ATL

AS-

CO

NF-

2016

-032

16Ju

ne20

16

ATLAS NOTEATLAS-CONF-2016-032

June 16, 2016

Search for new phenomena using events with b-jets and a pair ofsame-charge leptons in 3.2 fb−1 of pp collisions at

√s = 13 TeV

with the ATLAS detector

The ATLAS Collaboration

Abstract

An analysis is presented of events containing jets including at least one b-tagged jet, siz-able missing transverse momentum, and at least two charged leptons including a pair ofthe same electric charge, with the scalar sum of the jet and lepton transverse momenta be-ing large. Standard Model processes rarely produce these final states, but several modelsof physics beyond the Standard Model predict an enhanced production rate of such events.Specific models with this feature are considered here: vector-like T , B, and T5/3 quark pairproduction, and four-top-quark production under three scenarios (Standard Model, contactinteraction, and extra dimensions). A data sample of 3.2 fb−1of pp collisions at a center-of-mass energy of

√s = 13 TeV recorded by the ATLAS detector at the Large Hadron Collider

is used in this analysis. Several signal regions are defined, in which the consistency betweenthe data yield and the background-only hypothesis is checked, and 95% confidence levellimits are set on various signal models.

c© 2016 CERN for the benefit of the ATLAS Collaboration.Reproduction of this article or parts of it is allowed as specified in the CC-BY-4.0 license.

1 Introduction

The Standard Model (SM) has been repeatedly confirmed experimentally. Nonetheless there is a strongmotivation for physics beyond the SM (BSM) at about the TeV scale, with additional features that, forexample, specify the nature of dark matter and provide a mechanism to naturally stabilize the Higgsboson mass at its observed value of approximately 125 GeV [1, 2]. This paper reports on a search forBSM processes resulting in pairs of isolated high-transverse-momentum (high-pT) leptons1 with the sameelectric charge, hereafter denoted as same-sign leptons, missing transverse momentum, and b-jets. Thisis a promising search channel since the SM yields of such events are small and several types of BSMphysics may contribute.

Among the models that predict enhanced same-sign lepton production are those that postulate the exis-tence of vector-like quarks (VLQ) and those that predict additional production modes for the four-top-quark final state. A common data sample is used to search for each of these signatures.

Several extensions to the SM that regulate the Higgs boson mass in a natural way require the existenceof vector-like quarks [3–21], where ‘vector-like’ means that the left- and right-handed components trans-form identically under the SU(2)L weak isospin gauge symmetry. Since quarks with this structure do notreceive their mass from the Higgs mechanism, their existence would not enhance the Higgs boson pro-duction cross section and thus the motivation persists for a direct search [22]. There are several possiblevarieties of VLQs. Those having the same electric charge as the SM bottom and top quarks are denotedB and T . In addition, the exotic charge states T5/3 and B−4/3 may occur, where the subscripts indicate theelectric charge. Vector-like quarks may exist as isospin singlets, doublets, or triplets. Arguments basedon naturalness suggest that VLQs may not interact strongly with light SM quarks [23, 24]. Thus it is as-sumed for this analysis that VLQs decay predominantly to third-generation SM quarks. For the B and Tquarks, charged- and neutral-current decays may both occur (B → Wt, Zb, or Hb; T → W b, Zt, or Ht),providing many paths for same-sign lepton production for events with BB̄ or TT̄ pairs.

The branching fractions to each allowed final state are model-dependent, and the ones occurring in modelswhere the B and T exist as singlets [25] are used as a reference. For a B (T) quark with a mass of 0.75 TeV,these branching ratios are Wt : Zb : Hb = 46% : 28% : 26% (W b : Zt : Ht = 49% : 21% : 29%). Sincethe pair production of heavy quarks is mediated by the strong interaction, the cross section is identicalfor different types of vector-like quarks of a given mass. The next-to-next-to-leading-order (NNLO)cross sections from top++ v2.0 [26, 27] are used in this paper. The T5/3 quark must decay to W+t, andtherefore both single and pair production of this quark can result in same-sign lepton pairs; currently onlypair production is considered in this analysis.

Same-sign lepton pairs may also arise from the production of four top quarks (tt̄tt̄). The SM rate forthis production is small (≈ 9 fb [28, 29]), but there are several BSM scenarios that can enhance the rate,such as top compositeness models [30–32] or Randall-Sundrum models with SM fields in the bulk [33].For a certain range of model parameters, these can be described by an effective four-fermion contactinteraction with coupling strength C4t/Λ

2, where C4t is the coupling constant and Λ is the scale of theBSM physics [32]. The Lagrangian for this interaction is

L4t =C4t

Λ2

(t̄Rγ

µtR) (

t̄RγµtR)

(1)

1 “Lepton” means “charged lepton” is this paper. Only electrons and muons are considered in the search. Tau leptons are notexplicitly reconstructed, but electrons and muons from τ decay may enter the selected samples.

2

T

T̄W−, H, Z

b̄, t̄, t̄

t

H

g

g

(a)

g

g

t

t

t̄

t̄

t̄

t

(b)

Figure 1: Leading-order diagrams for (a) vector-like top quark pair production, and (b) four-top production fromthe contact interaction model.

where tR is the right handed top spinor and the γµ are the Dirac matrices. Direct constraints limit anycontact interaction between left-handed top quarks to be too small to be to be observed at the LHC. Amodel with two universal extra dimensions under the real projective plane geometry (2UED/RPP) [34] isalso considered. In this model, the compactification of the extra dimensions leads to discretization of themomenta along their directions. The model is parameterized by the radii R4 and R5 of the extra dimen-sions or, equivalently, by mKK = 1/R4 and ξ = R4/R5. This model predicts the pair production of tier2(1,1) Kaluza–Klein (KK) excitations of the photon (A(1,1)

µ ) with a leading-order mass of√

1 + ξ2 mKKthat decay to tt̄ with an unknown branching fraction, assumed here to be 100%.

Leading-order Feynman diagrams for the production in pp collisions of some of the signals searched forin this analysis are presented in figure 1.

Previous searches by the ATLAS collaboration using an integrated luminosity of 20.3 fb−1 of pp collisionsat a centre-of-mass energy

√s = 8 TeV [35], and the CMS collaboration using an integrated luminosity

19.5 fb−1 of pp collisions at√

s = 8 TeV [36] and 2.3 fb−1 at 13 TeV [37], did not observe a significantexcess of same-sign dilepton production. However, in the ATLAS search, a modest excess was observed,reaching 2.5 standard deviations in the set of signal regions defined for searching for four-top-quarkproduction. The ATLAS result was used to set limits at 95% confidence level on various models, includingon VLQ and four-top-quark production. The CMS result was also used to set limits on various models,including on SM four-top-quark production. The upper limit on the four-top-quark production crosssection, set by CMS, was 49 fb. In separate analyses the CMS collaboration used the same-sign leptonsignature as part of a search for T5/3 quarks [38], ruling out left-handed (right-handed) T5/3 quarks withmass below 0.94 (0.96) TeV, and as part of a broader search for vector-like T quarks [39], ruling out suchquarks with mass less than 0.69 TeV. Other searches by the ATLAS collaboration using pp collisions at√

s = 8 and 13 TeV [40, 41] with similar final states to those reported here were interpreted in the contextof supersymmetric models. The present analysis uses an analysis strategy that is similar to the

√s = 8

TeV ATLAS analysis, using a data set recorded at√

s = 13 TeV with an integrated luminosity of 3.2fb−1. This allows for a check of the modest excess observed at 8 TeV. Because of the increased crosssections, the sensitivity of the search was improved from the 8 TeV search, despite the smaller integratedluminosity of the data set.

2 A tier of the Kaluza–Klein towers is labeled by two integers, corresponding to the two extra dimensions.

3

2 Data and Monte Carlo simulations

The data were recorded by the ATLAS detector [42] in LHC pp collisions at√

s = 13 TeV betweenJune and November 2015, corresponding to an integrated luminosity of 3.2 fb−1. The uncertainty on theintegrated luminosity is ±2.1%. It is derived, following a methodology similar to that detailed in ref. [43],from a calibration of the luminosity scale using x-y beam-separation scans performed in August 2015.ATLAS is a general-purpose particle detector used to investigate a broad range of physics processes. Itincludes three major detector systems: inner tracking devices surrounded by a superconducting solenoid,electromagnetic and hadronic calorimeters, and a muon spectrometer with a toroidal magnetic field. Theinner detector (ID) consists of a high-granularity silicon pixel detector, including the newly-installedInsertable B-layer (IBL) [44], a silicon strip detector, and a straw tube tracker; it is situated inside a2-Tesla field from the solenoid and provides precision tracking of charged particles with pseudorapidity|η | < 2.5, where the pseudorapidity is defined in terms of the polar angle3 θ as η = − ln tan(θ/2).The straw tube detector also provides transition radiation measurements for electron identification. Thecalorimeter system covers the pseudorapidity range |η | < 4.9. It is composed of sampling calorimeterswith either liquid argon (LAr) or scintillator tiles as the active medium. The muon spectrometer (MS)provides muon identification and measurement for |η | < 2.7. The ATLAS detector has a two-level triggersystem to select events for offline analysis.

