Towards an understanding of the correlations in jet substructure
 1k Downloads
 38 Citations
Abstract
Over the past decade, a large number of jet substructure observables have been proposed in the literature, and explored at the LHC experiments. Such observables attempt to utilize the internal structure of jets in order to distinguish those initiated by quarks, gluons, or by boosted heavy objects, such as top quarks and W bosons. This report, originating from and motivated by the BOOST2013 workshop, presents original particlelevel studies that aim to improve our understanding of the relationships between jet substructure observables, their complementarity, and their dependence on the underlying jet properties, particularly the jet radius and jet transverse momentum. This is explored in the context of quark/gluon discrimination, boosted W boson tagging and boosted top quark tagging.
1 Introduction
The centerofmass energies at the Large Hadron Collider are large compared to the heaviest of known particles, even after accounting for parton density functions. With the start of the second phase of operation in 2015, the centerofmass energy will further increase from 7 TeV in 2010–2011 and 8 TeV in 2012 to 13 TeV. Thus, even the heaviest states in the Standard Model (and potentially previously unknown particles) will often be produced at the LHC with substantial boosts, leading to a collimation of the decay products. For fully hadronic decays, these heavy particles will not be reconstructed as several jets in the detector, but rather as a single hadronic jet with distinctive internal substructure. This realization has led to a new era of sophistication in our understanding of both standard Quantum Chromodynamics (QCD) jets, as well as jets containing the decay of a heavy particle, with an array of new jet observables and detection techniques introduced and studied to distinguish the two types of jets. To allow the efficient propagation of results from these studies of jet substructure, a series of BOOST Workshops have been held on an annual basis: SLAC (2009) [1], Oxford University (2010) [2], Princeton University (2011) [3], IFIC Valencia (2012) [4], University of Arizona (2013) [5], and, most recently, University College London (2014) [6]. Following each of these meetings, working groups have generated reports highlighting the most interesting new results, and often including original particlelevel studies. Previous BOOST reports can be found at [7, 8, 9].
This report from BOOST 2013 thus views the study and implementation of jet substructure techniques as a fairly mature field, and focuses on the question of the correlations between the plethora of observables that have been developed and employed, and their dependence on the underlying jet parameters, especially the jet radius R and jet transverse momentum (\(p_{T} \)). In new analyses developed for the report, we investigate the separation of a quark signal from a gluon background (q / g tagging), a W signal from a gluon background (Wtagging) and a top signal from a mixed quark/gluon QCD background (toptagging). In the case of toptagging, we also investigate the performance of dedicated toptagging algorithms, the HepTopTagger [10] and the Johns Hopkins Tagger [11]. We study the degree to which the discriminatory information provided by the observables and taggers overlaps by examining the extent to which the signalbackground separation performance increases when two or more variables/taggers are combined in a multivariate analysis. Where possible, we provide a discussion of the physics behind the structure of the correlations and the \(p_{T} \) and R scaling that we observe.
We present the performance of observables in idealized simulations without pileup and detector resolution effects; the relationship between substructure observables, their correlations, and how these depend on the jet radius R and jet \(p_{T} \) should not be too sensitive to such effects. Conducting studies using idealized simulations allows us to more clearly elucidate the underlying physics behind the observed performance, and also provides benchmarks for the development of techniques to mitigate pileup and detector effects. A full study of the performance of pileup and detector mitigation strategies is beyond the scope of the current report, and will be the focus of upcoming studies.
The report is organized as follows: in Sects. 2–4, we describe the methods used in carrying out our analysis, with a description of the Monte Carlo event sample generation in Sect. 2, the jet algorithms, observables and taggers investigated in our report in Sect. 3, and an overview of the multivariate techniques used to combine multiple observables into single discriminants in Sect. 4. Our results follow in Sects. 5–7, with q / gtagging studies in Sect. 5, Wtagging studies in Sect. 6, and toptagging studies in Sect. 7. Finally we offer some summary of the studies and general conclusions in Sect. 8.
The principal organizers of and contributors to the analyses presented in this report are: B. Cooper, S. D. Ellis, M. Freytsis, A. Hornig, A. Larkoski, D. Lopez Mateos, B. Shuve, and N. V. Tran.
2 Monte Carlo samples
Below, we describe the Monte Carlo samples used in the q/g tagging, Wtagging, and toptagging sections of this report. Note that no pileup (additional proton–proton interactions beyond the hard scatter) are included in any samples, and there is no attempt to emulate the degradation in angular and \(p_{T} \) resolution that would result when reconstructing the jets inside a real detector; such effects are deferred to future study.
2.1 Quark/gluon and Wtagging
Samples were generated at \(\sqrt{s} = 8 \,\mathrm{TeV}~\) for QCD dijets, and for \(W^+W^\) pairs produced in the decay of a scalar resonance. The W bosons are decayed hadronically. The QCD events were split into subsamples of gg and \(q\bar{q}\) events, allowing for tests of discrimination of hadronic W bosons, quarks, and gluons.
Individual gg and \(q\bar{q}\) samples were produced at leading order (LO) using MadGraph5 [12], while \(W^+W^\) samples were generated using the JHU Generator [13, 14, 15]. Both were generated using CTEQ6L1 PDFs [16]. The samples were produced in exclusive \(p_{T} \) bins of width 100 GeV, with the slicing parameter chosen to be the \(p_{T} \) of any final state parton or W at LO. At the parton level, the \(p_{T} \) bins investigated in this report were 300–400 GeV, 500–600 GeV and 1.0–1.1 TeV. The samples were then showered through Pythia8 (version 8.176) [17] using the default tune 4C [18]. For each of the various samples (\(W,\,q,\,g\)) and \(p_{T} \) bins, 500 k events were simulated.
2.2 Toptagging
Samples were generated at \(\sqrt{s}=14\) TeV. Standard Model dijet and top pair samples were produced with Sherpa 2.0.0 [19, 20, 21, 22, 23, 24], with matrix elements of up to two extra partons matched to the shower. The top samples included only hadronic decays and were generated in exclusive \(p_{T} \) bins of width 100 GeV, taking as slicing parameter the top quark \(p_{T} \). The QCD samples were generated with a lower cut on the leading partonlevel jet \(p_{T} \), where partonlevel jets are clustered with the anti\(k_T\) algorithm and jet radii of \(R= 0.4,\,0.8,\,1.2\). The matching scale is selected to be \(Q_\mathrm{cut}=40,\,60,\,80\,\mathrm{GeV}\) for the \(p_{T\,\text {min}}=600, 1000\), and \(1500\,\mathrm{GeV}\) bins, respectively. For the top samples, 100k events were generated in each bin, while 200 k QCD events were generated in each bin.
3 Jet algorithms and substructure observables
In Sects. 3.1, 3.2, 3.3 and 3.4, we describe the various jet algorithms, groomers, taggers and other substructure variables used in these studies. Over the course of our study, we considered a larger set of observables, but for presentation purposes we included only a subset in the final analysis, eliminating redundant observables.
We organize the algorithms into four categories: clustering algorithms, grooming algorithms, tagging algorithms, and other substructure variables that incorporate information about the shape of radiation inside the jet. We note that this labelling is somewhat ambiguous: for example, some of the “grooming” algorithms (such as trimming and pruning) as well as Nsubjettiness can be used in a “tagging” capacity. This ambiguity is particularly pronounced in multivariate analyses, such as the ones we present here, since a single variable can act in different roles depending on which other variables it is combined with. Therefore, the following classification is intended only to give an approximate organization of the variables, rather than as a definitive taxonomy.
