot separate these. Model stitching can: it links the frozen early layers of one network to the frozen late layers of another through a single trained linear map. As a positive control we use an incomplete Lund representation, the primary declustering branch, whose deficit is known independently. Stitched into ParT's downstream, it reaches a background rejection at least as much as its own downstream recovers, yet far below ParT's background rejection: the deficit is representational, not computational. A hypergraph reader over splitting triplets does no better, ruling out higher-order correlations as the cause. The gap is invisible in AUC, which stays within 1\% of its target value, and appears only in background rejection at low signal efficiency. It localizes to 0.6\% of background jets, at and above the top mass, with more constituents but fewer primary splittings than typical background. Rebuilding the representation on the full Cambridge/Aachen declustering tree repairs the deficit as predicted: standalone rejection rises, closing 72\% of the gap. Masking nodes at inference shows the repaired network still relies almost entirely on the primary branch; the secondary splittings make the representation learnable rather than supplying information it reads. Along the primary branch, the widest-angle splitting, near the top and W opening angles, is worth nearly twice a random one. For quark/gluon tagging the primary branch is far less deficient, retaining 77.6\% of ParT's rejection (46.9\% for top), and the full tree closes only 33\% of the gap (72\% for top).
Related
- Semi-analytical representation of the Gaunt-modified Landau-Lifshitz equation in a plane wave
- Wide-field-of-view naked-eye virtual reality 3D light field display system based on multichannel image-space stitching
- ResoSeg: Resonance Tagger using Transformer and Segment Model
- Multi-image Overlap Stitching and Automatic Image Construction for Coherent X-ray Imaging
- A compact, interrogable Tree Tensor Network Jet Tagger on Lund declustering trees
BibTeX
@misc{rai2026what,
title = {{What the Primary Lund Plane Omits: A Model-Stitching Diagnosis of a Jet-Tagger Representation Gap}},
author = {Saurabh Rai and Sanmay Ganguly},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.37137},
eprint = {2610.37137},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.37137},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - What the Primary Lund Plane Omits: A Model-Stitching Diagnosis of a Jet-Tagger Representation Gap
AU - Saurabh Rai
AU - Sanmay Ganguly
PY - 2026
DA - 2026/10/27
PB - arXiv
DO - 10.48550/arxiv.2610.37137
UR - https://arxiv.org/abs/2610.37137
AB - Lund-plane and Particle Transformer (ParT) jet taggers can differ in performance because one representation is weaker, or because a weaker downstream computation reads an adequate one. Standalone performance cannot separate these. Model stitching can: it links the frozen early layers of one network to the frozen late layers of another through a single trained linear map. As a positive control we use an incomplete Lund representation, the primary declustering branch, whose deficit is known independently. Stitched into ParT's downstream, it reaches a background rejection at least as much as its own downstream recovers, yet far below ParT's background rejection: the deficit is representational, not computational. A hypergraph reader over splitting triplets does no better, ruling out higher-order correlations as the cause. The gap is invisible in AUC, which stays within 1\% of its target value, and appears only in background rejection at low signal efficiency. It localizes to 0.6\% of background jets, at and above the top mass, with more constituents but fewer primary splittings than typical background. Rebuilding the representation on the full Cambridge/Aachen declustering tree repairs the deficit as predicted: standalone rejection rises, closing 72\% of the gap. Masking nodes at inference shows the repaired network still relies almost entirely on the primary branch; the secondary splittings make the representation learnable rather than supplying information it reads. Along the primary branch, the widest-angle splitting, near the top and W opening angles, is worth nearly twice a random one. For quark/gluon tagging the primary branch is far less deficient, retaining 77.6\% of ParT's rejection (46.9\% for top), and the full tree closes only 33\% of the gap (72\% for top).
ER -