Signal and the background sources that contain prompt same-sign leptons are modelled using Monte Carlo(MC) simulations. The remaining background sources are determined from data, as described in section 4.B, T , and T5/3 pair production is modelled using the Protos v2.2 [25] generator using the NNPDF v2.3LO [45] parton distribution functions (PDFs), with Pythia v8.1 [46] used to model extra gluon emissionand hadronization. Production of four top quarks is modelled under three scenarios: i) Standard Model,ii) contact interaction, and iii) 2UED/RPP. All three models are generated with MadGraph_aMC@NLOv5.2 [47] (where the leading-order calculation was used) using the NNPDF v2.3 PDFs followed by Pythiav8.1; in the case of 2UED/RPP the BRIDGE generator [48] is used to decay the pair-produced excitationsfrom MadGraph_aMC@NLO to tt̄. The simulated 2UED/RPP samples correspond to the tier (1,1) forthe symmetric (R4 = R5) case, with mKK ranging from 1000 to 1800 GeV. Constraints on the asymmetric(R4 > R5) case are derived by an extrapolation that uses kinematical considerations [49].

The background contributions from tt̄W and tt̄ Z (abbreviated as tt̄W/Z hereafter), tt̄W+W−, and three-topproduction are modelled with MadGraph_aMC@NLO v5.2 followed by Pythia v8.1, while W Z and Z Zplus jet production and W±W± j j production are modelled using Sherpa v2.1 [50]. The tt̄H backgroundis modelled with MadGraph_aMC@NLO v5.2 followed by Herwig++ v2.7 [51]. Background fromthe production of three vector bosons is modelled using Sherpa v2.1 and Pythia v8.1, and backgroundsfrom W H and Z H production are modelled using Pythia v8.1. The NNPDF v2.3 PDFs are used for thett̄W/Z , tt̄W+W−, three-top, four-top, W H and Z H samples, and the CT10 [52] PDFs are used for theW Z , Z Z , W±W± j j, three-vector-boson, and tt̄H samples. For the tt̄H samples, the CT10 PDFs are usedfor the matrix element calculation, and the CTEQ6L1 [53] PDFs are used for showering and multi-partoninteractions. For the major sources of background, such as tt̄W/Z production, the cross sections are scaledto match next-to-leading-order calculations [47].

A variable number of additional pp interactions are overlaid on simulated events to model the effect ofmultiple collisions during a single bunch crossing, and also the effect of the detector response to collisions

3 ATLAS uses a right-handed coordinate system with its origin at the nominal interaction point (IP) in the centre of the detectorand the z-axis along the beam axis. The x-axis points from the IP to the centre of the LHC ring, and the y-axis points upward.Cylindrical coordinates (r, φ) are used in the transverse plane, φ being the azimuthal angle around the beam axis.

4

from bunch crossings before or after the one containing the hard interaction. Events are then weighted toreproduce the distribution of the number of collisions per bunch crossing observed in data. The detectorresponse is modelled using a Geant4 [54, 55] simulation of the entire detector. The simulated events werereconstructed using the same algorithms that were applied to the collision data.

3 Event selection

The final states considered in this search are defined by the presence of two leptons with the same electriccharge, or three or more leptons of any charge. In addition, sizeable missing transverse momentum Emiss

Tand two or more jets are required, where at least one of the jets is consistent with hadronization of ab-quark. The criteria used for each of these objects are given below.

Each event must pass either an electron trigger where the chosen triggers require an electron with pT >

24 GeV or a muon trigger where the triggers chosen require either an isolated muon with pT > 20 GeV ora muon with pT > 50 GeV with no isolation requirement. The trigger efficiency for electrons is ∼ 95%while for muons it is ∼ 75%, resulting in trigger efficiencies that range from ≈ 95% for events with twomuons to > 99% for events with two electrons. In addition, events must have at least one reconstructedvertex, which is formed from at least two tracks with pT > 0.4 GeV. If multiple vertices are reconstructed,the vertex with the largest sum of the squared transverse momenta of its associated tracks is taken as theprimary vertex. For the final states considered in this analysis, the vertex reconstruction and selectionefficiency is expected to be close to 100%.

Electrons are identified by requiring a track to match an electromagnetic calorimeter energy cluster, sub-ject to several criteria on the shape of the shower, the quality of the track, and the consistency between theshower and the track. Electrons selected for this analysis must satisfy the tight requirement using a like-lihood discriminant that combines many individual variables [56]. The track associated with an electronmust have a transverse impact parameter with respect to the beamline d0 that satisfies |d0 |/σ(d0) < 5,where σ(d0) is the uncertainty on d0. The difference between the longitudinal position of the track alongthe beam line at the point where d0 is measured and the longitudinal position of the primary vertex, ∆z0,must satisfy |∆z0 sin θ | < 0.5 mm. The electron is required to have transverse momentum pT > 25 GeV,and the calorimeter energy cluster associated with the electron must have |η | < 2.47, with the bar-rel/endcap transition region 1.37 < |η | < 1.52 excluded. The electron candidate must be isolated fromadditional tracks within a cone of variable ∆R ≡

√(∆η)2 + (∆φ)2 = min(0.2,10 GeV/pT) [57], such that

the sum of the transverse momenta of the tracks within that cone must be less than 6% of the electron pT.The sum of energy of the calorimeter clusters within a cone of ∆R = 0.2 around the electron is requiredto be less than 6% of the electron pT . In addition, electrons must not share any tracks with a muon, andmust be separated from any jet by at least ∆R ≡

√(∆y)2 + (∆φ)2 = 0.4.4

Muons are identified from hits in the muon system matched to a central track. This analysis uses muonspassing the medium set of quality requirements on the muon and its associated track [58], with pT >

25 GeV and |η | < 2.5. The central track associated with the muon must satisfy |d0 |/σ(d0) < 3. The muoncandidate must be isolated from additional tracks within a cone of variable ∆R = min(0.3,10 GeV/pT).Muons are also required to be separated from any jet with more than two tracks by ∆R = max(0.4,0.04 +

10 GeV/pT). At least one of the selected leptons must match a lepton identified by the trigger.

4 ∆R is calculated from rapidity, y, instead of pseudorapidity, η. Rapidity is defined as: y ≡ (1/2) ln((E + pz )/(E − pz )). Inthe limit where the particle travels close to the speed of light, pseudorapidity converges to rapidity.

5

Electrons Muons JetsTrigger 1 electron, pT > 24 GeV 1 isolated muon, pT > 20 GeV

or 1 muon, pT > 50 GeVpT > 25 GeV > 25 GeV > 25 GeV|η | < 1.37 or 1.52 < |η | < 2.47 < 2.5 < 2.5Object ID tight medium –Vertex match |d0 |/σ(d0) < 5 |d0 |/σ(d0) < 3 JVT requirement

|∆z0 sin θ | < 0.5 mm |∆z0 sin θ | < 0.5 mm (if |η | < 2.4 andpT < 50 GeV)

Isolation track and calorimeter track –Multiplicity 2 same-charge leptons or ≥ 3 leptons –

Table 1: Summary of the ‘preselection’ criteria.

Jets are reconstructed from energy clusters in the calorimeter using the anti-kt algorithm [59–61] withradius parameter 0.4. Jets are removed if they are within ∆R = 0.2 of an electron or have fewer thanthree tracks and are within ∆R = 0.4 of a muon. To suppress jets that do not originate from the primaryvertex in the event, the jet vertex tagger (JVT) is defined, combining two variables that separate hardscattering jets from pileup jets [62]. Jets with pT < 50 GeV and |η | < 2.4 are selected based on their JVTvalue, keeping ≈ 92% of hard-scattering jets and rejecting ≈ 96% of jets not originating from the primaryvertex. All jets in this analysis are required to have pT > 25 GeV after energy calibration [63, 64] and|η | < 2.5.