Before describing the observables used in our analysis, we give our definition of jet constituents. As a starting point, we can think of the final state of an LHC collision event as being described by a list of “final state particles”. In the analyses of the simulated events described below (with no detector simulation), these particles include the sufficiently long lived protons, neutrons, photons, pions, electrons and muons with no requirements on \(p_{\mathrm {T}}\) or rapidity. Neutrinos are excluded from the jet analyses.
3.1 Jet clustering algorithms
This process of jet clustering serves to identify jets as (nonoverlapping) sublists of final state particles within the original eventwide list. The particles on the sublist corresponding to a specific jet are labeled the “constituents” of that jet, and most of the tools described here process this sublist of jet constituents in some specific fashion to determine some property of that jet. The concept of constituents of a jet can be generalized to a more detectorcentric version where the constituents are, for example, tracks and calorimeter cells, or to a perturbative QCD version where the constituents are partons (quarks and gluons). These different descriptions are not identical, but are closely related. We will focus on the MC based analysis of simulated events, while drawing insight from the perturbative QCD view. Note also that, when a detector (with a magnetic field) is included in the analysis, there will generally be a minimum \(p_{\mathrm {T}}\) requirement on the constituents so that realistic numbers of constituents will be smaller than, but presumably still proportional to, the numbers found in the analyses described here.
3.2 Jet grooming algorithms
Filtering Given a jet, recluster the constituents into subjets of radius \(R_\mathrm{filt}\) with the C/A algorithm. Redefine the jet to consist of only the hardest N subjets, where N is determined by the final state topology and is typically one more than the number of hard prongs in the resonance decay (to include the leading finalstate gluon emission) [35]. While we do not independently use filtering, it is an important step of the HEPTopTagger to be defined later.
3.3 Jet tagging algorithms
Johns Hopkins Tagger Recluster the jet using the C/A algorithm. The jet is iteratively declustered, and at each step the softer prong is discarded if its \(p_\mathrm{T}\) is less than \(\delta _p\,p_{\mathrm {T\,jet}}\). This continues until both prongs are harder than the \(p_\mathrm{T}\) threshold, both prongs are softer than the \(p_\mathrm{T}\) threshold, or if they are too close (\(\Delta \eta _{ij}+\Delta \phi _{ij}<\delta _R\)); the jet is rejected if either of the latter conditions apply. If both are harder than the \(p_\mathrm{T}\) threshold, the same procedure is applied to each: this results in 2, 3, or 4 subjets. If there exist 3 or 4 subjets, then the jet is accepted: the top candidate is the sum of the subjets, and W candidate is the pair of subjets closest to the W mass [11]. The output of the tagger is the mass of the top candidate (\(m_t\)), the mass of the W candidate (\(m_W\)), and \(\theta _\mathrm{h}\), a helicity angle defined as the angle, measured in the rest frame of the W candidate, between the top direction and one of the W decay products. The two free input parameters of the John Hopkins tagger in this study are \(\delta _p\) and \(\delta _R\), defined above, and their values are optimized for different jet kinematics and parameters in Sect. 7.
HEPTopTagger Recluster the jet using the C/A algorithm. The jet is iteratively declustered, and at each step the softer prong is discarded if \(m_1/m_{12}>\mu \) (there is not a significant mass drop). Otherwise, both prongs are kept. This continues until a prong has a mass \(m_i < m\), at which point it is added to the list of subjets. Filter the jet using \(R_\mathrm{filt}=\mathrm {min}(0.3,\Delta R_{ij})\), keeping the five hardest subjets (where \(\Delta R_{ij}\) is the distance between the two hardest subjets). Select the three subjets whose invariant mass is closest to \(m_t\) [10]. The top candidate is rejected if there are fewer than three subjets or if the top candidate mass exceeds 500 GeV. The output of the tagger is \(m_t\), \(m_W\), and \(\theta _\mathrm{h}\) (as defined in the Johns Hopkins Tagger). The two free input parameters of the HEPTopTagger in this study are m and \(\mu \), defined above, and their values are optimized for different jet kinematics and parameters in Sect. 7.
Toptagging with pruning or trimming In the studies presented in Sect. 7 we add a W reconstruction step to the pruning and trimming algorithms, to enable a fairer comparison with the dedicated top tagging algorithms described above. Following the method of the BOOST 2011 report [8], a W candidate is found as follows: if there are two subjets, the highestmass subjet is the W candidate (because the W prongs end up clustered in the same subjet), and the W candidate mass, \(m_W\), the mass of this subjet; if there are three subjets, the two subjets with the smallest invariant mass comprise the W candidate, and \(m_W\) is the invariant mass of this subjet pair. In the case of only one subjet, the top candidate is rejected. The top mass, \(m_t\), is the full mass of the groomed jet.
3.4 Other jet substructure observables
The jet substructure observables defined in this section are calculated using jet constituents prior to any grooming. This approach has been used in several analyses in the past, for example [38, 39], whilst others have used the approach of only considering the jet constituents that survive the grooming procedure [40]. We take the first approach throughout our analyses, as this approach allows a study of both the hard and soft radiation characteristic of signal vs. background. However, we do include the effects of initial state radiation and the underlying event, and unsurprisingly these can have a nonnegligible effect on variable performance, particularly at large \(p_{T} \) and jet R. This suggests that the differences we see between variable performance at large \(p_{T}/R\) will be accentuated in a high pileup environment, necessitating a dedicated study of pileup to recover as much as possible the “ideal” performance seen here. Such a study is beyond the scope of this paper.
4 Multivariate analysis techniques
Multivariate techniques are used to combine multiple variables into a single discriminant in an optimal manner. The extent to which the discrimination power increases in a multivariable combination indicates to what extent the discriminatory information in the variables overlaps. There exist alternative strategies for studying correlations in discrimination power, such as “truth matching” [46], but these are not explored here.
In all cases, the multivariate technique used to combine variables is a Boosted Decision Tree (BDT) as implemented in the TMVA package [47]. An example of the BDT settings used in these studies, chosen to reduce the effect of overtraining, is given in [47]. The BDT implementation including gradient boost is used. Additionally, the simulated data were split into training and testing samples and comparisons of the BDT output were compared to ensure that the BDT performance was not affected by overtraining.
5 Quark–gluon discrimination
In this section, we examine the differences between quark and gluoninitiated jets in terms of substructure variables. At a fundamental level, the primary difference between quark and gluoninitiated jets is the color charge of the initiating parton, typically expressed in terms of the ratio of the corresponding Casimir factors \(C_F/C_A = 4/9\). Since the quark has the smaller color charge, it radiates less than a corresponding gluon and the naive expectation is that the resulting quark jet will contain fewer constituents than the corresponding gluon jet. The differing color structure of the two types of jet will also be realized in the detailed behavior of their radiation patterns. We determine the extent to which the substructure observables capturing these differences are correlated, providing some theoretical understanding of these variables and their performance. The motivation for these studies arises not only from the desire to “tag” a jet as originating from a quark or gluon, but also to improve our understanding of the quark and gluon components of the QCD backgrounds relative to boosted resonances. While recent studies have suggested that quark/gluon tagging efficiencies depend highly on the Monte Carlo generator used [48, 49], we are more interested in understanding the scaling performance with \(p_{T} \) and R, and the correlations between observables, which are expected to be treated consistently within a single shower scheme.
Other examples of recent analytic studies of the correlations between jet observables relevant to quark jet versus gluon jet discrimination can be found in [43, 46, 50, 51].
5.1 Methodology and observable classes

