Search for t t ¯ resonances in the semileptonic final state using pp collisions at s √ = 8 TeV recorded with the CMS detector
SSearch for t ¯ t resonances in the semileptonic finalstate using pp collisions at √ s = 8 TeV recordedwith the CMS detector Anne-Laure Pequegnot
Institut de Physique Nucl´eaire de Lyon, FranceE-mail: [email protected]
Abstract.
Many extensions to the Standard Model predict gauge interactions with enhanced couplingsto third generation quarks, especially the extremely heavy top quark[1]. These phenomena canlead to resonances in the production of t ¯ t pairs. In this note, a model-independent search forthe production of heavy resonances decaying into top-antitop quark pairs in the semileptonicfinal state using pp collisions at √ s = 8 TeV recorded with the CMS detector[2] is presented.The analysis focuses on the semileptonic decay channels into electrons or muons and coversthe range of 0.5-3 TeV in the invariant mass of the t ¯ t pairs where the events present specificand different topologies. Therefore two dedicated analyses have been performed to achieve thebest sensitivity on the whole invariant mass range: the threshold analysis (0.5-2 TeV) which isoptimized for t ¯ t production at the kinematic production threshold, using standard top quarkreconstruction techniques, and the boosted analysis (1-3 TeV) optimized for t ¯ t produced withhigh Lorentz boosts, using dedicated techniques for boosted top quarks reconstruction.No evidence of signal has been found. Thus the following limits at 95% C.L. on theproduction of non-SM particles in specific models are set: topcolor Z (cid:48) bosons with a widthof 1.2 (10) % of its mass are excluded for masses below 2.1 TeV (up to 2.7 TeV). In addition,Kaluza-Klein excitations of a gluon with masses below 2.5 TeV in the Randall-Sundrum modelare excluded.
1. Event selection
Data samples corresponding to an integrated luminosity of 19.6 f b − of pp collisions at √ s = 8 TeV collected by the CMS experiment in 2012 are analyzed. The CMS detector, ageneral-purpose apparatus operating at the CERN LHC, is described in detail elsewhere [3].Concerning the threshold analysis (inclusive mass range of 0.5-2 TeV in m t ¯ t ), standardreconstruction techniques for top quarks produced with a small boost (all decay products areresolved) are used. The data analysed are recorded with triggers requiring one isolated leptonand 3 central jets. The following selection is applied: exactly one isolated lepton, at least 4jets with p T > / / / GeV , E missT >
20 GeV to reduce QCD background and at least oneb-tagged jet.Events are classified into 4 categories according to the flavour of the lepton and the numberof b-tagged jets (exactly one or two and more). a r X i v : . [ h e p - e x ] N ov or the boosted analysis (inclusive mass range of 1-3 TeV in m t ¯ t ), techniques dedicatedto boosted top quarks reconstruction are used. The data analysed are recorded with triggersrequiring one lepton and 2 central jets. For the selection, exactly one lepton without anyisolation specification and at least 2 jets with p T >
150 GeV for the leading jet and p T > E misst has to be greater than 50 GeV to reduce QCDbackground, and the scalar quantity T missT = E missT + p leptonT greater than 150 GeV. To reduceQCD background, especially in electron channel, several topological requirements are applied toensure the missing transverse momentum does not point along the transverse direction of theelectron or of the leading jet. Events are classified into 4 categories according to the flavour ofthe lepton and the number of b-tagged jets (exactly zero or one and more).
2. Event reconstruction
A first step consists in assigning the final-state objects in each event to either the leptonicor hadronic side of the t ¯ t decay. the charged lepton and E missT are assigned to the leptonic sideof the event, where E missT is assigned to the transverse component of the neutrinos momentum.After that, a χ sorting algorithm is used to assign the jets.In the threshold analysis, the χ is defined as follows: χ = χ m Lept + χ m Hadt + χ m Hadw + χ p t ¯ tT with χ x = ( x meas − x MC ) /σ MC (1) m Lept is the mass of the leptonic top quark ; m Hadt , the mass of the hadronic top quark ; m Hadw ,the mass of the hadronic W boson and p t ¯ tT , the transverse momentum of the t ¯ t system. Thechoice of the best jets combination corresponds to the one which gives the smallest χ value.In the boosted analysis, the χ is reduced at its two first terms. All mass hypotheses thathave exactly one jet assigned to the leptonic side, and at least one jet assigned to the hadronicside are then considered. The combination with the smallest value of χ is chosen for each event.This χ has to be smaller than 10 which rejects most of the W+jet background and maximizesthe sensitivity on the expected limits. Two particular models for the signal are considered: a generic spin 1 Z (cid:48) boson with a widthof 1.2 % (10 %) of the resonance mass [5]; and the Kaluza-Klein partner of the SM gluon [6].In the threshold analysis, the signal contribution is extracted from a maximum likelihoodfit to the data, using simultaneously the four categories described previously (see figure 1).The signal Probability Density Function in the fit is parametrized from the simulation, using asuperposition of Gaussian kernels to Tmodel the distribution. the background is estimated ondata using the following functional form: dσdm t ¯ t = (cid:16) − m √ s (cid:17) c (cid:16) m √ s (cid:17) c + c ln m √ s (2)The c , c , c parameters and the New Physics signal cross-section are floatting and areextracted directly from the fit on data.In the boosted analysis, the background is estimated from simulations (see figure 2). Toextract the number of signal events, a binned likelihood of the m t ¯ t distributions in the fourchannels is used. The number of events in bin i is assumed to follow a Poisson distribution with igure 1. Likelihood fit projection on data,including a signal with a mass hypothesisof 750 GeV, done simultaneously on the 4different categories for the threshold analysis.The expected shape of the Z (cid:48) signal isoverlaid, normalized to 1.0 pb.
