In searches for new physics in the energy regime of the LHC, it is becoming
increasingly important to distinguish single-jet objects that originate from the
merging of the decay products of W bosons produced with high transverse
momenta from jets initiated by single partons. Algorithms are defined to identify
such W jets for different signals of interest, using techniques that are also
applicable to other decays of bosons to hadrons that result in a single jet,
such as those from highly boosted Z and Higgs bosons. The efficiency for
tagging W jets is measured in data collected with the CMS detector at a
center-of-mass energy of 8 TeV, corresponding to an integrated luminosity of
19.7 fb
Article funded by SCOAP3
200$\GeV, the final state hadrons from the W$\to \cPaq\cPq'$ decay merge into a single jet, and the traditional analysis techniques relying on resolved jets are no longer applicable. However, in such cases the analysis of jet substructure can be used to identify those jets arising from decays of W, Z or H bosons. Because the values of the mass of the W and Z bosons are rather close to each other, we do not distinguish the two, and refer to such jets collectively as V jets, while the Higgs boson mass is significantly higher and can be distinguished. The focus of this paper is solely on the identification of W jets, however, we note that many of the procedures described are equally applicable for handling highly boosted Z and H bosons. Measurements of jet substructure observables related to identification of W bosons have been previously reported by CMS~\cite{jetmass,topwtag_pas} and ATLAS~\cite{atlasjetmass2,atlasjetmass}. Several searches at CMS have employed jet substructure techniques for identifying (``tagging") W jets and Z jets. These include searches in all-jet $\ttbar$ final states~\cite{hadronictop,hadronictop7TeV}, single and pair produced V bosons in inclusive dijet final states~\cite{hadronicvv,hadronicvv7TeV}, and searches in the VV final states, where one of the vector bosons decays leptonically~\cite{semileptonicZZ,semileptonicZZ7TeV}. In these searches, a variety of different observables have been used to identify the V jets. This paper aims to compare and measure the performance in 8\TeV pp collisions of various jet substructure techniques that can be used to distinguish V jets from more ordinary quark- and gluon-initiated jets, which we refer to as QCD jets. This paper is organized as follows. The CMS detector is described in section~\ref{sec:detector}. The procedures chosen for the reconstruction of events are described in section~\ref{sec:reco}. The data and simulated events used in our studies as well as the event selection criteria are presented in section~\ref{sec:samples}. In section~\ref{sec:algo}, through Monte Carlo (MC) simulation, we investigate the performance of jet substructure observables used to identify W jets, in order to find the best discriminants for such events. We compare these observables in different kinematic regimes, and examine factors that contribute to their performance. Their distributions in data are compared to those in MC simulations in section~\ref{sec:efficiency}, to learn how well current MC simulations can model the physical processes responsible for jet substructure. The methods used to extract data-to-simulation scale factors needed to correct W boson tagging efficiencies obtained from MC simulation are discussed in section~\ref{sec:efficiency}, and the mistagging rate of QCD jets in data is extracted. The goal being to provide these as reference tools for analyzing events with jets from V bosons in the final state. Finally, we give a summary of our studies in section~\ref{sec:summary}. ]]>
200\GeV$. Two algorithms are used to reconstruct muons~\cite{CMS-PAPER-MUO-10-004}: one proceeds from the inner tracker outwards, while the other starts from tracks measured in the muon chambers and matches them to those reconstructed in the silicon tracker. Muons are identified using selection criteria optimized for high-\pt muons~\cite{CMS-PAPER-MUO-10-004}. The selected muon candidates must be isolated from charged hadron activity in the detector by requiring the scaler sum of transverse momenta ($I_\mathrm{tk}$) of tracks within a cone of $\Delta R = \sqrt{(\Delta \phi)^2 + (\Delta \eta)^2} < 0.3$ around the muon track, divided by the muon \pt, to be $I_\mathrm{tk}/\pt < 0.1$. Electrons are reconstructed using a Gaussian-sum filter algorithm~\cite{Chatrchyan-ml-2013dga,CMS-PAS-EGM-10-004}, and each electron candidate must furthermore pass the identification and isolation criteria optimized for high \pt electrons~\cite{CMS-PAS-EGM-10-004}. ]]>