Number of constituents (\(n_\mathrm{constits}\)) in the jet.

Pruned Qjet mass volatility, \(\Gamma _\mathrm{Qjet}\).

1point energy correlation functions, \(C_1^{\beta }\) with \(\beta =0,\,1,\,2\).

1subjettiness, \(\tau _1^{\beta }\) with \(\beta =1,\,2\). The Nsubjettiness axes are computed using onepass \(k_t\) axis optimization.

Ungroomed jet mass, m.
We will demonstrate that, in terms of their jetbyjet correlations and their ability to separate quark jets from gluon jets, the above observables fall into five Classes. The first three observables, \(n_\mathrm{constits}\), \(\Gamma _\mathrm{Qjet}\) and \(C_1^{\beta =0}\), each constitutes a Class of its own (Classes I–III) in the sense that they each carry some independent information about a jet and, when combined, provide substantially better quark jet and gluon jet separation than any one observable alone. Of the remaining observables, \(C_1^{\beta =1}\) and \(\tau _1^{\beta =1}\) comprise a single class (Class IV) because their distributions are similar for a sample of jets, their jetbyjet values are highly correlated, and they exhibit very similar power to separate quark jets and gluon jets (with very similar dependence on the jet parameters R and \(p_T\)); this separation power is not improved when they are combined. The fifth class (Class V) is composed of \(C_1^{\beta =2}\), \(\tau _1^{\beta =2}\) and the (ungroomed) jet mass. Again the jetbyjet correlations are strong (even though the individual observable distributions are somewhat different), the quark versus gluon separation power is very similar (including the R and \(p_T\) dependence), and little is achieved by combining more than one of the Class V observables. This class structure is not surprising given that the observables within a class exhibit very similar dependence on the kinematics of the underlying jet constituents. For example, the members of Class V are constructed from of a sum over pairs of constituents using products of the energy of each member of the pair times the angular separation squared for the pair (this is apparent for the ungroomed mass when viewed in terms of a masssquared with small angular separations). By the same argument, the Class IV and Class V observables will be seen to be more similar than any other pair of classes, differing only in the power (\(\beta \)) of the dependence on the angular separations, which produces small but detectable differences. We will return to a more complete discussion of jet masses in Sect. 5.4.
5.2 Single variable discrimination
To more quantitatively study the power of each observable as a discriminator for quark/gluon tagging, Receiver Operating Characteristic (ROC) curves are built by scanning each distribution and plotting the background efficiency (to select gluon jets) vs. the signal efficiency (to select quark jets). Figure 2 shows these ROC curves for all of the substructure variables shown in Fig. 1 for \(R=0.4, 0.8\) and 1.2 jets (in the \(p_{T} =300\)–\(400\,\mathrm{GeV}\) bin). In addition, the ROC curve for a tagger built from a BDT combination of all the variables (see Sect. 4) is shown.
As suggested earlier, \(n_\mathrm{constits}\) is the best performing variable for all R values, although \(C_1^{\beta =0}\) is not far behind, particularly for \(R=0.8\). Most other variables have similar performance, with the main exception of \(\Gamma _\mathrm{Qjet}\), which shows significantly worse discrimination (this may be due to our choice of rigidity \(\alpha = 0.1\), with other studies suggesting that a smaller value, such as \(\alpha = 0.01\), produces better results [31, 32]). The combination of all variables shows somewhat better discrimination than any individual observable, and we give a more detailed discussion in Sect. 5.3 of the correlations between the observables and their impact on the combined discrimination power.
We now examine how the performance of the substructure observables varies with \(p_{T} \) and R. To present the results in a “digestible” fashion we focus on the gluon jet “rejection” factor, \(1/\varepsilon _\text {bkg}\), for a quark signal efficiency, \(\varepsilon _\text {sig}\), of \(50\,\%\). We can use the values of \(1/\varepsilon _\text {bkg}\) generated for the 9 kinematic points introduced above (\(R = 0.4, 0.8, 1.2\) and the 100 GeV \(p_{T} \) bins with lower limits \(p_T = 300\), 500, \(1000\,\text {GeV}\)) to generate surface plots. The surface plots in Fig. 3 indicate both the level of gluon rejection and the variation with \(p_{T} \) and R for each of the studied single observable. The color shading in these plots is defined so that a value of \(1/\varepsilon _\text {bkg}\simeq 1\) yields the color “violet”, while \(1/\varepsilon _\text {bkg}\simeq 20 \) yields the color “red”. The “rainbow” of colors in between vary linearly with \(\log _{10} (1/\varepsilon _\text {bkg})\).
We organize our results by the classes introduced in the previous subsection:
Class I The sole constituent of this class is \(n_\mathrm{constits}\). We see in Fig. 3a that, as expected, the numerically largest rejection rates occur for this observable, with the rejection factor ranging from 6 to 11 and varying rather dramatically with R. As R increases the jet collects more constituents from the underlying event, which are the same for quark and gluon jets, and the separation power decreases. At large R, there is some improvement with increasing \(p_{T} \) due to the enhanced QCD radiation, which is different for quarks vs. gluons.
Class II The variable \(\Gamma _\mathrm{Qjet}\) constitutes this class. Figure 3b confirms the limited efficacy of this single observable (at least for our parameter choices) with a rejection rate only in the range 2.5–2.8. On the other hand, this observable probes a very different property of jet substructure, i.e., the sensitivity to detailed changes in the grooming procedure, and this difference is suggested by the distinct R and \(p_{T} \) dependence illustrated in Fig. 3b. The rejection rate increases with increasing R and decreasing \(p_{T} \), since the distinction between quark and gluon jets for this observable arises from the relative importance of the one “hard” gluon emission configuration. The role of this contribution is enhanced for both decreasing \(p_{T} \) and increasing R. This general variation with \(p_{\mathrm {T}} \) and R is the opposite of what is exhibited in all of the other single variable plots in Fig. 3.
Class III The only member of this class is \(C_1^{\beta =0}\). Figure 3c indicates that this observable can itself provide a rejection rate in the range 7.8–8.6 (intermediate between the two previous observables), and again with distinct R and \(p_{T} \) dependence. In this case the rejection rate decreases slowly with increasing R, which follows from the fact that \(\beta = 0\) implies no weighting of \(\Delta R\) in the definition of \(C_1^{\beta =0}\), greatly reducing the angular dependence. The rejection rate peaks at intermediate \(p_{T} \) values, an effect visually enhanced by the limited number of \(p_{T} \) values included.
Class IV Figure 3d, e confirm the very similar properties of the observables \(C_1^{\beta =1}\) and \(\tau _1^{\beta =1}\) (as already suggested in Fig. 1d, e). They have essentially identical rejection rates (4.1–5.4) and identical R and \(p_{T} \) dependence (a slow decrease with increasing R and an even slower increase with increasing \(p_{T} \)).
Class V The observables \(C_1^{\beta =2}\), \(\tau _1^{\beta =2}\), and m have similar rejection rates in the range 3.5 to 5.3, as well as very similar R and \(p_{T} \) dependence (a slow decrease with increasing R and an even slower increase with increasing \(p_{T} \)).
Arguably, drawing a distinction between the Class IV and Class V observables is a fine point, but the color shading does suggest some distinction from the slightly smaller rejection rate in Class V. Again the strong similarities between the plots within the second and third rows in Fig. 3 speaks to the common properties of the observables within the two classes.
In summary, the overall discriminating power between quark and gluon jets tends to decrease with increasing R, except for the \(\Gamma _\mathrm{Qjet}\) observable, presumably in large part due to the contamination from the underlying event. Since the construction of the \(\Gamma _\mathrm{Qjet}\) observable explicitly involves pruning away the soft, large angle constituents, it is not surprising that it exhibits different R dependence. In general the discriminating power increases slowly and monotonically with \(p_{T} \) (except for the \(\Gamma _\mathrm{Qjet}\) and \(C_1^{\beta =0}\) observables). This is presumably due to the overall increase in radiation from high \(p_{T} \) objects, which accentuates the differences in the quark and gluon color charges and providing some increase in discrimination. In the following section, we study the effect of combining multiple observables.
5.3 Combined performance and correlations
Combining multiple observables in a BDT can give further improvement over cuts on a single variable. Since the improvement from combining correlated observables is expected to be inferior to that from combining uncorrelated observables, studying the performance of multivariable combinations gives insight into the correlations between substructure variables and the physical features allowing for quark/gluon discrimination. Based on our discussion of the correlated properties of observables within a single class, we expect little improvement in the rejection rate when combining observables from the same class, and substantial improvement when combining observables from different classes. Our classification of observables for quark/gluon tagging therefore motivates the study of particular combinations of variables for use in experimental analyses.
To quantitatively study the improvement obtained from multivariate analyses, we build quark/gluon taggers from every pairwise combination of variables studied in the previous section; we also compare the pairwise performance with the allvariables combination. To illustrate the results achieved in this way, we use the same 2D surface plots as in Fig. 3. Figure 4 shows pairwise plots for variables in (a) Class IV and (b) Class V, respectively. Comparing to the corresponding plots in Fig. 3, we see that combining \(C_1^{\beta =1}+\tau _{1}^{\beta =1}\) provides a small (\(\sim \)10 %) improvement in the rejection rate with essentially no change in the R and \(p_{T} \) dependence, while combining \(C_1^{\beta =2}+\tau _{1}^{\beta =2}\) yields a rejection rate that is essentially identical to the single observable rejection rate for all R and \(p_{T} \) values (with a similar conclusion if one of these observables is replaced with the ungroomed jet mass m). This confirms the expectation that the observables within a single class effectively probe the same jet properties.
The R and \(p_{T} \) dependence of the pairwise combinations is generally similar to the single observable with the most dependence on R and \(p_{T} \). The smallest R and \(p_{T} \) variation always occurs when pairing with \(C_1^{\beta =0}\). Changing any of the observables in these pairs with a different observable in the same class (e.g., \(C_1^{\beta =2}\) for \(\tau _1^{\beta =2}\)) produces very similar results.
Figure 5l shows the performance of a BDT combination of all the current observables, with rejection rates in the range 10.5–17.1. The performance is very similar to that observed for the pairwise \(n_\mathrm{constits}+ C_1^{\beta =1}\) and \(n_\mathrm{constits}+ \tau _1^{\beta =1}\) combinations, but with a somewhat narrower range and slightly larger maximum values. This suggests that almost all of the available information to discriminate quark and gluoninitiated jets is captured by \(n_\mathrm{constits}\) and \(C_1^{\beta =1}\) or \(\tau _1^{\beta =1}\) variables; this confirms the finding that nearoptimal performance can be obtained with a pair of variables from [52].
