Search for heavy resonances, and resonant diboson production with the ATLAS detector

Heavy resonances decaying into a pair of fundamental partic les such asjj, ll, γγ, andlν, are among the most common features to search for phenomena beyon d the standard model (SM). Electroweak boson pair production, such as W W or ZZ with subsequent decays to lνl′ν′ and lljj respectively, is a powerful test of the spontaneously broken gauge symmetry of the SM and can be also used to search for phenomena beyond the SM. There is a wide spectrum of theoreti cal models predicting these kinds of resonant signatures. This note covers several searches for these new phenomena conducted within ATLAS in 2011 and 2012 for the LHC 7 and 8 TeV center of mass energies respective ly. No significant deviations from the SM have been observed and therefore, limits are set on the chara cteristic parameters of several new physics models. These benchmark models include new heavy Z′/W ′ gauge bosons, chiral excitation of the SM weak gauge bosons,Z∗/W ∗ Randal-Sundrum and ADD gravitons, Composite models for qua ks, e.g.q∗ with substructure scaleΛ, Quantum black holes, TeV −1 Kaluza-Klein excitation ofγ/Z and more.


Models Publication
RS, ADD [5] This note is divided into two parts, one dealing with simple two-body resonances and the second dealing with diboson resonances. The separation within each part is signaturewise where for several signatures, there is an overlap with respect to the models interpretation of the limit on the models parameters.    are selected, where the backgrounds are γ/Z (estimated at NNLO in this analysis), dibosons, tt, multijet and W +jets. All backgrounds but the W +jets are estimated from Monte Carlo (MC) while the W +jets (used in the ee channel only) is estimated from the data. The γ/Z contribution can be treated as a part of the signal as it may interfere with the new physics part, as in the KK case, or as a part of the backgrounds if it does not, as in the RS graviton case. However, in cases where the interference with the new physics amplitude is very small, as in the Z ′ case, it can be safely neglected and thus the γ/Z contributions can be treated as a part of the backgrounds. The sum of backgrounds is normalized to the Z peak, in the range 70-110 GeV, to cancel out mass-independent uncertainties. The remaining dominant uncertainties are 20% from theory and 21% from the ee background estimation method. The data are found to be consistent with the backgroundonly hypothesis with p-values 8.6% (ee) and 69% (µµ). The benchmark model for this analysis is the sequential SM Z ′ . In the 8 TeV analysis [2] it was the only model considered along with several E 6 -inspired Z ′ s, where the limits on these are always lower than for the Z ′ SSM . In figures 1 and 2, two of the combined (ee and µµ channels) mass limits derived from the dilepton invariant mass distributions are presented. In figure 2, the limit on M KK , the mass of the first KK excitation of the γ/Z particles, is the highest to date -almost 1 TeV higher than previous limits obtained from indirect measurements. It is also the highest direct limit set on a resonance mass in ATLAS. Table 2 lists several representative limits on the parameters of various models, obtained from the two Table 2. A collection of representative limits on the parameters of various models, as derived from the dilepton mass distributions. The limit in the first row (Z ′ SSM ) was calculated also for 8 TeV [2], where the number in parentheses corresponds to 7 TeV [3]. The rest of the results are for 7 TeV only [3].
dilepton analyses. The dilepton signature can be utilized to search also for non-resonant signatures predicted by the set of CI models [4].