Identification of jets containing b-hadrons is performed with a multivariate algorithm based on the prop-erties of the tracks associated with the jet. A requirement is placed on the output of the algorithm suchthat ≈ 77% of b-quark jets, ≈ 22% of c-quark jets, and ≈ 1% of light-quark or gluon jets pass in inclusivesimulated tt̄ events [65, 66]. All jets that meet this criterion are called ‘b-tagged’ jets.

The missing transverse momentum (EmissT ) is calculated as the negative of the vector sum of the trans-

verse momenta of all identified physics objects (electrons, photons, muons, jets), and an additional softterm constructed from tracks associated with the primary vertex but not associated with any physics ob-jects [67].

If the same-sign leptons are both electrons, their invariant mass mee is required to be greater than 15 GeVand to satisfy |mee −mZ (= 91 GeV) | > 10 GeV. These criteria reject events from ψ, Υ and Z resonancesdecaying to an electron–positron pair where the charge of either the electron or positron is misidentified.

‘Preselection’ is defined as the set of all of the above criteria, as summarized in table 1.

4 Background estimation

Background arises from two distinct sources: SM processes that result in same-sign lepton pairs andinstrumental backgrounds where objects are misidentified or misreconstructed such that events appear tohave the required set of leptons. The former category includes production of tt̄W , tt̄ Z , tt̄W+W−, W Z ,Z Z , W±W± j j, tt̄H , W H , Z H , three vector bosons, or three top quarks. In addition, the four-top-quarkproduction predicted in the SM is included as a background to all searches for other signals. All of

6

these processes have small cross sections and their expected yields are predicted using MC simulation.Differences between the lepton selection and b-tag efficiencies between the MC simulation and data areestimated using dedicated control regions and correction factors for MC are derived. These correctionsare propagated to the simulated background yield predictions. The corrections are also applied to thesignal MC used to estimate the signal selection efficiency.

Instrumental backgrounds have contributions from two categories: i) events where one or more jets aremisidentified as leptons (‘fake’ leptons), or which contain non-prompt leptons, and ii) events that containtwo leptons of opposite charge, where one of the charges is mismeasured. The ‘matrix method’ [68] isused to estimate the contribution from events with fake or non-prompt leptons. In this method, the default(‘tight’) lepton identification criteria (section 3) are relaxed to form a ‘loose’ sample. Lepton isolationrequirements are not imposed for the loose sample for both electrons and muons. Their respective identi-fication quality, tight likelihood for electrons and medium for muons, remain the same. The loose sampleis thus a superset of the tight sample. The fraction of real leptons (meaning prompt leptons from thedecay of a W , Z , or H boson) passing the loose criteria that also pass the tight criteria is referred to as r .Similarly, the fraction of fake/non-prompt leptons passing the loose requirements that also pass the tightrequirements is referred to as f . Using measured values of r and f , the observed yields of dilepton eventsin the categories loose–loose, loose–tight, and tight–tight can be related to the real–real, real–fake, andfake–fake yields. An analogous procedure is applied to three lepton events starting with categories forall possible combinations of three loose or tight leptons, resulting in an estimate of the number of eventswith one or more misidentified leptons in the selected sample.

Single-lepton events are used to measure r and f . Here, for both the ‘loose’ and ‘tight’ leptons, theselection cuts on the impact parameter are omitted, so that fake/non-prompt enriched regions can bedefined using cuts on the impact parameter. The criteria used to select the single lepton events are differentfor electrons and muons due to the differences in the sources of fake/non-prompt leptons for each flavour.For electrons, r is measured using events with Emiss

T > 150 GeV, where the dominant contribution isfrom W → eν, and f is measured using events with the transverse mass of the Emiss

T -lepton system5

mT(W ) < 20 GeV and EmissT + mT(W ) < 60 GeV, where the dominant contribution is from multijet

production where one or more jets are misidentified as an electron. For muons, r is measured usingevents with mT(W ) > 100 GeV, a sample dominated by W → µν, and f is measured using events wherethe impact parameter of the muon with respect to the primary vertex is more than five standard deviationsfrom zero, consistent with muons arising from heavy-flavour hadron decays. The small contribution ofreal leptons to the control samples used to measure f is estimated from simulation and this contributionis subtracted from the sample. The values of r and f are parameterized with respect to properties of theleptons (e.g. |η | and pT) and of the event (e.g. the number of b-tagged jets). Typical values are r = 0.80– 0.90 and f = 0.15 – 0.50 for electrons and r = 0.80 – 1.00 and f = 0.10 – 0.40 for muons.

The triggers used for low-pT muons require isolation; since the tight and loose muon criteria differ intheir isolation requirements, fake/non-prompt muons in events where only the low-pT muon trigger firedare more isolated, on average, than those from an unbiased trigger, meaning that f for these muons issubstantially higher. Therefore, muon r and f are measured separately for samples collected with thedifferent triggers and the appropriate values are applied based on the muon triggers that fired in eachevent. To properly estimate the statistical uncertainty on the fake/non-prompt lepton contribution given

5 The transverse mass of the EmissT -lepton system is defined as mT(W ) ≡

√2pT`Emiss

T (1 − cos∆φ) where pT` is the lepton pTand ∆φ is the azimuthal angle between the lepton and the direction of the missing transverse momentum vector.

7

the small number of events, a Poisson likelihood for the estimate from the matrix method is used. Thisapproach also eliminates the possibility of a negative yield estimate.

Charge misidentification (‘Charge mis-ID’) is negligible for muons due to the small probability for muonsto radiate photons, the long lever arm to the muon system, and the fact that the charge is measured inboth the inner detector and the muon spectrometer. For electrons, the rate of charge misidentification iscalculated from a sample of Z → ee events selected using |mee − mZ | < 10 GeV, with no requirementplaced on the charge of the two electron tracks. It is assumed that the rate at which the charge of anelectron is misidentified varies with the |η | and pT of each electron but is uncorrelated between the twoelectrons in each event. Further assuming that the sample consists entirely of true opposite-sign electronpairs, the number of measured same-sign events N i j

ss where one electron is in the ith (|η |,pT) bin and theother in the j th bin is expected to be

N i jss = N i j (εi (1 − ε j ) + ε j (1 − εi )) (2)

where N i j is the total number of events in the i- j (|η |,pT) bin, and ε is the rate of charge mismeasurement.The value of ε in each (|η |,pT) bin is then extracted by maximizing the Poisson likelihood for the observednumber of same-sign pairs in each (|η |,pT) bin to be consistent with the expectation from equation 2.

To determine the number of events expected from charge mismeasurement in the signal region, a sam-ple is selected using the same criteria as for the analysis selection, except that an opposite-sign ratherthan same-sign ee or eµ pair is required. Then, for each event in this sample, the measured ε valuesof each electron are used to calculate an event weight, the sum of which determines the expected num-ber of mismeasured same-sign events in the analysis sample. One source of charge mismeasurement isfrom ‘trident’ electrons, where the electron emits a hard photon that subsequently produces an electron–positron pair, resulting in three tracks with small spatial separation. If the wrong track is matched to theEM cluster, the charge may be incorrect. However, photon conversion is also a source for misidentifiedelectrons and therefore events with trident electrons also contribute to the fake/non-prompt lepton back-ground. To avoid double-counting events with trident electrons in the background estimate, the chargemismeasurement rate is measured in a data sample where the non-prompt/fake contribution, estimatedusing the matrix method, has been removed.

Simulation was used to estimate the sources of events in the signal regions that have fake/non-promptleptons and/or electrons with mismeasured charge and it is found that tt̄ events provide the dominantcontribution.