Some features are more easily seen with an alternative presentation of the data. In Figs. 6 and 7 we fix R and \(p_{T} \) and simultaneously show the single and pairwise observables performance in a single matrix. The numbers in each cell are the same rejection rate for gluons used earlier, \(1/\varepsilon _\text {bkg}\), with \(\varepsilon _\text {sig}= 50\,\% \) (quarks). Figure 6 shows the results for \(p_{T} =1{}1.1\) TeV and \(R =0.4,0.8,1.2\), while Fig. 7 is for \(R = 0.4\) and the 3 \(p_{T} \) bins. The single observable rejection rates appear on the diagonal, and the pairwise results are off the diagonal. The largest pairwise rejection rate, as already suggested by Fig. 5e, appears at large \(p_{T} \) and small R for the pair \(n_ \mathrm{constits}+ \tau _1^{\beta =1}\) (with very similar results for \(n_ \mathrm{constits}+ C_1^{\beta =1}\)). The correlations indicated by the shading^{1} should be largely understood as indicating the organization of the observables into the nowfamiliar classes. The allobservable (BDT) result appears as the number at the lower right in each plot.
5.4 QCD jet masses
To close the discussion of q / gtagging, we provide some insight into the behavior of the masses of QCD jets initiated by both kinds of partons, with and without grooming. Recall that, in practice, an identified jet is simply a list of constituents, i.e., final state particles. To the extent that the masses of these individual constituents can be neglected (due to the constituents being relativistic), each constituent has a “well defined” 4momentum from its energy and direction. It follows that the 4momentum of the jet is simply the sum of the 4momenta of the constituents and its square is the jet mass squared. Simply on dimensional grounds, we know that jet mass must have an overall linear scaling with \(p_{T} \), with the remaining \(p_{T} \) dependence arising predominantly from the running of the coupling, \(\alpha _s(p_{T})\). The R dependence is also crudely linear as the jet mass scales approximately with the largest angular opening between any 2 constituents, which is set by R.
Several features of Fig. 8 can be easily understood. The distributions all cut off rapidly for \(m/p_T/R > 0.5\), which is understood as the precise limit (maximum mass) for a jet composed of just two constituents. As expected from the soft and collinear singularities in QCD, the mass distribution peaks at small mass values. The actual peak is “pushed” away from the origin by the socalled Sudakov form factor. Summing the corresponding logarithmic structure (singular in both \(p_{T} \) and angle) to all orders in perturbation theory yields a distribution that is highly damped as the mass vanishes. In words, there is precisely zero probability that a color parton emits no radiation (and the resulting jet has zero mass). Above the Sudakovsuppressed part of phase space, there are two structures in the distribution: the “shoulder” and the “peak”. The large mass shoulder (\(0.3 < m/p_T/R < 0.5\)) is driven largely by the presence of a single large angle, energetic emission in the underlying QCD shower, i.e., this regime is quite well described by loworder perturbation theory^{2} In contrast, we can think of the peak region as corresponding to multiple soft emissions. This simple, necessarily approximate picture provides an understanding of the bulk of the differences between the quark and gluon jet mass distributions. Since the probability of the single large angle, energetic emission is proportional to the color charge, the gluon distribution should be enhanced in this region by a factor of about \(C_A/C_F = 9/4\), consistent with what is observed in Fig. 8. Similarly the exponent in the Sudakov damping factor for the gluon jet mass distribution is enhanced by the same factor, leading to a peak “pushed” further from the origin. Therefore, compared to a quark jet, the gluon jet mass distribution exhibits a larger average jet mass, with a larger relative contribution arising from the perturbative shoulder region and a small mass peak that is further from the origin.
Together with the fact that the number of constituents in the jet is also larger (on average) for the gluon jet simply because a gluon will radiate more than a quark, these features explain much of what we observed earlier in terms of the effectiveness of the various observables to separate quark jets from gluons jets. They also give us insight into the difference in the distributions for the observable \(\Gamma _\mathrm{Qjet}\). Since the shoulder is dominated by a single large angle, hard emission, it is minimally impacted by pruning, which is designed to remove the large angle, soft constituents (as shown in more detail below). Thus, jets in the shoulder exhibit small volatility and they are a larger component in the gluon jet distribution. Hence gluon jets, on average, have smaller values of \(\Gamma _\mathrm{Qjet}\) than quark jets as in Fig. 1b. Further, this feature of gluon jets is distinct from the fact that there are more constituents, explaining why \(\Gamma _\mathrm{Qjet}\) and \(n_ \mathrm{constits}\) supply largely independent information for distinguishing quark and gluon jets.
Our final topic in this section is the residual R and \(p_{T} \) dependence exhibited in Figs. 8 and 9, which indicates a deviation from the naive linear scaling that has been removed by using the scaled variable \(m/p_T/R\). A helpful, intuitively simple, if admittedly imprecise, model of a jet is to separate the constituents of the jet into “hard” (with \(p_{T} \)’s that are of order the jet \(p_{T} \)) versus “soft” (with \(p_{T} \)’s small and fixed compared to the jet \(p_{T} \)), and “large” angle (with an angular separation from the jet direction of order R) versus “small” angle (with an angular separation from the jet direction smaller than and not scaling with R) components. As described above the Sudakov damping factor excludes constituents that are very soft or very small angle (or both). In this simple picture perturbative large angle, hard constituents appear rarely, but, as described above, they characterize the large mass jets that appear in the “shoulder” of the jet mass distribution where the mass scales approximately linearly with the jet \(p_{T} \) and with R. The hard, small angle constituents are somewhat more numerous and contribute to a jet mass that does not scale with R. The soft constituents are much more numerous (becoming more numerous with increasing jet \(p_{T} \)) and contribute to a jet mass that scales like \(\sqrt{p_{T,\text {jet}}}\). The small angle, soft constituents contribute to a jet mass that does not scale with R, while the large angle, soft constituents do contribute to a jet mass that scales like R and grow in number approximately linearly in R (i.e., with the area of the annulus at the outer edge of the jet). This simple picture allows at least a qualitative explanation of the behavior observed in Figs. 8 and 9.
As already suggested, the residual \(p_{T} \) dependence can be understood as arising primarily from the slow decrease of the strong coupling \(\alpha _s(p_{T})\) as \(p_{T} \) increases. This leads to a corresponding decrease in the (largely perturbative) shoulder regime for both distributions at higher \(p_{T} \), i.e., a decrease in the number of hard, large angle constituents. At the same time, and for the same reason, the Sudakov damping is less strong with increasing \(p_{T} \) and the peak moves in towards the origin. While the number of soft constituents increases with increasing jet \(p_{T} \), their contributions to the scaled jet mass distribution shift to smaller values of \(m/p_{T} \) (decreasing approximately like \(1/\sqrt{p_{T}}\)). Thus the overall impact of increasing \(p_{T} \) for both distributions is a (gradual) shift to smaller values of \(m/p_T/R\). This is just what is observed in Figs. 8 and 9, although the numerical size of the effect is reduced in the pruned case.
The residual R dependence is somewhat more complicated. The perturbative large angle, hard constituent contribution largely scales in the variable \(m/p_T/R\), which is why we see little residual R dependence in either figure at higher masses (\(m/p_T/R > 0.4\)). The contribution of the small angle constituents (hard and soft) contribute at fixed m and thus shift to the left versus the scaled variable as R increases. This presumably explains the small shifts in this direction at small mass observed in both figures. The large angle, soft constituents contribute to mass values that scale like R, and, as noted above, tend to increase in number as R increases (i.e., as the area of the jet grows). Such contributions yield a scaled jet mass distribution that shifts to the right with increasing R and presumably explain the behavior at small \(p_{T} \) in Fig. 8. Since pruning largely removes this contribution, we observe no such behavior in Fig. 9.
5.5 Conclusions
In Sect. 5 we have seen that a variety of jet observables provide information about the jet that can be employed to effectively separate quarkinitiated from gluoninitiated jets. Further, when used in combination, these observables can provide superior separation. Since the improvement depends on the correlation between observables, we use the multivariable performance to separate the observables into different classes, with each class containing highly correlated observables. We saw that the best performing single observable is simply the number of constituents in the jet, \(n_ \mathrm{constits}\), while the largest further improvement comes from combining with \(C_1^{\beta =1}\) (or \(\tau _1^{\beta =1}\)). The performance of this combined tagger is strongly dependent on \(p_{T} \) and R, with the best performance being observed for smaller R and higher \(p_{T} \). The smallest R and \(p_{T} \) dependence arises from combining \(n_ \mathrm{constits}\) with \(C_1^{\beta = 0}\). Some of the commonly used observables for q / g tagging are highly correlated and do not provide extra information when used together. We have found that adding further variables to the \(n_ \mathrm{constits}\) + \(C_1^{\beta =1}\) or \(n_ \mathrm{constits}\) + \(\tau _1^{\beta =1}\) BDT combination results in only a small improvement in performance, suggesting that almost all of the available information to discriminate quark and gluoninitiated jets is captured by \(n_\mathrm{constits}\) and \(C_1^{\beta =1}\) (or \(\tau _1^{\beta =1}\)) variables. In addition to demonstrating these correlations, we have provided a discussion of the physics behind the structure of the correlation. Using the jet mass as an example, we have given arguments to explicitly explain the differences between jet observables initiated by each type of parton.
Finally, we remind the reader that the numerical results were derived for a particular color configuration (qq and gg events), in a particular implementation of the parton shower and hadronization. Color connections in more complex event configurations, or different Monte Carlo programs, may well exhibit somewhat different efficiencies and rejection factors. The value of our results is that they indicate a subset of variables expected to be rich in information about the partonic origin of finalstate jets. These variables can be expected to act as valuable discriminants in searches for new physics, and could also be used to define modelindependent finalstate measurements which would nevertheless be sensitive to the shortdistance physics of quark and gluon production.
6 Boosted Wtagging
In this section, we study the discrimination of a boosted, hadronically decaying W boson (signal) against a gluoninitiated jet background, comparing the performance of various groomed jet masses and substructure variables. A range of different distance parameters for the anti\(k_T\) jet algorithm are explored, in a range of different leading jet \(p_{T} \) bins. This allows us to determine the performance of observables as a function of jet radius and jet boost, and to see where different approaches may break down. The groomed mass and substructure variables are then combined in a BDT as described in Sect. 4, and the performance of the resulting BDT discriminant explored through ROC curves to understand the degree to which variables are correlated, and how this changes with jet boost and jet radius. Using BDT combinations of substructure variables to improve W tagging has been studied earlier in [61].
6.1 Methodology
These studies use the WW samples as signal and the dijet gg as background, described previously in Sect. 2. Whilst only gluonic backgrounds are explored here, the conclusions regarding the dependence of the performance and correlations on the jet boost and radius are not expected to be substantially different for quark backgrounds; we will see that the differences in the substructure properties of quark and gluoninitiated jets, explored in the last section, are significantly smaller than the differences between Winitiated and gluoninitiated jets.