Diphoton
In this analysis [5], the γγ results are combined with the ℓℓ results [3] and [4] (see section 2.1) for the applicable models (7 TeV only). This is another clear signature where one looks on the invariant mass of the two highest-E T (>25 GeV) isolated photons in an event. The benchmark model in this analysis is the RS graviton. The backgrounds for this analysis are divided into two: (a) irreducible: SM γγ (estimated at NLO from MC), and (b) reducible: γj, jγ and jj with one or two jets faking photons (estimated from the data). The first two options differ by the identity of the leading-p T object -the photon or the jet (fake photon). The sum of backgrounds is normalized in the control region (142 < m γγ < 409 GeV) to cancel out mass-independent uncertainties. The reducible background is extrapolated to high masses using a smooth function f (m γγ ) = p 1 · m p2+p3 log mγγ γγ . The dominant uncertainties in this analysis are: 9% (photon identification and isolation), 5-15% (mostly due to parton distribution functions (PDFs)). The data are found to be consistent with background-only hypothesis with p-value 86%. Figure 3 shows the limit on the RS graviton mass for few values of the dimensionless couplings, κ/M Pl . Combining the result with dileptons yields a limit of 1.03 and 2.23 TeV on the RS graviton mass, for κ/M Pl = 0.01 and 0.1 respectively. The diphoton channel can be utilized to search also for non-resonant signatures such as those predicted by the ADD model for example. In that case, the limit on the number of ADD signal events in the search region at m γγ > 1217 GeV is 7.21 and the lower limits on the ultraviolet cutoff of the KK spectrum, M S , range between 2.79 and 4.18 TeV depending on the number of EDs and the theoretical formalism (see [4]). Mass [GeV] 2000 3000 4000 5000 [pb] xA × σ

Dijet
In this analysis [6-8], the mass and angular distribution of the two most high-p T jets in an event are used. The benchmark model used is the excited quark, q * . The analysis takes advantage of the fact that the dominant t-channel QCD interactions lead to angular distribution that peak at small scattering angles while the signal is expected to be more isotropic. Therefore, the angular distributions are used (in addition to the dijet mass distribution). A data driven background estimation leading to the m jj spectrum is parametrized by the function where, x ≡ m jj / √ s, while the background estimate for the angular analyses relies on QCD MC. The dominant uncertainties for the mass analysis are: 4% for p jet T > 1 TeV from the jet energy scale (JES) and 3.6-3.9% from the luminosity uncertainties. The dominant uncertainties for the angular analysis are: <8% due to NLO QCD renormalization and factorization scales, and <15% due to JES. The angular analysis (7 TeV) employs ratio observables called χ and F χ (see [7] for definition) to reduce its sensitivity to systematic uncertainties (JES, PDFs, luminosity) and is more sensitive to non-resonant signals than the mass analysis. The data are found to be consistent with background-only hypothesis with p-value 61%. Figures 4 and 5 show the limit on the cross section times acceptance vs. the resonance mass for q * and for a generic Gaussian model (see [7,8] on how to interpret the Gaussian limits). Table 3 lists several representative  limits obtained from both the mass and the angular distribution analyses.

Charged lepton and a neutrino
This analysis [9] requires exactly one isolated, high-p T muon or electron with p T >25 and 85 GeV respectively. A missing transverse energy (E miss T ) with same thresholds is also required. The kinematic variable used to identify a W ′ /W * signal is the transverse mass, . These two signals are distinguishable with respect to both the m T and the angular distributions. The dominant backgrounds are: SM W bosons (estimated at NNLO), Z → ℓℓ with one lepton not reconstructed, τ 's from W/Z and diboson production, tt and single-top production and QCD where a hadron decays semileptonically or a jet is misidentified as an electron. The dominant uncertainties are: 12% (cross sections) and <5% (experiment). The data are found to be consistent with the background-only hypothesis. Figure 6 shows the limit on the cross section times branching fraction vs. M W ′ where the corresponding scenario for the W * model is comparable. For the W ′ theory curve in figure 6, no interference with SM W is taken. This is again a sequential SM scenario. For the W * limit, EPJ Web of Conferences the W * is taken with q and g coupling strengths normalized to reproduce the W ′ width. The limit on M W ′ (M W * ) is 2.55 (2.42) TeV.

