Study of the dijet mass spectrum in pp to W + jets events at sqrt(s) = 7 TeV

We report an investigation of the invariant mass spectrum of the two jets with highest transverse momentum in pp to W+2-jet and W+3-jet events to look for resonant enhancements. The data sample corresponds to an integrated luminosity of 5.0 inverse femtobarns collected with the CMS detector at sqrt(s) = 7 TeV. We find no evidence for the anomalous structure reported by the CDF Collaboration, and establish an upper limit of 5.0 pb at 95% confidence level on the production cross section for a generic Gaussian signal with mass near 150 GeV. Additionally, we exclude two theoretical models that predict a CDF-like dijet resonance near 150 GeV.


1
The CDF Collaboration reported evidence for an excess in the mass range 120-160 GeV in the invariant mass (m jj ) spectrum of the two leading transverse-momentum (p T ) jets produced in pp → W+2-jet events with a cross section of 4 pb [1]. The DØ Collaboration carried out a similar analysis but did not confirm the CDF result, instead setting a 95% confidence level (CL) upper limit of 1.9 pb on the cross section [2]. This Letter details the search for a bumplike enhancement in the m jj spectrum in events with a W boson using 5.0 fb −1 of data collected from pp collisions at √ s = 7 TeV with the Compact Muon Solenoid (CMS) detector at the CERN Large Hadron Collider (LHC) during 2010 and 2011.
We search for a resonance with a width consistent with detector resolution as reported by CDF. We further investigate three representative models, a technicolor π T from the decay of a technicolor ρ T [3], a leptophobic Z decaying to two jets [4], and the standard model (SM) Higgs boson (m H = 150 GeV) produced in association with a W boson (referred to as WH production) and decaying to a pair of jets. For the unknown state with detector resolution, we follow the convention used at the Tevatron of using the conservative WH simulation for analysisdependent quantities like efficiencies and acceptances. The WH production cross section at the LHC is negligible compared to contributions from other SM processes, which overwhelm any contribution to this analysis from WH → νjj decays for m H ≈ 125 GeV [5,6]. must be greater than 50 GeV, where and φ and φ E / T are the azimuthal angles of the lepton and E / T , respectively. Events with more than one identified lepton are vetoed.
We reconstruct jets and E / T [9, 10] with the particle-flow algorithm [11], which combines information from several subdetectors. The jet finding uses the anti-k T clustering algorithm [12] with a distance parameter of 0.5. We require |η jet | < 2.4 to ensure that they lie within the tracker acceptance, and a minimum jet p T of 30 GeV. Jets must satisfy identification criteria that eliminate jet candidates originating from noisy channels in the hadron calorimeter [13]. Jet-energy corrections are applied to account for the non-linear response of the calorimeters to the particle energies and other instrumental effects. These corrections are based on in-situ measurements using dijet, γ + jet, and Z+jet data samples [14]. Overlapping minimum-bias events from other pp collisions (pile-up) and the underlying event can contribute additional energy to the reconstructed jets. The median energy density due to pile-up is evaluated in each event and the corresponding energy is subtracted from each jet [15]. In addition, tracks that do not originate from the primary vertex are not considered for jet clustering [16]. We verify that the procedures successfully remove the dependence of jet response on the number of interactions in a single event. The jet p T resolution varies from 15% at p T = 40 GeV to 6% at p T = 400 GeV [14]. We evaluate the mass resolution σ jj for a selected jet pair using simulation and verify it using hadronic W decays in data. We find σ jj to be 10% of m jj for masses around 150 GeV.
We require p j 1 T + p j 2 T > 45 GeV and |∆η(j 1 , j 2 )| < 1.2, where the jets are numbered in order of decreasing p T . We retain events with exactly two or exactly three jets satisfying p T > 30 GeV and with the leading jet having p T > 40 GeV and pointing more than 0.4 rad in azimuth from the direction of the E / T . The selected jets and the lepton from the W decay must originate from the same primary vertex. Additionally, we impose 0.3 < p j 2 T /m jj < 0.7 to take advantage of the Jacobian nature of resonant dijet production as observed in simulation studies compared with nonresonant W plus jets production. W production with two or more jets dominates the selected sample. Smaller contributions come from top-pair and single-top decays, Drell-Yan events with two or more jets, multijet production, and WW and WZ diboson production where one W decays into leptons and the other W or Z decays into quarks.