Ungroomed, trimmed (\(m_{\text {trim}}\)), and pruned (\(m_{\text {prun}}\)) jet masses.

Mass output from the modified mass drop tagger (\(m_{\text {mmdt}}\)).

Soft drop mass with \(\beta =2\) (\(m_{\mathrm {sd}}\)).

2point energy correlation function ratio \(C_2^{\beta =1}\) (we also studied \(\beta =2\) but do not show its results because it showed poor discrimination power).

Nsubjettiness ratio \(\tau _2 / \tau _1\) with \(\beta =1\) (\(\tau _{21}^{\beta =1}\)) and with axes computed using onepass \(k_t\) axis optimization (we also studied \(\beta =2\) but did not show its results because it showed poor discrimination power).

Pruned Qjet mass volatility, \(\Gamma _\mathrm{Qjet}\).
6.2 Single variable performance
In this section we explore the performance of the various groomed jet mass and substructure variables in separating signal from background. Since we have not attempted to optimise the grooming parameter settings of each grooming algorithm, we do not place much emphasis here on the relative performance of the groomed masses, but instead concentrate on how their performance changes depending on the kinematic bin and jet radius considered.
Figure 10 compares the signal and background in terms of the different groomed masses explored for the anti\(k_T\) \(R=0.8\) algorithm in the \(p_{T} \) = 500–600 GeV bin. One can clearly see that, in terms of separating signal and background, the groomed masses are significantly more performant than the ungroomed anti\(k_T\) \(R=0.8\) mass. Using the same jet radius and \(p_{T} \) bin, Fig. 11 compares signal and background for the different substructure variables studied.
Figures 12, 13 and 14 show the single variable ROC curves for various \(p_{T} \) bins and values of R. The single variable performance is also compared to the ROC curve for a BDT combination of all the variables (labelled “allvars”). In all cases, the “allvars” option is significantly more performant than any of the individual single variables considered, indicating that there is considerable complementarity between the variables, and this is explored further in Sect. 6.3.
In Figs. 15, 16 and 17 the same information is shown in a format that more readily allows for a quantitative comparison of performance for different R and \(p_{T} \); matrices are presented which give the background rejection for a signal efficiency of 70 %^{3} for single variable cuts, as well as two and threevariable BDT combinations. The results are shown separately for each \(p_{T} \) bin and jet radius considered. Most relevant for our immediate discussion, the diagonal entries of these plots show the background rejections for a single variable BDT using the labelled observable, and can thus be examined to get a quantitative measure of the individual single variable performance, and to study how this changes with jet radius and momenta. The offdiagonal entries give the performance when two variables (shown on the xaxis and on the yaxis, respectively) are combined in a BDT. The final column of these plots shows the background rejection performance for threevariable BDT combinations of \(m_{sd}^{\beta =2} + C_2^{\beta =1} + X\). These results will be discussed later in Sect. 6.3.3.
In general, the most performant single variables are the groomed masses. However, in certain kinematic bins and for certain jet radii, \(C_2^{\beta =1}\) has a background rejection that is comparable to or better than the groomed masses.
6.3 Combined performance
Studying the improvement in performance (or lack thereof) when combining single variables into a multivariate analysis gives insight into the correlations among jet observables. The offdiagonal entries in Figs. 15, 16 and 17 can be used to compare the performance of different BDT twovariable combinations, and see how this varies as a function of \(p_{T} \) and R. By comparing the background rejection achieved for the twovariable combinations to the background rejection of the “all variables” BDT, one can also understand how discrimination can be improved by adding further variables to the twovariable BDTs.
In general the most powerful twovariable combinations involve a groomed mass and a nonmass substructure variable (\(C_2^{\beta =1}\), \(\Gamma _\mathrm{Qjet}\) or \(\tau _{21}^{\beta =1}\)). Twovariable combinations of the substructure variables are not as powerful in comparison. Which particular mass \(+\) substructure variable combination is the most powerful depends strongly on the \(p_{T} \) and R of the jet, as discussed in the sections to follow.
There is also modest improvement in the background rejection when different groomed masses are combined, indicating that there is complementary information between the different groomed masses (first shown in [62]). In addition, there is an improvement in the background rejection when the groomed masses are combined with the ungroomed mass, indicating that grooming removes some useful discriminatory information from the jet. These observations are explored further in the section below.
Generally, the \(R=0.8\) jets offer the best twovariable combined performance in all \(p_{T} \) bins explored here. This is despite the fact that in the highest \(p_{T} \) = 1.0–1.1 TeV bin the average separation of the quarks from the W decay is much smaller than 0.8, and well within 0.4. This conclusion could of course be susceptible to pileup, which is not considered in this study. It is in marked contrast to the R dependence of the q / g tagging performance shown in Sect. 5, where a monotonic improvement in performance with reducing R is observed.
6.3.1 Mass + substructure performance
As already noted, the largest background rejection at 70 % signal efficiency are in general achieved using those twovariable BDT combinations which involve a groomed mass and a nonmass substructure variable. We now investigate the \(p_{T} \) and R dependence of the performance of these combinations.
For both \(R=0.8\) and \(R=1.2\) jets, the rejection power of these twovariable combinations increases substantially with increasing \(p_{T} \), at least within the \(p_{T} \) range considered here.
However, when we switch to a jet radius of \(R=1.2\) the picture for \(C_2^{\beta =1}\) combinations changes dramatically. These become significantly less powerful, and the most powerful variable in groomed mass combinations becomes \(\tau _{21}^{\beta =1}\) for all jet \(p_{T} \) considered. Figure 23 shows the correlation between \(m_{sd}^{\beta =2}\) and \(C_2^{\beta =1}\) in the \(p_{T} \) = 1.0–1.1 TeV bin for the various jet radii considered. Figure 24 is the equivalent set of distributions for \(m_{sd}^{\beta =2}\) and \(\tau _{21}^{\beta =1}\). One can see from Fig. 23 that, due to the sensitivity of the observable to soft, wideangle radiation, as the jet radius increases \(C_2^{\beta =1}\) increases and becomes more and more smeared out for both signal and background, leading to worse discrimination power. This does not happen to the same extent for \(\tau _{21}^{\beta =1}\). We can see from Fig. 24 that the negative correlation between \(m_{sd}^{\beta =2}\) and \(\tau _{21}^{\beta =1}\) that is clearly visible for \(R=0.4\) decreases for larger jet radius, such that the groomed mass and substructure variable are far less correlated and \(\tau _{21}^{\beta =1}\) offers improved discrimination within a \(m_{sd}^{\beta =2}\) mass window.
6.3.2 Mass + mass performance
6.3.3 “All variables” performance
Figures 15, 16 and 17 report the background rejection achieved by a combination of all the variables considered into a single BDT discriminant. In all cases, the rejection power of this “all variables” BDT is significantly larger than the best twovariable combination. This indicates that, beyond the best twovariable combination, there is still significant complementary information available in the remaining observables to improve the discrimination of signal and background. How much complementary information is available appears to be \(p_{T} \) dependent. In the lower \(p_{T} \) = 300–400 and 500–600 GeV bins, the background rejection of the “all variables” combination is a factor \(\sim \)1.5 greater than the best twovariable combination, but in the highest \(p_{T} \) bin it is a factor \(\sim \)2.5 greater.
The final column in Figs. 15, 16 and 17 allows us to further explore the all variables performance relative to the pairwise performance. It shows the background rejection for threevariable BDT combinations of \(m_\mathrm{sd}^{\beta =2} + C_2^{\beta =1} + X\), where X is the variable on the yaxis. For jets with \(R=0.4\) and \(R=0.8\), the combination \(m_\mathrm{sd}^{\beta =2} + C_2^{\beta =1}\) is (at least close to) the best performant twovariable combination in every \(p_{T} \) bin considered. For \(R=1.2\) this is not the case, as \(C_2^{\beta =1}\) is superseded by \(\tau _{21}^{\beta =1}\) in performance, as discussed earlier. Thus, in considering the threevariable combination results, it is simplest to focus on the \(R=0.4\) and \(R=0.8\) cases. Here we see that, for the lower \(p_{T} \) = 300–400 and 500–600 GeV bins, adding the third variable to the best twovariable combination brings us to within \(\sim \)15 % of the “all variables” background rejection. However, in the highest \(p_{T} \) = 1.0–1.1 TeV bin, whilst adding the third variable does improve the performance considerably, we are still \(\sim \)40 % from the observed “all variables” background rejection, and clearly adding a fourth or maybe even fifth variable would bring considerable gains. In terms of which variable offers the best improvement when added to the \(m_\mathrm{sd}^{\beta =2} + C_2^{\beta =1}\) combination, it is hard to see an obvious pattern; the best third variable changes depending on the \(p_{T} \) and R considered.
It appears that there is a rich and complex structure in terms of the degree to which the discriminatory information provided by the set of variables considered overlaps, with the degree of overlap apparently decreasing at higher \(p_{T} \). This suggests that in all \(p_{T} \) ranges, but especially at higher \(p_{T} \), there are substantial performance gains to be made by designing a more complex multivariate W tagger.
6.4 Conclusions
We have studied the performance, in terms of the separation of a hadronically decaying W boson from a gluoninitiated jet background, of a number of groomed jet masses, substructure variables, and BDT combinations of the above. We have used this to gain insight into how the discriminatory information contained in the variables overlaps, and how this complementarity between the variables changes with jet \(p_{T} \) and anti\(k_T\) distance parameter R.
In terms of the performance of individual variables, we find that, in agreement with other studies [40], the groomed masses generally perform best, with a background rejection power that increases with larger \(p_{T} \), but which is more consistent with respect to changes in R. We have explained the dependence of the groomed mass performance on \(p_{T} \) and R using the understanding of the QCD mass distribution developed in Sect. 5.4. Conversely, the performance of other substructure variables, such as \(C_2^{\beta =1}\) and \(\tau _{21}^{\beta =1}\), is more susceptible to changes in radius, with background rejection power decreasing with increasing R. This is due to the inherent sensitivity of these observables to soft, wide angle radiation.
The best twovariable performance is obtained by combining a groomed mass with a substructure variable. Which particular substructure variable works best in combination strongly depends on \(p_{\mathrm {T}}\) and R. The variable \(C_2^{\beta =1}\) offers significant complementarity to groomed mass for the smaller values of R investigated (\(R=0.4\) and 0.8), owing to the small degree of correlation between the variables. However, the sensitivity of \(C_2^{\beta =1}\) to soft, wideangle radiation leads to worse discrimination power at \(R=1.2\), where \(\tau _{21}^{\beta =1}\) performs better in combination. The best twovariable performance in each \(p_{T} \) bin examined is obtained for \(C_2^{\beta =1}\) in combination with a groomed mass, using \(R=0.8\), with a performance that is better at higher \(p_{T} \). Our studies also demonstrate the potential for enhancing discrimination by combining groomed and ungroomed mass information, although the use of ungroomed mass in this may be limited in practice by the presence of pileup that is not considered in these studies.
By examining the performance of a BDT combination of all variables considered, it is clear that there are potentially substantial performance gains to be made by designing a more complex multivariate W tagger, especially at higher \(p_{T} \).
7 Top tagging
In this section, we investigate the identification of boosted top quarks using jet substructure. Boosted top quarks result in largeradius jets with complex substructure, containing a bsubjet and a boosted W. As a consequence of the many kinematic differences between top and QCD jets, top taggers are typically complex, with a couple of input parameters necessary for any given algorithm. We study the variation in performance of top tagging techniques with respect to jet \(p_{T} \) and R, reoptimizing the tagger inputs for each kinematic range and jet radius considered. We also investigate the effects of combining dedicated top tagging algorithms with other jet substructure variables, giving insight into the correlations among toptagging variables.
7.1 Methodology
We use the top quark MC samples for each bin described in Sect. 2.2. The analysis relies on FastJet 3.0.3 for jet clustering and calculation of jet substructure variables. Jets are clustered using the anti\(k_T\) algorithm, and only the leading jet is used in each analysis. To ensure similar \(p_{T} \) spectra in each bin an upper and lower \(p_{T} \) cut are applied to each sample after jet clustering. The bins in leading jet \(p_{T} \) for top tagging are 600–700 GeV, 1–1.1 TeV , and 1.5–1.6 TeV. Jets are clustered with radii \(R=0.4\), 0.8, and 1.2; \(R=0.4\) jets are only studied in the 1.5–1.6 TeV bin because the top decay products are all contained within an \(R=0.4\) jet for top quarks with this boost.
 1.
HEPTopTagger
 2.
Johns Hopkins Tagger (JH)
 3.
Trimming with Widentification
 4.
Pruning with Widentification