The background estimates are validated using samples where one or more of the signal selection criteria,described in section 5, are inverted so that the samples are statistically independent and the expected yieldfrom signal events is small. One of these validation regions, called the ‘low HT+1b’ region, is defined byapplying the preselection, and adding the requirements that the scalar sum of all jet and lepton transversemomenta (HT) satisfies 100 GeV < HT < 400 GeV, and that there be at least one jet in the event.This validation region is particularly useful because the background composition is similar to that of thesignal selection, including the fact that tt̄ events are the dominant source of the fake/non-prompt leptonand charge mismeasurement background contributions. Figure 2 shows the distributions of data eventsand the expected background in HT, Emiss

T , and the number of b-tagged jets Nb , as well as the significanceof the disagreement between the expected and observed yields in each bin, considering both the statisticaland major systematic uncertainties. The predicted and observed yields in this validation region are givenin tables 2 and 3. Events with three leptons are considered explicitly in table 3 since the fake/non-promptlepton background contribution from trilepton events is not negligible, and it is important to check that

8

Sample ee eµ µµ

Charge mis-ID 88 ± 2 ± 15 87 ± 2 ± 18 —Fake/Non-prompt 26 ± 6 ± 11 75 ± 7 ± 32 27 ± 3 ± 12tt̄W/Z/W+W− 2.18 ± 0.03 ± 0.14 6.39 ± 0.05 ± 0.41 3.68 ± 0.04 ± 0.25Dibosons 3.0 ± 0.5 ± 0.1 7.7 ± 0.5 ± 0.2 4.0 ± 0.4 ± 0.1Other bkg. 0.4 ± 0.1 1.4 ± 0.1 0.6 ± 0.1Total bkg. 120 ± 6 ± 19 178 ± 7 ± 37 35 ± 3 ± 12Data 135 144 30

Table 2: Observed and expected numbers of events in the low-HT+1b validation region for the same-sign dileptonchannels. The first uncertainty is statistical and the second is systematic. Systematic uncertainties are described insection 6.

Sample eee eeµFake/Non-prompt 3.0 ± 1.2 ± 1.3 6.0 ± 1.7 ± 2.6tt̄W/Z/W+W− 0.90 ± 0.02 ± 0.07 1.61 ± 0.02 ± 0.10Dibosons 1.9 ± 0.3 ± 0.1 2.7 ± 0.4 ± 0.1Other bkg. 0.11 ± 0.03 0.28 ± 0.03Total bkg. 5.9 ± 1.2 ± 1.3 10.6 ± 1.8 ± 2.6Data 11 8

Sample eµµ µµµ

Fake/Non-prompt 6.4 ± 1.7 ± 2.8 1.7 ± 1.0 ± 0.7tt̄W/Z/W+W− 1.97 ± 0.03 ± 0.12 1.65 ± 0.02 ± 0.12Dibosons 2.8 ± 0.3 ± 0.1 4.1 ± 0.4 ± 0.1Other bkg. 0.25 ± 0.04 0.11 ± 0.03Total bkg. 11.4 ± 1.8 ± 2.8 7.5 ± 1.0 ± 0.7Data 9 5

Table 3: Observed and expected numbers of events in the low-HT+1b validation region for the trilepton channels.The first uncertainty is statistical and the second is systematic. Systematic uncertainties are described in section 6.

this component of the background is well understood. The ‘other bkg.’ category includes W H/Z H , tt̄H ,three vector bosons, three top quarks, and SM four top quarks.

5 Signal region definition

The distributions of the total background estimate in HT, EmissT , Nb and the jet multiplicity after applying

the preselection are shown in figure 3, along with the distributions for some of the BSM signals consid-ered in this analysis. Based on the differences between the signal and background distributions, signalregions are defined to select events with at least one b-tagged jet and large HT and Emiss

T . The vector-likequark and four-top-quark signatures share a similar final-state topology, but the distribution of events in

9

[GeV]TH

0 100 200 300 400 500 600

Eve

nts

/ 40

GeV

0

20

40

60

80

100

120Data 2015

Fake/Non-prompt

Charge mis-ID-W++W/Z/Wtt

Dibosons

Other bkg

Stat. Unc.

Total Unc.

1 b-jet≥<400 GeV , T

100<HSS dilepton / trilepton

PreliminaryATLAS -1=13 TeV, 3.2 fbs

[GeV]TH0 100 200 300 400 500 600

Dat

a / B

kg

00.5

11.5

22.5 Total Unc. Stat. Unc.

(a)

[GeV]TmissE

0 50 100 150 200 250 300

Eve

nts

/ 20

GeV

0

20

40

60

80

100

120

140

160Data 2015

Fake/Non-prompt

Charge mis-ID-W++W/Z/Wtt

Dibosons

Other bkg

Stat. Unc.

Total Unc.

1 b-jet≥<400 GeV , T

100<HSS dilepton / trilepton

PreliminaryATLAS -1=13 TeV, 3.2 fbs

[GeV]TmissE

0 50 100 150 200 250 300

Dat

a / B

kg

00.5

11.5

22.5 Total Unc. Stat. Unc.

(b)

bN

0 1 2 3 4 5

Eve

nts

0

100

200

300

400

500 Data 2015

Fake/Non-prompt

Charge mis-ID-W++W/Z/Wtt

Dibosons

Other bkg

Stat. Unc.

Total Unc.

1 b-jet≥<400 GeV , T

100<HSS dilepton / trilepton

PreliminaryATLAS -1=13 TeV, 3.2 fbs

bN0 1 2 3 4 5

Dat

a / B

kg

00.5

11.5

22.5 Total Unc. Stat. Unc.

(c)

Figure 2: Distributions of (a) the scalar sum of the jet and lepton transverse momenta HT, (b) EmissT , and (c) Nb for

data and the expected background for events in the low-HT+1b validation region. The bottom panels display theratios of data to the total background prediction (‘Bkg’).

Definition Namee±e± + e±µ± + µ±µ± + eee + eeµ + eµµ + µµµ, Njets ≥ 2

400 < HT < 700 GeVNb = 1 SR0Nb = 2 Emiss

T > 40 GeV SR1Nb ≥ 3 SR2

HT ≥ 700 GeV

Nb = 140 < Emiss

T < 100 GeV SR3Emiss

T ≥ 100 GeV SR4

Nb = 240 < Emiss

T < 100 GeV SR5Emiss

T ≥ 100 GeV SR6Nb ≥ 3 Emiss

T > 40 GeV SR7

Table 4: Definitions of the different signal regions. The jet and b-jet multiplicities are denoted by Njets and Nb

respectively.

these variables differs between them. Therefore eight event categories are defined, placing different re-quirements on HT, Emiss

T , and Nb , as shown in table 4. Splitting the sample in this manner provides goodoverall efficiency for signal events, while allowing regions with the highest signal-to-background ratios tobe treated separately in the analysis, thereby enhancing the sensitivity to BSM physics. One consequenceof defining several signal categories is that the background estimates are subject to large statistical fluc-tuations. To mitigate this, all lepton flavours are summed within each category. The same signal regionsthat were defined in the 8 TeV ATLAS analysis [35] are used, in order to perform a direct check of themodest excess that was observed.

6 Systematic uncertainties

Tables 5 and 6 show the sources of systematic uncertainties that contribute more than 1% uncertainty onthe expected background or signal yield for each of the signal regions considered. For the yields derived

10

[GeV]TH

0 500 1000 1500 2000 2500

Eve

nts

/ 150

GeV

0

200

400

600

800

1000

1200

1400 178)× CI (100 fb) (tttt

206)× SM (100 fb) (tttt

93)×UED 1.2 TeV (75 fb) (

123)× 0.75 TeV (295 fb) (TT

Total Bkg

PreselectionSS dilepton / trilepton

PreliminaryATLAS -1=13 TeV, 3.2 fbs

(a)

[GeV]TmissE

0 100 200 300 400 500 600

Eve

nts

/ 40

GeV

0

200

400

600

800

1000

1200

1400

1600 178)× CI (100 fb) (tttt

206)× SM (100 fb) (tttt

93)×UED 1.2 TeV (75 fb) (

123)× 0.75 TeV (295 fb) (TT

Total Bkg

PreselectionSS dilepton / trilepton

PreliminaryATLAS -1=13 TeV, 3.2 fbs

(b)

bN

0 1 2 3 4 5

Eve

nts

0200400600800

10001200140016001800200022002400

178)× CI (100 fb) (tttt

206)× SM (100 fb) (tttt

93)×UED 1.2 TeV (75 fb) (

123)× 0.75 TeV (295 fb) (TT

Total Bkg

PreselectionSS dilepton / trilepton

PreliminaryATLAS -1=13 TeV, 3.2 fbs

(c)

jetsN

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14

Eve

nts

0

200

400

600

800

1000

1200

1400 178)× CI (100 fb) (tttt

206)× SM (100 fb) (tttt

93)×UED 1.2 TeV (75 fb) (

123)× 0.75 TeV (295 fb) (TT

Total Bkg

PreselectionSS dilepton / trilepton

PreliminaryATLAS -1=13 TeV, 3.2 fbs

(d)

Figure 3: Distributions of (a) scalar sum of jet and lepton transverse momenta HT, (b) EmissT , (c) Nb and (d) the

jet multiplicity for the expected background and for some of the signals considered in the analysis, after applyingthe preselection criteria. All of the signal cross sections are scaled so that the signals are normalized to the samenumber of events as the total background.