The shapes of the m jj distributions for background processes are modeled using samples of simulated events. The MADGRAPH5 1.3.30 [17] event generator produces parton-level events with a W boson and up to four partons on the basis of matrix-element (ME) calculations. (The Tevatron experiments used the ALPGEN generator [18].) The ME-parton shower matching scale µ is taken to be 20 GeV [19], and the factorization and renormalization scales are set to q 2 = M 2 W + p 2 T,W . Samples of tt and Drell-Yan events are also generated with MADGRAPH. Single-top production is modeled with POWHEG 1.0 [20]. Multijet and diboson samples (WW, WZ, ZZ) are generated with PYTHIA 6.422 [21]. PYTHIA provides the parton shower simulation in all cases, with parameters of the underlying event set to the Z2 tune [22]. The set of parton distribution functions used is CTEQ6LL [23]. A GEANT4-based simulation [24] of the CMS detector is used in the production of all Monte Carlo (MC) samples. Multiple proton-proton interactions within a bunch crossing are simulated, and the triggers are emulated. All simulated events are reconstructed and analyzed with the same software as data.
We generate signal samples for the WH model using PYTHIA, with parameters corresponding We determine the contributions of the known SM processes to the observed m jj spectrum by means of an extended unbinned maximum-likelihood fit in the range between 40 GeV and 400 GeV. We fit separately in four event categories, {µ, e} × {2-jet, 3-jet}, because the background compositions differ. The m jj signal region, 123 to 186 GeV, corresponding to ±2σ jj , is excluded from this fit in order to arrive at an unbiased estimate of a possible resonant enhancement in this region. Table 1 lists the SM processes included in the fit. The W plus jets normalization is a free fit parameter because it is by far the dominant background. We allow the normalizations of the other background components to vary within Gaussian constraints around the central values also listed in Table 1. The central values for all processes except multijet come from next-to-leadingorder (NLO), next-to-next-to-leading-log (NNLL) or next-to-NLO (NNLO) calculations, and the constraints reflect the published uncertainties. We derive templates for the m jj distribution for each background from simulation except for the multijet events, which contribute when jets are misidentified as leptons. In a separate fit to events that fail the lepton isolation requirements, we determine the central value of the multijet normalization, the constraint on the normalization and the template for the m jj distribution [9]. The fit to data determines the correlations among the various fit parameters.
The default CMS MADGRAPH sample of the dominant W plus jets background does not describe well the m jj spectrum in the m jj sidebands. Four alternative samples of W events, with the scales µ and q increased and reduced by a factor two with respect to those of the default, fail to provide significant improvement. Thus, we employ an empirically-driven combination of three shapes to describe this component in the fit model: , where F W+jets denotes the m jj shape from simulation. The parameters µ 0 (µ ) and q 0 (q ) correspond to the default (alternative) values of µ and q, respectively, while fractional contributions α and β are free to vary between 0 and 1. We take µ = 2µ 0 or 0.5µ 0 (q = 2q 0 or 0.5q 0 ), depending on which alternative sample provides a better fit to data. Furthermore, we verify, via pseudo-experiment simulations generated with an alternate shape, that the function in the above equation has sufficient freedom to describe the W plus jets shape.  with the fitted projections of the contributions of various SM processes. Figure 1(b) shows the same distribution after subtraction of all SM contributions from data except electroweak diboson WW/WZ events. No peak is visible in the spectrum except that near 80 GeV due to diboson events. Figure 1(c) shows the bin-by-bin pull. Table 2 presents the yields of the SM components obtained from the fit. The sum of all the contributions is compared to the number of observed events. All numbers except those in the last two rows are for the m jj range of 40 to 400 GeV. The last two rows compare the observed number of events and the number predicted by the fit in the m jj range of 123 to 186 GeV. The data agree with the SM expectations, and we find no significant excess in the signal region. We observe a sizable deficit in the muon 2-jet data with respect to the prediction from our model. We do not observe similar deviations in the other three categories, suggesting it is a fluctuation and not a systematic bias.