The ungroomed jet mass.

Nsubjettiness ratios \(\tau _{21}^{\beta =1}\) and \(\tau _{32}^{\beta =1}\), using the “winnertakesall” axes definition.

2point energy correlation function ratios \(C_2^{\beta =1}\) and \(C_3^{\beta =1}\).

The pruned Qjet mass volatility, \(\Gamma _\mathrm{Qjet}\).

HEPTopTagger \(m\in [30,100]\) GeV, \(\mu \in [0.5,1]\)

JH Tagger \(\delta _p\in [0.02,0.15]\), \(\delta _R\in [0.07,0.2]\)

Trimming \(f_\mathrm{cut}\in [0.02,0.14]\), \(R_\mathrm{trim}\in [0.1,0.5]\)

Pruning \(z_\mathrm{cut}\in [0.02,0.14]\), \(R_\mathrm{cut}\in [0.1,0.6]\)
7.2 Single variable performance
We begin by investigating the behaviour of individual jet substructure variables. Because of the rich, threepronged structure of the top decay, it is expected that combinations of masses and jet shapes will far outperform single variables in identifying boosted tops. However, a study of the toptagging performance of single variables facilitates a direct comparison with the W tagging results in Sect. 6, and also allows a straightforward examination of the performance of each variable for different \(p_{T} \) and jet radius.
We also see in Fig. 28b that the top mass from the JH tagger and the HEPTopTagger has superior performance relative to either of the grooming algorithms; this is because the pruning and trimming algorithms do not have inherent Widentification steps and are not optimized for this purpose. Indeed, because of the lack of a Widentification step, grooming algorithms are forced to strike a balance between undergrooming the jet, which broadens the signal peak due to underlying event contamination and features a larger background rate, and overgrooming the jet, which occasionally throws out the bjet and preserves only the W components inside the jet. We demonstrate this effect in Figs. 29 and 30, showing that with 30 % signal efficiency, the optimal performance of the tagger overgrooms a substantial fraction of the jets (\(\sim \)20–30 %), leading to a spurious second peak at \(m_{W}\). This effect is more pronounced at large R and \(p_{T} \), since more aggressive grooming is required in these limits to combat the increased contamination from underlying event and QCD radiation.
7.3 Performance of multivariable combinations
We now consider various BDT combinations of the single variables considered in the last section, using the techniques described in Sect. 4. In particular, we consider the performance of individual taggers such as the JH tagger and HEPTopTagger, which output information about the top and W candidate masses and the helicity angle; for each tagger, all three output variables are combined in a BDT. For trimming and pruning, the output candidate \(m_{W}\) and \(m_{t}\) are combined in a BDT. Finally, we consider the combination of the full set of outputs of each of the above taggers/groomers with the shape variables, as well also a combination of the outputs of the HEPTopTagger and JH tagger. This allows us to determine the degree of complementary information in taggers/groomers and shape variables, as well as between the top tagging algorithms themselves. For all variables with tuneable input parameters, we scan and optimize over realistic values of such parameters, as described in Sect. 7.1.
In Fig. 37, we directly compare the performance of the HEPTopTagger, the JH tagger, trimming, and pruning, in the \(p_{T} = 1{}1.1\) TeV bin with \(R=0.8\), where both \(m_{t}\) and \(m_{W}\) are used in the groomers. Generally, we find that pruning, which does not naturally incorporate subjets into the algorithm, does not perform as well as the others. Interestingly, trimming, which does include a subjetidentification step, performs comparably to the standard HEPTopTagger over much of the range, possibly due to the backgroundshaping observed in Sect. 7.2, although this can change with recent proposed updates to the HEPTopTagger [63, 64]. By contrast, the JH tagger outperforms the other standard algorithms. To determine whether there is complementary information in the mass outputs from different top taggers, we also consider in Fig. 37a multivariable combination of all of the JH and HEPTopTagger outputs. The maximum efficiency of the combined JH and HEPTopTaggers is limited, as some fraction of signal events inevitably fails either one or other of the taggers. We do see a 20–50 % improvement in performance when combining all outputs, which suggests that the different algorithms used to identify the top and W for different taggers contains complementary information.
In Fig. 39 we present the results for multivariable combinations of groomer outputs with and without shape variables. As with the tagging algorithms, combinations of groomers with shape variables improves their discriminating power; combinations with \(\tau _{32}+\tau _{21}\) perform comparably to those with \(C_3+C_2\), and both of these are superior to combinations with the mass volatility, \(\Gamma _\mathrm{Qjet}\). Substantial further improvement is possible by combining the groomers with all shape variables. Not surprisingly, the taggers that lag behind in performance enjoy the largest gain in signalbackground discrimination with the addition of shape variables. Once again, in Fig. 39c, we find that the differences between pruning and trimming are erased when combined with shape information.
Up to this point, we have considered only the combined multivariable performance in the \(p_{T} \) = 1.0–1.1 TeV bin with jet radius \(R=0.8\). We now compare the BDT combinations of tagger outputs, with and without shape variables, at different \(p_{T} \). The taggers are optimized over all input parameters for each choice of \(p_{T} \) and signal efficiency. As with the singlevariable study, we consider anti\(k_T\) jets clustered with \(R=0.8\) and compare the outcomes in the \(p_{T} \) = 500–600 GeV, \(p_{T} \) = 1–1.1 TeV, and \(p_{T} \) = 1.5–1.6 TeV bins. The comparison of the taggers/groomers is shown in Fig. 41. The behaviour with \(p_{T} \) is qualitatively similar to the behaviour of the \(m_{t}\) variable for each tagger/groomer shown in Fig. 32; this suggests that the \(p_{T} \) behaviour of the taggers is dominated by the topmass reconstruction. As before, the standard HEPTopTagger performance degrades slightly with increased \(p_{T} \) due to the background shaping effect (which may be mitigated by recently proposed updates), while the JH tagger and groomers modestly improve in performance.
7.4 Performance at suboptimal working points
Up until now, we have reoptimized our tagger and groomer parameters for each \(p_{T} \), R, and signal efficiency working point. In reality, experiments will choose a finite set of working points to use. When this is taken into account, how will the toptagging performance compare to the optimal results already shown? To address this concern, we replicate our analyses, but optimize the top taggers only for a single \(p_{T} \) bin, single jet radius R, or single signal efficiency, and subsequently apply the same parameters to other scenarios. This allows us to determine the extent to which reoptimization is necessary to maintain the high signaltobackground discrimination power seen in the toptagging algorithms we studied. In this section, we focus on the taggers and groomers, and their combination with shape variables, as the shape variables alone typically do not have any input parameters to optimize.
Optimizing at a single R In Fig. 48, we show the performance of the reconstructed top mass for \(R=0.4\) and 0.8, with all input parameters optimized to \(R=1.2\) TeV bin (and \(p_{T} \) = 1.5–1.6 TeV throughout). This is normalized to the performance using the optimized tagger inputs at each R. While the performance of each variable degrades at small \(\varepsilon _\mathrm{sig}\) compared to the optimized search, the HEPTopTagger fares the worst. It is not surprising that a tagger whose top mass reconstruction is susceptible to backgroundshaping at large R and \(p_{T} \) would require a more careful optimization of parameters to obtain the best performance; recent updates to the tagger algorithm [63, 64] may mitigate the need for this more careful optimization.
