Maximizing Boosted Top Identification by Minimizing N-subjettiness (1108.2701v2)

Abstract: N-subjettiness is a jet shape designed to identify boosted hadronic objects such as top quarks. Given N subjet axes within a jet, N-subjettiness sums the angular distances of jet constituents to their nearest subjet axis. Here, we generalize and improve on N-subjettiness by minimizing over all possible subjet directions, using a new variant of the k-means clustering algorithm. On boosted top benchmark samples from the BOOST2010 workshop, we demonstrate that a simple cut on the 3-subjettiness to 2-subjettiness ratio yields 20% (50%) tagging efficiency for a 0.23% (4.1%) fake rate, making N-subjettiness a highly effective boosted top tagger. N-subjettiness can be modified by adjusting an angular weighting exponent, and we find that the jet broadening measure is preferred for boosted top searches. We also explore multivariate techniques, and show that additional improvements are possible using a modified Fisher discriminant. Finally, we briefly mention how our minimization procedure can be extended to the entire event, allowing the event shape N-jettiness to act as a fixed N cone jet algorithm.

• The paper demonstrates that minimizing N-subjettiness via iterative clustering significantly improves boosted top quark tagging efficiency while reducing misidentification rates.
• It employs discriminators like τ3/τ2, achieving tangible results such as 20% efficiency at 0.23% fake rate and 50% efficiency at 4.1% fake rate in benchmark analyses.
• The study highlights the potential to extend the optimized N-subjettiness approach to broader jet algorithms and exotic particle identification challenges.

A Study on Maximizing Boosted Top Identification through Minimization of N-subjettiness

This paper focuses on enhanced techniques for identifying boosted top quarks via optimization of the $N$-subjettiness observable, a useful jet shape designed to quantify the $N$-prong substructure of jets. By generalizing $N$-subjettiness and employing a clustering minimization technique akin to $k$-means, significant improvements are demonstrated in tagging performance for boosted tops in particle collider experiments.

Overview of N-subjettiness and Its Optimization

$N$-subjettiness ($\tau_N$) measures the alignment of radiation within a jet along $N$ subjet axes, which are determined using subject recognition algorithms. Here, the authors propose refining the identification process by selecting subjet axes such that $\tau_N$ is minimized rather than relying on fixed axes obtained from previous clustering methods, like the exclusive $k_T$ algorithm.

Their approach involves applying a clustering algorithm that minimizes $\tau_N$ through iterative refinement of subjet axes. They explore different angular weighting exponents, particularly focusing on $\beta = 1$ (the jet broadening measure), which exhibits superior discrimination capability for jet tagging.

Experimental Performance and Empirical Findings

The paper assesses the improved methodology's efficacy using benchmark samples from the BOOST2010 report, comparing enhanced $N$-subjettiness against established top tagging methods. They utilize two powerful discriminators: the ratios $\tau_3/\tau_2$ and $\tau_2/\tau_1$. A single-variable cut of $\tau_3/\tau_2$ demonstrates strong tagging efficiency (20% with a 0.23% fake rate and 50% with a 4.1% fake rate). This is notable as projects involving hadronic top quarks often grapple with minimizing misidentification rates while maximizing tagging efficiencies.

Beyond unitary cut procedures, the paper explores multivariate methods, highlighting augmented performance through a modified Fisher discriminant that optimally combines multiple substructure variables. These outcomes underscore the robustness and adaptability of the $N$-subjettiness framework in complex collider environments.

Implications and Future Prospects

The implications of this research are significant for theoretical and practical advancements in jet substructure analysis. By reinforcing boosted top tagging capabilities, the approach aids in deepening our understanding of beyond-the-Standard Model phenomena in high-energy physics. The analytical techniques presented potentially extend to other substructure problems, like identifying Higgs bosons and other exotic particles.

An intriguing theoretical prospect raised involves extending $N$-subjettiness minimization to entire events, transforming it into a novel fixed $N$ cone jet algorithm. This could bridge fixed-cone and compositional clustering methods, possibly refining next-generation jet algorithms used in particle colliders.

Furthermore, the paper enriches the narrative around the evolving landscape of machine learning and optimization techniques in particle physics, suggesting a broader applicability of clustering and minimization strategies with an eye toward precision particle identification tasks.

In conclusion, while behavior under different Monte Carlo settings requires further exploration, $N$-subjettiness as elucidated here stands as a compelling addition to the toolkit of methods used for jet identification at high-energy colliders. The research promises impactful future developments in streamlining complex event reconstructions and improving the precision of substructure analyses.