We validate the fit procedure by performing pseudo-experiments. In each experiment, we generate the m jj pseudo-data of the SM processes, including the correlations taken from the fit to data, and then fit each pseudo-data sample. The results indicate that the bias on the total yield is below 0.2% and that the fit underestimates the total yield uncertainty by about 30%. These effects are corrected for in the final result. Uncertainties in the jet energy are estimated using a sample of W bosons decaying hadronically in a pure sample of semileptonic tt events. The mean and resolution of the reconstructed dijet mass distribution in data agree within 0.6% with the expectation from simulation. A small difference in E / T resolution [10] between data and simulation affects the signal acceptance for the new physics models under consideration at the 0.5% level. Further systematic uncertainties are due to the uncertainty of the trigger efficiency estimates (1%) and the estimate of lepton reconstruction and selection efficiency (2%) [9]. The uncertainty on the integrated luminosity is 2.2% [31].
We scrutinize the dijet mass spectrum near 150 GeV, searching for a technicolor, leptophobic Z , or WH resonant enhancement. We also use a generic signal model obtained by convolving a delta function centered at m jj = 150 GeV with a Gaussian function having width equal to σ jj . Figure 1(b) shows this generic signal shape. The expected number of signal events at the LHC for a given cross section at the Tevatron can be estimated by considering the ratio of the  [21] 0.059 0.060 0.019 0.038 0.013 predicted cross sections for our reference process, WH production with M H = 150 GeV. This process is dominated by quark-antiquark (qq) annihilation. As qq processes have the smallest increase in parton luminosity from the Tevatron to the LHC, this choice provides a conservative limit. We therefore assume where σ WH LHC = 300.1 fb [32] and σ WH Tevatron = 71.8 fb [33]. A generic Gaussian signal normalized to σ Tevatron = 4 pb corresponds to σ LHC = 16.7 pb. Table 3 contains the values of σ LHC times the branching fraction to jets and of the overall efficiency times acceptance εA for the models considered.
Since we observe no resonant enhancement, we proceed to set exclusion limits using a modified frequentist CL S method [34,35] with profile likelihood as the test statistic. Inputs to the limitsetting procedure are the m jj distribution obtained by combining the SM components from the fit, the observed distribution in data, the expectation from the dijet resonance model under consideration and the uncertainties associated with these quantities. Figure 2(a) shows the observed and expected CL S values versus cross section for a generic Gaussian signal, after combining the results of all four event categories. We set a 95% CL upper limit of 5.0 pb and a 99.9% CL upper limit of 8.5 pb on the dijet production cross section for a generic resonance with WH-like εA. Because we have minimal sensitivity to WH, we compare the limit in Fig. 2(b) to 100 times the SM cross section as an illustration.
In summary, we have studied the invariant mass spectrum of the two jets with highest transverse momentum in pp → W+2-jet and W+3-jet events, with the W decaying leptonically to a muon or electron. The analyzed data sample corresponds to an integrated luminosity of 5.0 fb −1 at √ s = 7 TeV. We find no evidence for a resonant enhancement near a dijet mass of 150 GeV, as reported by the CDF Collaboration, and set upper limits on the dijet production cross section of 5.0 pb at 95% CL and 8.5 pb at 99.9% CL. Two theoretical models, leptophobic Z and technicolor, which predict the presence of a resonant enhancement near 150 GeV, are excluded.
We thank Adam Martin and Matthew Buckley for help with simulation of technicolor and Z models, respectively. We congratulate our colleagues in the CERN accelerator departments for the excellent performance of the LHC machine. We thank the technical and administrative staff at CERN and other CMS institutes, and acknowledge support from FMSR (Austria); FNRS and (b) Observed and expected 95% CL upper limits, with one-and two-sigma error bands, on the cross section divided by the expected values for various signal models. The limits are calculated using the CL S method. A value of the excluded cross section over the predicted cross section of less than one indicates that the model is excluded at 95% CL. Table 3 lists the cross sections for these models.   [5] CMS Collaboration, "Observation of a new boson at a mass of 125 GeV with the CMS experiment at the LHC", (2012). arXiv:1207.7235. Submitted to Phys. Lett. B.

References
[6] ATLAS Collaboration, "Observation of a new particle in the search for the Standard Model Higgs boson with the ATLAS detector at the LHC", (2012). arXiv:1207.7214. Submitted to Phys. Lett. B.