Optimizing at a single efficiency The strongest assumption we have made so far is that the taggers can be reoptimized for each signal efficiency point. This is useful for making a direct comparison of the power of different toptagging algorithms, but is not particularly practical for LHC analyses. We now consider the scenario in which the tagger inputs are optimized once, in the \(\varepsilon _\mathrm{sig}=0.3\)–0.35 bin, and then used for all signal efficiencies. We do this in the \(p_{T} \) = 1.0–1.1 TeV bin and with \(R=0.8\).
The performance of each tagger, normalized to its performance optimized in each signal efficiency bin, is shown in Fig. 50 for cuts on the top mass and W mass, and in Fig. 51 for BDT combinations of tagger outputs and shape variables. In both plots, it is apparent that optimizing the taggers in the \(\varepsilon _\mathrm{sig}=0.3\)–0.35 efficiency bin gives comparable performance over efficiencies ranging from 0.2 to 0.5, although performance degrades at substantially different signal efficiencies. Pruning appears to give especially robust signalbackground discrimination without reoptimization, most likely due to the fact that there are no absolute distance or \(p_{T} \) scales that appear in the algorithm. Figures 50 and 51 suggest that, while optimization at all signal efficiencies is a useful tool for comparing different algorithms, it is not crucial to achieve good toptagging performance in experiments.
7.5 Conclusions
We have studied the performance of various jet substructure variables, groomed masses, and top taggers to study the performance of top tagging with different \(p_{T} \) and jet radius parameters. At each \(p_{T} \), R, and signal efficiency working point, we optimize the parameters for those variables with tuneable inputs. Overall, we have found that these techniques, individually and in combination, continue to perform well at high \(p_{T} \), at least at the particlelevel, which is important for future LHC running. In general, the John Hopkins tagger performs best, while jet grooming algorithms underperform relative to the best top taggers due to the lack of an optimized Widentification step. Tagger performance can be improved by a further factor of 2–4 through combination with jet substructure variables such as \(\tau _{32}\), \(C_3\), and \(\Gamma _\mathrm{Qjet}\). When combined with jet substructure variables, the performance of various groomers and taggers becomes very comparable, suggesting that, taken together, the variables studied are sensitive to nearly all of the physical differences between top and QCD jets at particlelevel. A small improvement is also found by combining the Johns Hopkins and HEPTopTaggers, indicating that different taggers are not fully correlated. The degree to which these findings continue to hold under more realistic pileup and detector configurations is, however, not addressed in this analysis and left to future study.
Comparing results at different \(p_{T} \) and R, toptagging performance is generally better at smaller R due to less contamination from uncorrelated radiation. Similarly, most variables perform better at larger \(p_{T} \) due to the higher degree of collimation of radiation. Some variables fare worse at higher \(p_{T} \), such as the Nsubjettiness ratio \(\tau _{32}\) and the Qjet mass volatility \(\Gamma _\mathrm{Qjet}\), as higher\(p_{T} \) QCD jets have more and harder emissions that fake the topjet substructure. The standard HEPTopTagger algorithm is also worse at high \(p_{T} \) due to the tendency of the tagger to shape backgrounds around the top mass. This is unsurprising, given that the HepTopTagger was specifically designed for a lower \(p_{\mathrm {T}}\) range than that considered here; recently proposed updates may improve performance at high \(p_{T} \) and R [63, 64]. The \(p_{T} \) and Rdependence of the multivariable combinations is dominated by the \(p_{T} \) and Rdependence of the top mass reconstruction component of the tagger/groomer.
Finally, we consider the performance of various tagger and jet substructure variable combinations under the more realistic assumption that the input parameters are only optimized at a single \(p_{T} \), R, or signal efficiency, and then the same inputs are used at other working points. Remarkably, the performance of all variables is typically within a factor of 2 of the fully optimized inputs, suggesting that while optimization can lead to substantial gains in performance, the general behavior found in the fully optimized analyses extends to more general applications of each variable. In particular, the performance of pruning typically varies the least when comparing suboptimal working points to the fully optimized tagger due to the scaleinvariant nature of the pruning algorithm.
8 Summary and conclusions
We focused on the discrimination of quark jets from gluon jets, and the discrimination of boosted W bosons and top quarks from the QCD backgrounds. For each, we have identified the bestperforming jet substructure observables at particle level, both individually and in combination with other observables. In doing so, we have also provided a physical picture of why certain sets of observables are (un)correlated. Additionally, we have investigated how the performance of jet substructure observables varies with R and \(p_{T} \), identifying observables that are particularly robust against or susceptible to these changes. In the case of q / g tagging, it seems that the ideal performance can be nearly achieved by combining the most powerful discriminant, the number of constituents of a jet, with just one other variable, \(C_1^{\beta =1}\) (or \(\tau _1^{\beta =1}\)). Many of the other variables considered are highly correlated and provide little additional discrimination. For both top and W tagging, the groomed mass is a very important discriminating variable, but one that can be substantially improved in combination with other variables. There is clearly a rich and complex relationship between the variables considered for W and top tagging, and the performance and correlations between these variables can change considerably with changing jet \(p_{T} \) and R. In the case of W tagging, even after combining groomed mass with two other substructure observables, we are still some way short of the ultimate tagger performance, indicating the complexity of the information available, and the complementarity between the observables considered. In the case of top tagging, we have shown that the performance of both the John Hopkins and HEPTopTagger can be improved when their outputs are combined with substructure observables such as \(\tau _{32}\) and \(C_{3}\), and that the performance of a discriminant built from groomed mass information plus substructure observables is very comparable to the performance of the taggers. We have optimized the top taggers for particular values of \(p_{T} \), R, and signal efficiency, and studied their performance at other working points. We have found that the performance of observables remains within at most a factor of two of the optimized value, suggesting that the performance of jet substructure observables is not significantly degraded when tagger parameters are only optimized for a few select benchmark points.
In all of q / g, W and top tagging, we have observed that the tagging performance improves with increasing \(p_{T} \). However, whereas for q / g and top tagging the performance improves with decreasing R (for the range of R considered here), the dependence on R for W tagging is more complex, with a peak performance at \(R=0.8\) for each \(p_{T} \) bin considered.
Our analyses were performed with ideal detector and pileup conditions in order to most clearly elucidate the underlying physical scaling with \(p_{T} \) and R. At higher boosts, detector resolution effects will become more important, and with the higher pileup expected at Run II of the LHC, pileup mitigation will be crucial for future jet substructure studies. Future studies will be needed to determine which of the observables we have studied are most robust against pileup and detector effects, and our analyses suggest particularly useful combinations of observables to consider in such studies.
At the new energy frontier of Run II of the LHC, boosted jet substructure techniques will be more central to our searches for new physics than ever before. By achieving a deeper understanding of the underlying structure of quark, gluon, W and topinitiated jets, as well as the relations between observables sensitive to their respective structures, it is hoped that more sophisticated analyses can be performed that will maximally extend the reach for new physics.