11

SourceSignal region

SR0 SR1 SR2 SR3 SR4 SR5 SR6 SR7Cross section 8 11 26 13 9 27 23 57

Jet energy scale 1 1 3 1 1 3 2 4

Jet energy resolution <1 2 2 2 < 1 1 <1 3

b-tagging efficiency 1 2 5 3 1 2 2 7

Luminosity 1 1 1 1 1 1 1 1

Fake/non-prompt leptons 17 7 15 13 26 13 17 17

Charge misID 8 3 7 5 3 6 5 8

Table 5: Uncertainty (in %) on the total background yield due to the leading sources of systematic uncertainty.

SourceSignal region

SR0 SR1 SR2 SR3 SR4 SR5 SR6 SR7Jet energy scale 2 12 2 6 4 3 3 3

Jet energy resolution 16 6 7 16 14 11 1 2

b-tagging efficiency 8 5 5 21 14 15 5 5

Lepton ID efficiency 1 1 1 4 2 2 2 1

Luminosity 2 2 2 2 2 2 2 2

Table 6: Uncertainty (in %) on the yield of a representative signal (four top quark production with SM kinematics)due to the leading sources of experimental systematic uncertainty.

from simulation, the largest source of uncertainty is usually the cross section times acceptance calcula-tion. This uncertainty is estimated based on variations in the PDFs and variations of the renormalizationand factorization scales [47]. These uncertainties, applied to the event yields shown in table 7, result inthe overall cross section uncertainties reported in table 5. The uncertainty on the integrated luminosityis 2.1%, derived using similar methodology as used for 8 TeV [43]. This uncertainty applies only to thebackgrounds estimated from simulation, not to the data-driven estimates of the fake/non-prompt leptonand electron charge mismeasurement backgrounds, so the overall contribution of the luminosity uncer-tainty shown in table 5 is less than 2.1%. The largest detector-specific uncertainties arise from the jetenergy scale [63], the b-tagging efficiency [65], and the lepton identification efficiency [56, 58].

Systematic uncertainties on the background contributions estimated from data are evaluated separately.Four effects are considered when assigning the systematic uncertainty on the predicted yield of eventsfrom electron charge mismeasurement: i) the statistical uncertainty on the probability for an electron tohave its charge mismeasured, ii) the difference observed in simulated Z boson events between the truecharge mismeasurement rate and the rate obtained by applying the same method as is used for the data, iii)the variation in the result observed when the upper and lower cut on the ee invariant mass defining the Zpeak region are increased and decreased by 10 GeV respectively, and iv) the uncertainty on the correctionfor the overlap in the measurement of charge misidentification and fake-electron background estimates,

12

SR index

0 1 2 3 4 5 6 7

Eve

nts

1−10

1

10

210

310

410Data 2015Fake/Non-prompt

Charge mis-ID-

W+W/Z/Wtt

Dibosons

Other bkg CI (100 fb)tttt

0.75TeV (295 fb)TT

Stat. Unc.

Total Unc.

SS dilepton / trilepton + b-jets PreliminaryATLAS -1=13 TeV, 3.2 fbs

SR index0 1 2 3 4 5 6 7

Dat

a / B

kg

00.5

11.5

22.5 Total Unc. Stat. Unc.

Figure 4: Expected background yields and observed data events in various signal regions. Uncertainties includeboth the statistical and systematic errors added in quadrature. The bottom panel displays the ratio of data to thetotal background prediction (‘Bkg’).

which is anti-correlated with the uncertainty on the fake-electron background. The magnitude of the totaluncertainty is ≈25%, with small variations between the different validation and signal regions due todependence on the event characteristics, as presented in tables 2 and 7. The expected yield of fake/non-prompt leptons is subject to uncertainties in the real and fake/non-prompt lepton efficiencies that arisefrom i) variations in the values of r and f when different control regions are used to measure them, ii)the finite statistics in the control regions used to measure r and f , and iii) the MC model used to subtractthe real lepton contribution from the fake/non-prompt lepton control region. When assessing effect i,the following alternative control regions are used: for electrons, the alternative fake/non-prompt controlregion requires one loose electron and Emiss

T < 20 GeV, while for muons, the alternative control regionrequires one loose muon, mT(W ) < 20 GeV and Emiss

T + mT(W ) < 60 GeV. In both cases the expectedcontribution from real leptons in the control region is subtracted using simulation. The alternative controlregions for r are formed by increasing the requirement on Emiss

T from > 150 GeV to > 175 GeV forelectrons and by increasing the requirement on mT(W ) from > 100 GeV to > 110 GeV for muons. Effectsi)–iii) sum to an overall systematic uncertainty of 54% for the set of events with at least one b-taggedjet and HT > 400 GeV. This uncertainty is applied to the fake/non-prompt background in all eight of thesignal regions.

13

SR0 SR1 SR2 SR3 SR4Fake/Non-prompt 16.3± 9.5 4.2 ±3.3 1.0 ± 0.9 1.8 ± 1.4 7.1±4.5Charge mis-ID 18.1± 4.1 14.9±3.5 1.2 ± 0.3 1.5 ± 0.4 2.1±0.5tt̄W/Z/W+W− 10.1± 1.4 9.2 ±1.3 1.0 ± 0.3 2.2 ± 0.3 3.1±0.5Dibosons 5.8 ± 1.0 0.5 ±0.2 0.03±0.07 1.6 ± 0.4 1.8±0.4Other bkg. 2.0 ± 1.0 1.7 ±0.9 0.3 ± 0.2 0.3 ± 0.2 0.5±0.3Total bkg. 52 ± 11 31 ± 5 3.6 ± 1.0 7.4 ± 1.5 15 ± 5tt̄tt̄ (SM) 0.5 ± 0.1 0.8 ±0.1 0.9 ± 0.1 0.2 ± 0.1 0.5±0.1tt̄tt̄ (CI) 0.26±0.04 0.6 ±0.1 0.6 ± 0.1 0.24±0.05 0.9±0.1UED 1.2 TeV <0.01 <0.01 <0.01 0.3 ± 0.1 3.8±0.8TT̄ 0.75 TeV 0.2 ± 0.1 0.31±0.1 0.04±0.04 0.9 ± 0.2 3.7±0.4Data 51 37 3 4 11

SR5 SR6 SR7Fake/Non-prompt 1.4±0.9 2.6±1.8 0.0 ±0.6Charge mis-ID 1.4±0.4 1.6±0.5 0.6 ±0.2tt̄W/Z/W+W− 2.3±0.6 3.0±0.7 0.8 ±0.4Dibosons 0.3±0.1 0.2±0.1 0.0 ±0.1Other bkg. 0.4±0.2 0.7±0.4 0.5 ±0.3Total bkg. 5.8±1.2 8.1±2.0 1.9 ±0.8tt̄tt̄ (SM) 0.7±0.1 1.8±0.2 3.6 ±0.4tt̄tt̄ (CI) 0.6±0.1 2.2±0.2 5.2 ±0.4UED 1.2 TeV 0.6±0.1 6.6±0.7 10.1±0.8TT̄ 0.75 TeV 1.3±0.2 5.0±0.5 3.2 ±0.4Data 6 3 2

Table 7: Observed and expected number of events with the total uncertainties for the signal regions. The expectedUED and TT̄ yields are normalized to their theoretical cross sections of 75 fb and 295 fb, respectively, while theexpected tt̄tt̄ (SM) and tt̄tt̄ (CI) are both normalized to a cross section of 100 fb.

[GeV]TH

500 1000 1500 2000 2500

Eve

nts

/ 300

GeV

1−10

1

10

210

310

Data 2015

Fake/Non-prompt

Charge mis-ID-W+W/Z/Wtt

Dibosons

Other bkg

CI (100 fb)tttt

0.75TeV (295 fb)TT

Stat. Unc.

Total Unc.

1 b-jet≥2 jets, ≥>40 GeV

T

miss>400 GeV, ETH

SS dilepton / trilepton PreliminaryATLAS -1=13 TeV, 3.2 fbs

[GeV]TH500 1000 1500 2000 2500

Dat

a / B

kg

00.5

11.5

22.5 Total Unc. Stat. Unc.

(a)

[GeV]TmissE

0 100 200 300 400 500 600

Eve

nts

/ 75

GeV

1−10

1

10

210

310Data 2015

Fake/Non-prompt

Charge mis-ID-W+W/Z/Wtt

Dibosons

Other bkg

CI (100 fb)tttt

0.75TeV (295 fb)TT

Stat. Unc.