30\GeV$ and $\abs{\eta} < 2.4$. The two jets of highest $\pt$ are required to have a pseudorapidity separation $\abs{\Delta\eta}<1.3$, which rejects a large fraction of QCD multijet events. Finally, the dijet invariant mass is required to be larger than 890\GeV. This threshold is chosen such that the trigger selection for events with dijet masses above this threshold is 99\% efficient. W-tagging is studied using the leading jet in the selected dijet events, with additional requirements set on jet $\pt$. The main goal of the kinematic selection of the W+jet sample is to isolate a sample of events with a highly boosted topology consistent with a leptonically decaying W boson recoiling against a high \pt jet. The W+jet sample, as well as the $\ttbar$ sample discussed below, are collected using single-lepton triggers. The lepton \pt thresholds of these triggers are 40 and 80\GeV for the muon and electron channels, respectively. Offline, at least one muon or one electron, with respective $\pt >50$\GeV or $\pt >90$\GeV, is required within respective $\abs{\eta}<2.1$ or $\abs{\eta}<2.5.$ Events containing additional muons with $\pt > 20$\GeV and $\abs{\eta}<2.4$ or additional electrons with $\pt > 20$\GeV and $\abs{\eta}<2.5$ are rejected, in order to improve the purity of W+jet events. A requirement on the imbalance in transverse momentum (\ETmiss) is used to reduce the QCD multijet background. The \ETmiss is computed from the negative transverse component of the vector sum of all PF candidate momenta, and is required to be above 50\GeV or 80\GeV for the muon and electron channel. The threshold is higher in the electron channel to further suppress the larger background from multijet processes. The \pt of the leptonically decaying W boson and of the CA8 jet with highest \pt, are required to be $>$200\GeV. Additional criteria are applied to ensure that the leptonic W boson and the CA8 jet are mostly back-to-back in the transverse plane: $\Delta R$ between the lepton and the jet must be greater than $\pi/2$; the azimuthal distance $\Delta \phi$ between \ETmiss and the jet must be greater than 2.0 radians; and the azimuthal distance $\Delta \phi$ between the leptonically decaying W boson and the CA8 jet must also be greater than 2.0 radians. Finally, a cutoff on additional jet activity in the event is applied to reduce the amount of \ttbar\ background. We identify additional b jet candidates in the event by requiring that an AK5 jet, with an angular distance of $\Delta R>0.8$ to the CA8 jet, passes the CSV b-tagging discriminant~\cite{Chatrchyan-ml-1494669} using a medium working point. To suppress \ttbar\ background in the W+jet selections described above, we require that no such b jets be present in the event. To select the $\ttbar$ sample, we use the kinematic selection described above for the W+jet topology, but instead require that there is at least one AK5 b jet, with an angular distance of $\Delta R>0.8$ to the CA8 jet considered as W jet candidate. To increase the statistical precision of the sample, we select the CA8 jet with the largest mass and with $\Delta\phi$ between the lepton and the jet greater than $\pi/2$ as W jet candidate, rather than the highest \pt CA8 jet. ]]>
200\GeV$. Simulation with different parton shower models of \PYTHIA~6, \PYTHIA~8 and \HERWIG{++} are also compared. In figure~\ref{fig:dj_datmc_mass}, the pruned jet mass distribution is shown for both data and simulation in the dijet and W+jet samples that probe the W-tagging variables using QCD jets. We find that the agreement is good between data and simulation, but \HERWIG{++} agrees better than \PYTHIA~6, and \PYTHIA~8 shows best agreement. Similar findings have been reported in refs.~\cite{topwtag_pas, jetmass, atlasjetmass}. The $\tau_2 / \tau_1$ variable is also shown and found to agree better with \HERWIG{++} and best with \PYTHIA~8. \begin{figure} \centerline{ \includegraphics[width=0.45\textwidth]{jet_mass_pr_0.pdf} \includegraphics[width=0.45\textwidth]{jet_tau2tau1_0.pdf}} \centerline{ \includegraphics[width=0.45\textwidth]{substructure_pas_mass_2.pdf} \includegraphics[width=0.45\textwidth]{substructure_pas_tau21_aftermass_2.pdf} } ]]>
1.3$. The uncertainty from pileup is determined by moving the minimum bias cross section within its measured uncertainty of 6\%~\cite{minbiasxsec}. The scale factors for lepton and b jet identification are also changed within their uncertainties. Finally, uncertainties in the energy and momentum scale and resolution of leptons and jets in the event are propagated to an uncertainty on the \MET. The results are summarized in table~\ref{tab:systematics}. The dominant systematic effect on the scale factor for the efficiency is from modeling of the parton shower and PDF, with a systematic uncertainty of 6.0\%. The quadratic sum of systematic uncertainties of 7.6\% is comparable to the statistical uncertainty on the scale factor of 6.4\%. \begin{table} \centering\renewcommand\arraystretch{1.3} \begin{tabular}{|c|c|}\hline Source & Effect on the scale factor \\ \hline Parton showering + PDF & 6.0\% \\ Underlying event & $<$0.5\% \\ W-polarization & 2.0\% \\ Pileup & 1.8\% \\ Nearby jets & 2.4\% \\ Jet mass scale & $<$0.5\% \\ Jet mass resolution & 1.9\% \\ Jet energy scale & 1.9\% \\ Jet energy resolution & 0.9\% \\ Lepton ID & $<$0.5\% \\ b-tagging & $<$0.5\% \\ \MET & $<$0.5\% \\ \hline Total systematic & 7.6\% \\ \hline Statistical & 6.4\% \\ \hline Total & 9.9\% \\\hline \end{tabular} ]]>