Footnotes
 1.
 2.
The shoulder label will become more clear when examining groomed jet mass distributions.
 3.Note that we here choose to report the rejection for a higher signal efficiency than the 50 % that was used in the q / g tagging studies of Sect. 5, because the rejection rates in W tagging are considerably higher.
 4.
Notes
Acknowledgments
We thank the Department of Physics at the University of Arizona for hosting and providing support for the BOOST 2013 workshop, and the US Department of Energy for their support of the workshop. We especially thank Vivian Knight (University of Arizona) for her help with the organization of the of the workshop. We also thank Prof. J. Boelts of the University of Arizona School of Art VisCom program and his Fall 2012 ART 465 class for organizing the design competition for the workshop poster. In particular, we thank the winner of the competition, Ms. Hallie Bolonkin, for creating the final design.
References
 1.Boost, SLAC National Accelerator Laboratory, 9–10 July 2009 (2009). http://wwwconf.slac.stanford.edu/Boost2009
 2.Boost, University of Oxford, 22–25 June 2010 (2010). http://www.physics.ox.ac.uk/boost2010
 3.Boost, Princeton University, 22–26 May 2011 (2011). https://indico.cern.ch/event/138809/
 4.Boost, IFIC Valencia, 23–27 July 2012 (2012). http://ific.uv.es/boost2012
 5.Boost, University of Arizona, 12–16 August 2013 (2013). https://indico.cern.ch/event/215704/
 6.Boost, University College London, 18–22 August 2014 (2014). http://www.hep.ucl.ac.uk/boost2014/
 7.A. Abdesselam, E.B. Kuutmann, U. Bitenc, G. Brooijmans, J. Butterworth, et al., Eur. Phys. J. C 71, 1661 (2011). doi: 10.1140/epjc/s100520111661y
 8.A. Altheimer, S. Arora, L. Asquith, G. Brooijmans, J. Butterworth, et al., J. Phys. G 39, 063001 (2012). doi: 10.1088/09543899/39/6/063001
 9.A. Altheimer, A. Arce, L. Asquith, J. Backus Mayes, E. Bergeaas Kuutmann, et al., Eur. Phys. J. C 74(3), 2792 (2014). doi: 10.1140/epjc/s1005201427928
 10.T. Plehn, M. Spannowsky, M. Takeuchi, D. Zerwas, JHEP 1010, 078 (2010). doi: 10.1007/JHEP10(2010)078
 11.D.E. Kaplan, K. Rehermann, M.D. Schwartz, B. Tweedie, Phys. Rev. Lett. 101, 142001 (2008). doi: 10.1103/PhysRevLett.101.142001
 12.J. Alwall, M. Herquet, F. Maltoni, O. Mattelaer, T. Stelzer, JHEP 1106, 128 (2011). doi: 10.1007/JHEP06(2011)128
 13.Y. Gao, A.V. Gritsan, Z. Guo, K. Melnikov, M. Schulze, et al., Phys. Rev. D 81, 075022 (2010). doi: 10.1103/PhysRevD.81.075022
 14.S. Bolognesi, Y. Gao, A.V. Gritsan, K. Melnikov, M. Schulze, et al., Phys. Rev. D 86, 095031 (2012). doi: 10.1103/PhysRevD.86.095031
 15.I. Anderson, S. Bolognesi, F. Caola, Y. Gao, A.V. Gritsan, et al., Phys. Rev. D 89, 035007 (2014). doi: 10.1103/PhysRevD.89.035007
 16.J. Pumplin, D. Stump, J. Huston, H. Lai, P.M. Nadolsky, et al., JHEP 0207, 012 (2002). doi: 10.1088/11266708/2002/07/012
 17.T. Sjostrand, S. Mrenna, P.Z. Skands, Comput. Phys. Commun. 178, 852 (2008). doi: 10.1016/j.cpc.2008.01.036
 18.A. Buckley, J. Butterworth, S. Gieseke, D. Grellscheid, S. Hoche, et al., Phys. Rept. 504, 145 (2011). doi: 10.1016/j.physrep.2011.03.005
 19.T. Gleisberg, S. Hoeche, F. Krauss, M. Schonherr, S. Schumann, et al., JHEP 0902, 007 (2009). doi: 10.1088/11266708/2009/02/007
 20.S. Schumann, F. Krauss, JHEP 0803, 038 (2008). doi: 10.1088/11266708/2008/03/038
 21.F. Krauss, R. Kuhn, G. Soff, JHEP 0202, 044 (2002). doi: 10.1088/11266708/2002/02/044
 22.T. Gleisberg, S. Hoeche, JHEP 0812, 039 (2008). doi: 10.1088/11266708/2008/12/039
 23.S. Hoeche, F. Krauss, S. Schumann, F. Siegert, JHEP 0905, 053 (2009). doi: 10.1088/11266708/2009/05/053
 24.M. Schonherr, F. Krauss, JHEP 0812, 018 (2008). doi: 10.1088/11266708/2008/12/018
 25.S. Bethke, et al., Phys. Lett. B 213, 235 (1988). doi: 10.1016/03702693(88)910325
 26.M. Cacciari, G.P. Salam, G. Soyez, JHEP 0804, 063 (2008). doi: 10.1088/11266708/2008/04/063
 27.Y.L. Dokshitzer, G. Leder, S. Moretti, B. Webber, JHEP 9708, 001 (1997). doi: 10.1088/11266708/1997/08/001
 28.M. Wobisch, T. Wengler, in Proceedings of the Monte Carlo Generators for HERA Physics Workshop, Hamburg, ed. by A. Doyle (1998)Google Scholar
 29.S. Catani, Y.L. Dokshitzer, M. Seymour, B. Webber, Nucl. Phys. B 406, 187 (1993). doi: 10.1016/05503213(93)90166M
 30.S.D. Ellis, D.E. Soper, Phys. Rev. D 48, 3160 (1993). doi: 10.1103/PhysRevD.48.3160
 31.S.D. Ellis, A. Hornig, T.S. Roy, D. Krohn, M.D. Schwartz, Phys. Rev. Lett. 108, 182003 (2012). doi: 10.1103/PhysRevLett.108.182003
 32.S.D. Ellis, A. Hornig, D. Krohn, T.S. Roy, JHEP 1501, 022 (2015). doi: 10.1007/JHEP01(2015)022
 33.S.D. Ellis, C.K. Vermilion, J.R. Walsh, Phys. Rev. D 81, 094023 (2010). doi: 10.1103/PhysRevD.81.094023
 34.D. Krohn, J. Thaler, L.T. Wang, JHEP, 084 (2010). doi: 10.1007/JHEP02(2010)084
 35.J.M. Butterworth, A.R. Davison, M. Rubin, G.P. Salam, Phys. Rev. Lett. 100, 242001 (2008). doi: 10.1103/PhysRevLett.100.242001
 36.A.J. Larkoski, S. Marzani, G. Soyez, J. Thaler, JHEP 1405, 146 (2014). doi: 10.1007/JHEP05(2014)146
 37.M. Dasgupta, A. Fregoso, S. Marzani, G.P. Salam, JHEP 1309, 029 (2013). doi: 10.1007/JHEP09(2013)029
 38.V. Khachatryan, et al., JHEP 1408, 173 (2014). doi: 10.1007/JHEP08(2014)173
 39.G. Aad, et al., New J. Phys. 16(11), 113013 (2014). doi: 10.1088/13672630/16/11/113013
 40.Performance of Boosted W Boson Identification with the ATLAS Detector. Tech. Rep. ATLPHYSPUB2014004, CERN, Geneva (2014)Google Scholar
 41.J. Thaler, K. Van Tilburg, JHEP 1103, 015 (2011). doi: 10.1007/JHEP03(2011)015
 42.A.J. Larkoski, D. Neill, J. Thaler, JHEP 1404, 017 (2014). doi: 10.1007/JHEP04(2014)017
 43.A.J. Larkoski, J. Thaler, JHEP 1309, 137 (2013). doi: 10.1007/JHEP09(2013)137
 44.A.J. Larkoski, G.P. Salam, J. Thaler, JHEP 1306, 108 (2013). doi: 10.1007/JHEP06(2013)108
 45.S. Chatrchyan, et al., JHEP 1204, 036 (2012). doi: 10.1007/JHEP04(2012)036
 46.A.J. Larkoski, J. Thaler, W.J. Waalewijn, JHEP 1411, 129 (2014). doi: 10.1007/JHEP11(2014)129
 47.A. Hoecker, P. Speckmayer, J. Stelzer, J. Therhaag, E. von Toerne, H. Voss, An example of the BDT settings used in these studies are as follows: NTrees=1000; BoostType=Grad; Shrinkage=0.1; UseBaggedGrad=F; nCuts=10000; MaxDepth=3; UseYesNoLeaf=F; nEventsMin=200, PoS ACAT, 040 (2007)Google Scholar
 48.G. Aad, et al., Eur. Phys. J. C 74(8), 3023 (2014). doi: 10.1140/epjc/s100520143023z
 49.J. Gallicchio, M.D. Schwartz, JHEP 1304, 090 (2013). doi: 10.1007/JHEP04(2013)090
 50.A.J. Larkoski, I. Moult, D. Neill, JHEP 1409, 046 (2014). doi: 10.1007/JHEP09(2014)046
 51.M. Procura, W.J. Waalewijn, L. Zeune, JHEP 1502, 117 (2015). doi: 10.1007/JHEP02(2015)117
 52.J. Gallicchio, M.D. Schwartz, Phys. Rev. Lett. 107, 172001 (2011). doi: 10.1103/PhysRevLett.107.172001
 53.C. Collaboration (2013)Google Scholar
 54.H.n. Li, Z. Li, C.P. Yuan, Phys. Rev. D 87, 074025 (2013). doi: 10.1103/PhysRevD.87.074025
 55.M. Dasgupta, K. KhelifaKerfa, S. Marzani, M. Spannowsky, JHEP 1210, 126 (2012). doi: 10.1007/JHEP10(2012)126
 56.M. Dasgupta, A. Fregoso, S. Marzani, A. Powling, Eur. Phys. J. C 73(11), 2623 (2013). doi: 10.1140/epjc/s1005201326233
 57.Y.T. Chien, R. Kelley, M.D. Schwartz, H.X. Zhu, Phys. Rev. D 87(1), 014010 (2013). doi: 10.1103/PhysRevD.87.014010
 58.T.T. Jouttenus, I.W. Stewart, F.J. Tackmann, W.J. Waalewijn, Phys. Rev. D 88(5), 054031 (2013). doi: 10.1103/PhysRevD.88.054031
 59.Z.L. Liu, C.S. Li, J. Wang, Y. Wang, JHEP 1504, 005 (2015). doi: 10.1007/JHEP04(2015)005
 60.S.D. Ellis, C.K. Vermilion, J.R. Walsh, Phys. Rev. D 80, 051501 (2009). doi: 10.1103/PhysRevD.80.051501
 61.Y. Cui, Z. Han, M.D. Schwartz, Phys. Rev. D 83, 074023 (2011). doi: 10.1103/PhysRevD.83.074023
 62.D.E. Soper, M. Spannowsky, JHEP 1008, 029 (2010). doi: 10.1007/JHEP08(2010)029
 63.C. Anders, C. Bernaciak, G. Kasieczka, T. Plehn, T. Schell, Phys. Rev. D 89, 074047 (2014). doi: 10.1103/PhysRevD.89.074047
 64.G. Kasieczka, T. Plehn, T. Schell, T. Strebler, G.P. Salam (2015), JHEP 1506, 203 (2015). doi: 10.1007/JHEP06(2015)203
 65.S. Schaetzel, M. Spannowsky, Phys. Rev. D 89(1), 014007 (2014). doi: 10.1103/PhysRevD.89.014007
Copyright information
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.
Funded by SCOAP^{3}