Total Unc.

1 b-jet≥2 jets, ≥>40 GeV

T

miss>400 GeV, ETH

SS dilepton / trilepton PreliminaryATLAS -1=13 TeV, 3.2 fbs

[GeV]TmissE

0 100 200 300 400 500 600

Dat

a / B

kg

00.5

11.5

22.5 Total Unc. Stat. Unc.

(b)

bN

0 1 2 3 4 5

Eve

nts

1−10

1

10

210

310

Data 2015

Fake/Non-prompt

Charge mis-ID-W+W/Z/Wtt

Dibosons

Other bkg

CI (100 fb)tttt

0.75TeV (295 fb)TT

Stat. Unc.

Total Unc.

1 b-jet≥2 jets, ≥>40 GeV

T

miss>400 GeV, ETH

SS dilepton / trilepton PreliminaryATLAS -1=13 TeV, 3.2 fbs

bN0 1 2 3 4 5

Dat

a / B

kg

00.5

11.5

22.5 Total Unc. Stat. Unc.

(c)

Figure 5: Distributions of (a) the scalar sum of the jet and lepton transverse momenta HT, (b) EmissT , and (c) Nb for

data and the expected background for events in the eight signal regions. The bottom panels display the ratios ofdata to the total background prediction (‘Bkg’).

14

7 Results

The observed yields for each signal selection are given in table 7 and figure 4. Figure 5 shows thedistributions of data events and the expected background in HT, Emiss

T , and Nb , summed over all eightsignal regions. The CLs method [69] is used to assess the consistency between the observed yields andeach potential BSM physics signal, where a profile log-likelihood ratio is used as the test statistic. Foreach model, the test statistic is defined as

qµ = −2 ln *.,

L(µ, ˆ̂θµ )

L( µ̂, θ̂)+/-

(3)

where the likelihood is a product of the Poisson likelihood describing the statistical fluctuations of theselected sample and the Gaussian distributions modelling the systematic uncertainties. L( µ̂, θ̂) is uncon-ditionally maximized with respect to both the signal-strength parameter µ, a multiplicative factor appliedto the predicted signal cross section, and θ, a set of nuisance parameters that encodes the effect of sys-tematic uncertainties on the signal and background expectations. L(µ, ˆ̂θµ ) is conditionally maximized atconstant µ (µ = 1 in the test statistic used for limit setting). Probability distributions of qµ under eachhypothesis can be obtained either by generating pseudo-experiments taking into account the statisticaland systematical fluctuations, or by using the asymptotic approximation [70, 71]. The latter was used inthis analysis. The quantities CLs+b and CLb are defined as the fractions of signal-plus-background andbackground-only pseudo-experiments with qµ larger than a reference value qµ,r . Signal cross sectionsfor which CLs = CLs+b/CLb < 0.05 are deemed to be excluded at the 95% confidence level (CL).Expected exclusion limits are computed by setting qµ,r to the median of the qµ probability distributionunder the background-only hypothesis; these are the basis for assessing the intrinsic sensitivity of theanalysis. Observed exclusion limits are obtained by setting qµ,r to the value of qµ for the data.

95% CL limits (upper limits on cross sections, or lower limits on masses) relevant for each model forthe VLQ and four-top-quark production are calculated. Limits on the VLQ pair-production cross section,assuming the branching fractions to W , Z , and H modes prescribed by the singlet model, are shown infigure 6. Comparison with the calculated cross section results in lower limits on the B-quark mass of0.83 TeV and on the T quark mass of 0.78 TeV at 95% CL. The expected limits in the absence of a signalcontribution are 0.75 TeV for the B-quark mass and 0.73 TeV for the T-quark mass. If the three branchingfractions are allowed to vary independently, subject to the constraint that they sum to one, the data canbe interpreted as excluding at 95% CL some of the possible sets of branching ratios for a given B- orT-quark mass. These exclusions are shown in figures 7 and 8. Limits on T5/3 production are set for pairproduction only. The limits are shown in figure 9 and correspond to a mass limit of 0.99 TeV (0.92 TeVexpected).

The upper limit on the cross section for four-top-quark production is 95 fb assuming SM kinematics and67 fb for production with a BSM-physics contact interaction (expected limits are respectively 107 fb and79 fb). The cross section limit for the contact interaction case is lower than for the SM since the contactinteraction tends to result in final-state objects with larger pT, which increases the selection efficiency.The limits are also interpreted in the context of specific BSM physics models. For the contact interactionmodel, the upper limit on |C4t |/Λ

2 is 3.5 TeV−2, as illustrated in figure 10(a). The observed limits on thecross section times branching ratio for the 2UED/RPP signal are shown in figure 10(b). From these, a limiton mKK can be extracted for the symmetric case (R4 = R5) under the assumption that the A(1,1)

µ decays tott̄ with 100% branching fraction: mKK > 1.4 TeV (where the expected limit is 1.3 TeV). Figure 11 showsthe limit in the (mKK, ξ) plane.

15

[GeV]Bm

400 600 800 1000 1200 1400

) [p

b]B

B→

(pp

σ

3−10

2−10

1−10

1

10

210

310Observed limitExpected limit

σ 1 ±σ 2 ±

Theory (NNLO)

-1 = 13 TeV, 3.2 fbs

SS dilepton / trilepton + b-jetsSU(2) Singlet

ATLAS Preliminary

(a)

[GeV]Tm

400 600 800 1000 1200 1400

) [p

b]T

T→

(pp

σ

3−10

2−10

1−10

1

10

210

310Observed limitExpected limit

σ 1 ±σ 2 ±

Theory (NNLO)

-1 = 13 TeV, 3.2 fbs

SS dilepton / trilepton + b-jetsSU(2) Singlet

ATLAS Preliminary

(b)

Figure 6: Observed (solid line) and expected (dashed line) 95% CL upper limits on (a) the BB̄ cross section as afunction of the B mass, and (b) the TT̄ cross section as a function of the T mass. These limits assume branchingratios given by the model where the B and T quarks exist as weak isospin singlets [25]. The surrounding shadedbands correspond to ±1 and ±2 standard deviations around the expected limit. The blue line shows the theoreticalprediction.

Wt)→BR(B

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Hb)

→B

R(B

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

95%

CL

mas

s lim

it [G

eV]

400

500

600

700

800

900

1000ATLAS Preliminary

-1 = 13 TeV, 3.2 fbsSS dilepton / trilepton + b-jets

Expected limit

650

700

750

800

850

(a)

Wt)→BR(B

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Hb)

→B

R(B

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

95%

CL

mas

s lim

it [G

eV]

400

500

600

700

800

900

1000ATLAS Preliminary

-1 = 13 TeV, 3.2 fbsSS dilepton / trilepton + b-jets

Observed limit

650

700

750

800

850

900

(b)

Figure 7: (a) Expected and (b) observed 95% CL lower limit on the mass of the B quark in the plane of BR(B → Hb)versus BR(B → Wt). Contour lines of constant mass limit values are are provided to guide the eye.

16

Wb)→BR(T

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Ht)

→B

R(T

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

95%

CL

mas

s lim

it [G

eV]

400

500

600

700

800

900

1000ATLAS Preliminary

-1 = 13 TeV, 3.2 fbsSS dilepton / trilepton + b-jets

Expected limit

650

700

750750

800

(a)

Wb)→BR(T

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Ht)

→B

R(T

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

95%

CL

mas

s lim

it [G

eV]

400

500

600

700

800

900

1000ATLAS Preliminary

-1 = 13 TeV, 3.2 fbsSS dilepton / trilepton + b-jets

Observed limit

650

700

750

800

850

(b)

Figure 8: (a) Expected and (b) observed 95% CL lower limit on the mass of the T quark in the plane of BR(T → Ht)versus BR(T → W b). Contour lines of constant mass limit values are are provided to guide the eye.