Figure 2.
Comparison between dataand SM prediction for reconstructed m t ¯ t distributions for the boosted semileptonicanalysis with muon and ≥ (cid:48) samples.mean λ i , given by the sum over all considered background processes and the signal. The signalis scaled with a signal strength modifier µ , which corresponds to the signal cross section in pb: λ i = µS i + (cid:88) k B k,i (3)where k runs over all considered background processes, B k is the background template forbackground k, and S is the signal template, scaled according to luminosity and a signal crosssection of 1 pb.
3. Systematic uncertainties
The treatment of the systematic uncertainties varies between the threshold and boostedanalysis owing to the differences in the techniques. The table 1 summarizes the systematicuncertainties considered, specifying if they affect the signal, the background or both.
4. Results
The statistical treatment doesn’t indicate any excess of events above the expected yield ofthe SM processes, thus limits are set. A bayesian statistical method is used to extract the 95%C.L. upper limits. The systematic uncertainties taken into account as nuisance parameters,integrated with a log normal prior. The transition between the two analyses is based on theexpected limits sensitivity. An analysis for the all-hadronic channel has also been performed(only in the boosted regime), but is not described here (see [4] for more details). This analysisis combined with the semileptonic boosted one.All these combinations lead to the following limits on the production of non-SM particles inspecific models. Topcolor Z (cid:48) bosons with a width of 1.2 % (10%) of its mass are excluded at hreshold analysis Boosted analysisSystematic uncertainty signal background signal backgroundEvent pileup x x xLuminosity x x xLepton ID and trigger x x xJES and JER x x xSignal fit pdf xBackground fit pdf xBackground cross section xParton distribution functions negl. x xBackground modelling x
Table 1.
Summary of systematic uncertainties considered. A cross indicates when a certainsystematic uncertainty was applied, ”negl.” indicates that the systematic uncertainty was foundto be negligible.95% CL for masses below 2.1 TeV (2.7 TeV) (see figure 3 for the Z (cid:48) with narrow decay width).Kaluza-Klein excitations of a gluon with masses below 2.5 TeV in the Randall-Sundrum modelare excluded.
Figure 3.
Final result of the combination of the 3 analysis: the 95% CL upper limits onthe product of the production cross section σ Z (cid:48) and the branching fraction B of hypothesizedresonances that decay into t ¯ t as a function of the invariant mass of the resonance. The Z (cid:48) production with Γ Z (cid:48) /m Z (cid:48) = 1.2% compared to predictions based on Ref. [5]. The ± ± References [1] R. Frederix et F. Maltoni. Top pair invariant mass distribution : A Window on new physics. JHEP,0901:047,2009.[2] S. Chatrchyan et al. (CMS Collaboration) Searches for new physics using the t ¯ t invariant mass distribution inpp collisions at √ s = 8 TeV. Phys. Rev. Lett. 111 (2013) 211804 ; Erratum Phys. Rev. Lett. 112, 119903(2014)[3] CMS Collaboration. The CMS experiment at the CERN LHC. Journal of Instrumentation 3 no.08, (2008)S08004.[4] CMS Collaboration. Search for Anomalous Top Quark Pair Production in the Boosted All-Hadronic FinalState using pp Collisions at √ s = 8 TeV. CMS-PAS-B2G-12-005, (2013).[5] R. M. Harris et S. Jain. Cross Sections for Leptophobic Topcolor Z (cid:48) Decaying to tt
Final result of the combination of the 3 analysis: the 95% CL upper limits onthe product of the production cross section σ Z (cid:48) and the branching fraction B of hypothesizedresonances that decay into t ¯ t as a function of the invariant mass of the resonance. The Z (cid:48) production with Γ Z (cid:48) /m Z (cid:48) = 1.2% compared to predictions based on Ref. [5]. The ± ± References [1] R. Frederix et F. Maltoni. Top pair invariant mass distribution : A Window on new physics. JHEP,0901:047,2009.[2] S. Chatrchyan et al. (CMS Collaboration) Searches for new physics using the t ¯ t invariant mass distribution inpp collisions at √ s = 8 TeV. Phys. Rev. Lett. 111 (2013) 211804 ; Erratum Phys. Rev. Lett. 112, 119903(2014)[3] CMS Collaboration. The CMS experiment at the CERN LHC. Journal of Instrumentation 3 no.08, (2008)S08004.[4] CMS Collaboration. Search for Anomalous Top Quark Pair Production in the Boosted All-Hadronic FinalState using pp Collisions at √ s = 8 TeV. CMS-PAS-B2G-12-005, (2013).[5] R. M. Harris et S. Jain. Cross Sections for Leptophobic Topcolor Z (cid:48) Decaying to tt ¯ tt