Infrared safe definition of jet flavour

TL;DR

The paper tackles the lack of infrared-safe jet flavour definitions by introducing the flavour-kt algorithm, a flavour-aware modification of the ${k_t}$ clustering distance that treats soft-quark emissions differently from gluon emissions. It extends the method to hadron-hadron collisions with dynamic beam scales and a bland variant to control multiply-flavoured clusters, and validates IR safety through NLO tests in ${e^+e^-}$ and extensive Monte Carlo studies. The approach enables consistent separation of quark and gluon jets in multi-jet environments, aiding fixed-order/ resummation matching and heavy-quark jet studies. This work provides a practical framework for reliable flavour tagging in QCD analyses at both $e^+e^-$ and hadron colliders and points to future integration of flavour information in NLO codes.

Abstract

It is common, in both theoretical and experimental studies, to separately discuss quark and gluon jets. However, even at parton level, widely-used jet algorithms fail to provide an infrared safe way of making this distinction. We examine the origin of the problem, and propose a solution in terms of a new "flavour-kt" algorithm. As well as being of conceptual interest this can be a powerful tool when combining fixed-order calculations with multi-jet resummations and parton showers. It also has applications to studies of heavy-quark jets.

Figures (7)

• Figure 1: (a) Specific $q\bar{q} \to q\bar{q}$ flavour channel for a $2\to2$ parton scattering process; (b) higher-order diagram that can be seen as a correction to (a); (c) higher-order diagram that can be seen as a correction to the process $q\bar{q} \to gg$, but with the same final-state partons as (b).
• Figure 2: A large-angle soft gluon splitting to a large-angle soft $q\bar{q}$ pair ($k_3$, $k_4$) with the $q$ and $\bar{q}$ then clustered into different jets ($k_1$, $k_2$).
• Figure 3: NLO differential cross section for $e^+e^- \to q\bar{q}$ events that after jet clustering have their flavour badly identified, i.e. identified as consisting of two gluon jets (that is, each of zero net flavour) or two jets each of net flavour larger than 1; the coefficient of $(\alpha_s/2\pi)^2$, as generated with Event2 CataniSeymour, is plotted as a function of the Durham $y_3$ three-jet resolution threshold; results are shown for the Durham and flavour algorithms (for two values of $\alpha$).
• Figure 4: Fraction of events (generated by Herwig Herwig at parton level) whose flavour is badly identified by various jet algorithms, shown as a function of the Durham $y_3^D$ jet resolution threshold; a large value of $Q$ has been chosen for illustrational purposes, so as to provide a correspondingly large range in $y_3^D$; the left-hand plot shows results for $e^+e^- \to q\bar{q}$, while the right-hand plot shows fake "$e^+e^- \to gg$" process as generated by Herwig (code=107).
• Figure 5: Plot of $k_{tB}$ and $k_{t\bar{B}}$ for a multi-jet parton-level LHC event, generated by Herwig; also shown is the histogram of the rapidity distribution of transverse momenta.