[GeV]5/3Tm

400 600 800 1000 1200 1400

) [p

b]5/

3T

5/3

T→

(pp

σ

3−10

2−10

1−10

1

10

210

310Observed limitExpected limit

σ 1 ±σ 2 ±

Theory (NNLO)

-1 = 13 TeV, 3.2 fbsSS dilepton / trilepton + b-jets

ATLAS Preliminary

Figure 9: Observed (solid line) and expected (dashed line) 95% CL upper limits on the T5/3T5/3 cross section as afunction of the T5/3 mass. The surrounding shaded bands correspond to ±1 and ±2 standard deviations around theexpected limit. The blue line shows the theoretical prediction..

17

[TeV]Λ

4 5 6 7 8 9 10 11 12 13

|4t

|C

0

50

100

150

200

250

300

Excluded

Observed

Expected

σ 1 ±σ 2 ±

JHEP 1510 (2015) 150SS dilepton / trilepton + b-jets, Run 1

ATLAS-CONF-2016-020l+jets, Run 2

ATLAS Preliminary-1 = 13 TeV, 3.2 fbs

SS dilepton / trilepton + b-jets

(a)

[TeV]KKm

0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9

) [p

b]ttt t

→(1

,1)

A(1

,1)

A→

(pp

σ

3−10

2−10

1−10

1

10Observed limitExpected limit

σ 1 ±σ 2 ±

Theory (LO)

-1 = 13 TeV, 3.2 fbs

SS dilepton / trilepton + b-jets2UED/RPP model

ATLAS Preliminary

(b)

Figure 10: Limits obtained from the search for four-top-quark production. (a) Expected and observed limits on thecoupling constant |C4t | in the contact interaction model for four-top production as a function of the BSM physicsenergy scale Λ. The region in the upper left, corresponding to |C4t |/Λ

2 > 3.5 TeV−2, is excluded at 95% CL. Thelower bound on Λ is chosen so that the fraction of signal events with a partonic four-top-quark invariant mass abovethat value of Λ is below 30%, in order to to stay within the range of validity of the EFT approach. (b) Expected andobserved limits on the four-top-quark production rate for the 2UED/RPP model in the symmetric case. The theoryline corresponds to the production of four-top-quark events by tier (1,1) with a branching ratio of A(1,1) to tt̄ of100%. The surrounding shaded bands correspond to ±1 and ±2 standard deviations around the expected limit.

[TeV]4=1/RKKm

0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8

5/R 4

=R

ξ

1

1.2

1.4

1.6

1.8

2

2.2

2.4 Excluded region

Observed

Expected

σ 1 ±

σ 2 ±

PreliminaryATLAS-1 = 13 TeV, 3.2 fbs

2UED/RPP modelSS dilepton / trilepton + b-jets

)=100%t t→(1,1)BR(A

JHEP 1510 (2015) 150SS dilepton / trilepton + b-jets, Run 1

ATLAS-CONF-2016-020l+jets, Run 2

Figure 11: Expected and observed limit in the (mKK = 1/R4, ξ = R4/R5) plane for the 2UED/RPP model. Thesurrounding shaded bands correspond to ±1 and ±2 standard deviations around the expected limit.

18

8 Conclusion

A search for BSM physics has been performed using pp collisions at√

s = 13 TeV corresponding toan integrated luminosity of 3.2 fb−1 recorded by the ATLAS detector at the LHC, where events with atleast two leptons, including a pair of the same electric charge, at least one b-tagged jet, sizable missingtransverse momentum, and large HT were considered. Several BSM processes could enhance the yieldof such events over the small SM expectation. The search was performed in the context of several BSMscenarios, with a set of eight signal regions defined for different models. Good agreement between thedata yield and the expected background was observed in all of these regions, meaning that the modestexcess observed in a similar analysis of the ATLAS data at

√s = 8 TeV is not confirmed. The regions

of parameter space excluded by the data are quantified by setting 95% CL limits, as follows: the massesof vector-like B and T quarks are constrained to mB > 0.83 TeV, mT > 0.78 TeV (assuming branchingfractions to the W , Z , and H decay modes arising from a singlet model), the mass of the T5/3 quark isgreater than 0.99 TeV, the four-top production cross section is less than 95 fb and 67 fb assuming SM andBSM kinematics respectively, and the Kaluza–Klein mass in the context of models with two universalextra dimensions is greater than 1.4 TeV.

References

[1] ATLAS Collaboration, Observation of a new particle in the search for the Standard Model Higgsboson with the ATLAS detector at the LHC, Phys. Lett. B 716 (2012) 1, arXiv: 1207.7214[hep-ex].

[2] CMS Collaboration, Observation of a new boson at a mass of 125 GeV with the CMS experimentat the LHC, Phys. Lett. B 716 (2012) 30, arXiv: 1207.7235 [hep-ex].

[3] B. A. Dobrescu and C. T. Hill, Electroweak symmetry breaking via a top condensation seesawmechanism, Phys. Rev. Lett. 81 (1998) 2634, arXiv: hep-ph/9712319 [hep-ph].

[4] R. S. Chivukula et al., Top quark seesaw theory of electroweak symmetry breaking, Phys. Rev. D59 (1999) 075003, arXiv: hep-ph/9809470 [hep-ph].

[5] H.-J. He, C. T. Hill, and T. M. Tait, Top quark seesaw model, vacuum structure, and electroweakprecision constraints, Phys. Rev. D 65 (2002) 055006, arXiv: hep-ph/0108041 [hep-ph].

[6] C. T. Hill and E. H. Simmons, Strong dynamics and electroweak symmetry breaking, Phys. Rept.381 (2003) 235, arXiv: hep-ph/0203079 [hep-ph].

[7] R. Contino, L. Da Rold, and A. Pomarol, Light custodians in natural composite Higgs models,Phys. Rev. D 75 (2007) 055014, arXiv: hep-ph/0612048 [hep-ph].

[8] C. Anastasiou, E. Furlan, and J. Santiago, Realistic composite Higgs models, Phys. Rev. D 79(2009) 075003, arXiv: 0901.2117 [hep-ph].

[9] K. Kong, M. McCaskey, and G. W. Wilson, Multi-lepton signals from the top-prime quark at theLHC, JHEP 1204 (2012) 079, arXiv: 1112.3041 [hep-ph].

[10] A. Carmona, M. Chala, and J. Santiago, New Higgs production mechanism in composite Higgsmodels, JHEP 1207 (2012) 049, arXiv: 1205.2378 [hep-ph].

[11] M. Gillioz, et al., Higgs low-energy theorem (and its corrections) in composite models, JHEP 1210(2012) 004, arXiv: 1206.7120 [hep-ph].

19

[12] A. Davidson and K. C. Wali, Family mass hierarchy from universal seesaw mechanism, Phys. Rev.Lett. 60 (18 May 1988) 1813.

[13] K. S. Babu and R. N. Mohapatra, Solution to the strong CP problem without an axion, Phys. Rev.D 41 (4 Feb. 1990) 1286.

[14] B. Grinstein, M. Redi, and G. Villadoro, Low scale flavor gauge symmetries, JHEP 1011 (2010) 067,arXiv: 1009.2049 [hep-ph].

[15] D. Guadagnoli, R. N. Mohapatra, and I. Sung, Gauged flavor group with left-right symmetry, JHEP1104 (2011) 093, arXiv: 1103.4170 [hep-ph].

[16] N. Arkani-Hamed, A. G. Cohen, E. Katz, A. E. Nelson, The littlest Higgs, JHEP 0207 (2002) 034,arXiv: hep-ph/0206021 [hep-ph].

[17] T. Han et al., Phenomenology of the little Higgs model, Phys. Rev. D 67 (2003) 095004, arXiv:hep-ph/0301040 [hep-ph].

[18] M. Perelstein, M. E. Peskin, and A. Pierce, Top quarks and electroweak symmetry breaking in littleHiggs models, Phys. Rev. D 69 (2004) 075002, arXiv: hep-ph/0310039 [hep-ph].

[19] M. Schmaltz and D. Tucker-Smith, Little Higgs theories, Ann. Rev. Nucl. Part. Sci. 55 (2005) 229,arXiv: hep-ph/0502182 [hep-ph].

[20] M. S. Carena et al., Collider signatures for new T-parity-odd quarks in little Higgs models, Phys.Rev. D 75 (2007) 091701, arXiv: hep-ph/0610156 [hep-ph].

[21] S. Matsumoto, T. Moroi, and K. Tobe, Testing the Littlest Higgs Model with T-parity at the CERNLarge Hadron Collider, Phys. Rev. D 78 (2008) 055018, arXiv: 0806.3837 [hep-ph].

