Towards an Understanding of the Correlations in Jet Substructure

157 citations (93 excluding self-citations). The fourth BOOST workshop report, shifting focus from benchmarking individual tools to understanding why different substructure observables are correlated and how to combine them optimally.

The Approach

By 2013, jet substructure had matured from a novel idea to a standard tool at the LHC. The question was no longer “which tagger works best” but “why do different taggers give similar results, and how can we combine observables to extract more information?” The report presents particle-level studies of the relationships between substructure observables, their complementarity for quark/gluon discrimination and boosted-object tagging, and their sensitivity to underlying jet properties like color charge and mass. This understanding of correlations laid groundwork for the machine-learning approaches that followed, where knowing which features carry independent information is essential for building effective classifiers.

Recollections

[To be filled in.]

• DOI
• arXiv:1504.00679
• INSPIRE