WW resonance (ℓνℓ ′ ν ′ )
In this analysis [10], the signature is less pronounced than for the simple two-body resonances because of the subsequent decay of the W bosons produced in the hard interaction. More over, the presence of two W particles implies a large and non-resolved E miss T . In this analysis, one requires exactly two oppositely-charged isolated, high-p T (>25 GeV) leptons, and a large E miss T (>30, 60 and 65 GeV for eµ, ee and µµ). In order to reduce Z and top backgrounds, it is required that m ℓℓ > 106 GeV and events with b-jets are discarded (the tagging efficiency of b-jets is ǫ tag b ∼ 85%). The W W transverse mass, m W W as a discriminant variable. The RS graviton is again the benchmark model. However, since this analysis concentrates on the possibility of decay into a pair of heavy W particles, the sensitivity for another theoretical variation of the ordinary RS model is greater. In the "bulk" RS graviton, G * bulk model, where the ED setup is slightly different than for the ordinary RS model, the graviton has enhanced couplings to the heavier particles, leading to large branching fractions for these states, e.g.
The dominant backgrounds are: SM W W (estimated at NNLO) and W Z/ZZ with only two reconstructed leptons, W γ where the γ is reconstructed as a lepton, tt (estimated at NLO, with zero b-jets) and single-top (zero b-jets), W/Z+jets and QCD multi-jet production (both estimated from data). The main uncertainties are: <5% (due to muon resolution correction), 2-9% (JES), 3.5% (E miss T energy scale), 6-21% (ǫ tag b estimation) and 5-10% (cross sections) and 10-30% (W +jets estimation). The data are found to be consistent with background-only hypothesis with p-value >8%. Figure 7 shows the limit on the cross section times branching fraction vs. M G * bulk . The limits on the ordinary RS and the "bulk" RS graviton mass are 1.23 and 0.84 TeV respectively.

ZZ resonance (ℓℓjj)
In this analysis [11], the signature is again less pronounced than for the simple two-body resonances because of the subsequent hadronic decay of one of the produced Z bosons. One requires exactly two isolated, high-p T , same flavor leptons and two high-p T jets (not within ∆R = 0.3 around a lepton). The presence of leptons reduce the multijet background with respect to the fully-hadronic final state and allow a complete kinematic reconstruction of the intermediate states. The "bulk" RS graviton is the only signal considered in this analysis. For highly boosted Z bosons, the η − φ distance between the two quarks can be parametrized as R qq ≈ 2M Z /p Z T . Therefore, in resonances with masses above ∼ 900 GeV, the two quarks can fall within a ∆R = ∆φ 2 + ∆η 2 = 0.4 cone, resulting in a single reconstructed massive jet where m jj (m j ) ≃ M Z . As a result, two signal selections must be defined, (a) resolved and (b) merged. The background modeling is tested in these two selections for the m ℓℓjj (m ℓℓj ) distributions respectively. Muons are required to be oppositelycharged and the ℓℓ invariant mass is required to be in the range 66<m ℓℓ <116 GeV in order to ensure a Z origin. The dominant backgrounds are: SM Z+jets (estimated at NLO), tt, SM diboson production, W +jets and QCD (estimated from data). The final background is estimated by fitting the m ℓℓjj (m ℓℓj ) distributions in the data to a smooth function f (x) = p 1 (1 − x) p2 x −p3−p4 ln(x) where x ≡ m ℓℓjj / √ s or x ≡ m ℓℓj / √ s for the two selections respectively. The main uncertainties are: 5% for m ℓℓjj <800 GeV and 10-40% for m ℓℓj (due to the background fit), 11-15% (overall uncertainty on the signal acceptance due to jet mass scale, luminosity, JES and ISR/FSR modeling). The data are found to be consistent with backgroundonly hypothesis. Figure 8 shows the limit on the cross section times branching fraction vs. M G * bulk where the two signal selections are combined by showing only limits for one of them in the range where the expected limit is better. The limit on the "bulk" RS graviton mass is 0.84 TeV.

Summary
This note covers the most recent ATLAS searches for heavy resonances in 8 different analyses and for 6 different signatures. The most massive observed event in the data,