[22] J. A. Aguilar-Saavedra, R. Benbrik, S. Heinemeyer, M. Pérez-Victoria, Handbook of vectorlikequarks: Mixing and single production, Phys. Rev. D 88 (2013) 094010, arXiv: 1306 . 0572[hep-ph].

[23] J. A. Aguilar-Saavedra, Effects of mixing with quark singlets, Phys. Rev. D 67 (3 Feb. 2003) 035003.

[24] Y. Okada and L. Panizzi, LHC signatures of vector-like quarks, Adv. High Energy Phys. 2013(2013) 364936, arXiv: 1207.5607 [hep-ph].

[25] J. A. Aguilar-Saavedra, Identifying top partners at LHC, JHEP 0911 (2009) 030, arXiv: 0907.3155 [hep-ph].

[26] M. Czakon, P. Fiedler, and A. Mitov, Total top-quark pair-production cross section at hadroncolliders through O(α4

S ), Phys. Rev. Lett. 110 (2013) 252004, arXiv: 1303.6254 [hep-ph].

[27] M. Czakon and A. Mitov, Top++: A program for the calculation of the top-pair cross-section athadron colliders, Comput. Phys. Commun. 185 (2014) 2930, arXiv: 1112.5675 [hep-ph].

[28] V. Barger, A. L. Stange, and R. J. N. Phillips, Four-heavy-quark hadroproduction, Phys. Rev. D44 (1991) 1987.

[29] V. Barger, W.-Y. Keung, and B. Yencho, Triple-top signal of new physics at the LHC, Phys. Lett.B 687 (2010) 70, arXiv: 1001.0221 [hep-ph].

[30] B. Lillie, J. Shu, and T. M. P. Tait, Top Compositeness at the Tevatron and LHC, JHEP 0804(2008) 087, arXiv: 0712.3057 [hep-ph].

[31] A. Pomarol and J. Serra, Top quark compositeness: Feasibility and implications, Phys. Rev. D 78(2008) 074026.

20

[32] C. Degrande et al., Non-resonant new physics in top pair production at hadron colliders, JHEP1103 (2011) 125, arXiv: 1010.6304 [hep-ph].

[33] M. Guchait, F. Mahmoudi, and K. Sridhar, Associated production of a Kaluza-Klein excitation ofa gluon with a tt̄ pair at the LHC, Phys. Lett. B 666 (2008) 347, arXiv: 0710.2234 [hep-ph].

[34] G. Cacciapaglia, A. Deandrea, and J. Llodra-Perez, A Dark Matter candidate from Lorentz Invari-ance in 6D, JHEP 1003 (2010) 083, arXiv: 0907.4993 [hep-ph].

[35] ATLAS Collaboration, Search for anomalous production of prompt same-sign lepton pairs andpair-produced doubly charged Higgs bosons with

√s = 8 TeV pp collisions using the ATLAS

detector, JHEP 1503 (2015) 041, arXiv: 1412.0237 [hep-ex].

[36] CMS Collaboration, Search for new physics in events with same-sign dileptons and jets in ppcollisions at

√s = 8 TeV, JHEP 1401 (2014) 163, arXiv: 1311.6736.

[37] CMS Collaboration, Search for new physics in same-sign dilepton events in proton-proton colli-sions at

√s = 13 TeV, (2016), arXiv: 1605.03171 [hep-ex].

[38] CMS Collaboration, Search for top quark partners with charge 5/3 at√

s = 13 TeV, CMS-PAS-B2G-15-006, 2015, url: https://cds.cern.ch/record/2114805.

[39] CMS Collaboration, Inclusive search for a vector-like T quark with charge 23 in pp collisions at

√s = 8 TeV, Phys. Lett. B 729 (2014) 149, arXiv: 1311.7667 [hep-ex].

[40] ATLAS Collaboration, Search for supersymmetry at√

s=8 TeV in final states with jets and twosame-sign leptons or three leptons with the ATLAS detector, JHEP 1406 (2014) 035, arXiv:1404.2500 [hep-ex].

[41] ATLAS Collaboration, Search for supersymmetry at√

s = 13 TeV in final states with jets and twosame-sign leptons or three leptons with the ATLAS detector, Eur. Phys. J. C 76 (2016) 259, arXiv:1602.09058 [hep-ex].

[42] ATLAS Collaboration, The ATLAS experiment at the CERN Large Hadron Collider, JINST 3(2008) S08003.

[43] ATLAS Collaboration, Improved luminosity determination in pp collisions at√

s = 7 TeV usingthe ATLAS detector at the LHC, Eur. Phys. J. C 73 (2013) 2518, arXiv: 1302.4393 [hep-ex].

[44] ATLAS Collaboration, ATLAS Insertable B-Layer Technical Design Report, CERN-LHCC-2010-013, ATLAS-TDR-19, Sept. 2010, url: https://cds.cern.ch/record/1291633.

[45] S. Carrazza, S. Forte, and J. Rojo, “Parton Distributions and Event Generators,” Proceedings, 43rdInternational Symposium on Multiparticle Dynamics (ISMD 13), 2013 89, arXiv: 1311.5887[hep-ph].

[46] T. Sjostrand, S. Mrenna, and P. Z. Skands, A Brief Introduction to PYTHIA 8.1, Comput. Phys.Commun. 178 (2008) 852, arXiv: 0710.3820 [hep-ph].

[47] J. Alwall et al., The automated computation of tree-level and next-to-leading order differentialcross sections, and their matching to parton shower simulations, JHEP 1407 (2014) 079, arXiv:1405.0301 [hep-ph].

[48] P. Meade and M. Reece, BRIDGE: Branching ratio inquiry / decay generated events, (2007),arXiv: hep-ph/0703031 [hep-ph].

[49] G. Cacciapaglia, A. Deandrea, and J. Llodra-Perez, The Universal Real Projective Plane: LHCphenomenology at one loop, JHEP 1110 (2011) 146, arXiv: 1104.3800 [hep-ph].

21

[50] T. Gleisberg et al., Event generation with SHERPA 1.1, JHEP 0902 (2009) 007, arXiv: 0811.4622[hep-ph].

[51] M. Bahr et al., Herwig++ Physics and Manual, Eur. Phys. J. C 58 (2008) 639, arXiv: 0803.0883[hep-ph].

[52] H.-L. Lai et al., New parton distributions for collider physics, Phys. Rev. D 82 (2010) 074024,arXiv: 1007.2241 [hep-ph].

[53] J. Pumplin et al., New generation of parton distributions with uncertainties from global QCDanalysis, JHEP 0207 (2002) 012, arXiv: hep-ph/0201195 [hep-ph].

[54] GEANT4 Collaboration, S. Agostinelli et al., GEANT4: A Simulation toolkit, Nucl. Instrum. Meth.A 506 (2003) 250.

[55] ATLAS Collaboration, The ATLAS simulation infrastructure, Eur. Phys. J. C 70 (2010) 823, arXiv:1005.4568 [physics.ins-det].

[56] ATLAS Collaboration, Electron identification measurements in ATLAS using√

s = 13 TeV datawith 50 ns bunch spacing, ATL-PHYS-PUB-2015-041, Sept. 2015, url: https://cds.cern.ch/record/2048202.

[57] K. Rehermann and B. Tweedie, Efficient identification of boosted semileptonic top quarks at theLHC, JHEP 1103 (2011) 059, arXiv: 1007.2221 [hep-ph].

[58] ATLAS Collaboration, Muon reconstruction performance in early√

s = 13 TeV data, ATL-PHYS-PUB-2015-037, Aug. 2015, url: https://cds.cern.ch/record/2047831.

[59] M. Cacciari, G. P. Salam, and G. Soyez, The anti-kt jet clustering algorithm, JHEP 0804(2008) 063, arXiv: 0802.1189 [hep-ph].

[60] M. Cacciari and G. P. Salam, Dispelling the N3 myth for the kt jet-finder, Phys. Lett. B 641(2006) 57, arXiv: hep-ph/0512210 [hep-ph].

[61] M. Cacciari, G. P. Salam, and G. Soyez, FastJet User Manual, Eur. Phys. J. C 72 (2012) 1896,arXiv: 1111.6097 [hep-ph].

[62] ATLAS Collaboration, Performance of pile-up mitigation techniques for jets in pp collisions withthe ATLAS detector, Nucl. Instrum. Meth. A 824 (2016) 367, arXiv: 1510.03823 [hep-ex].

[63] ATLAS Collaboration, Jet Calibration and Systematic Uncertainties for Jets Reconstructed inthe ATLAS Detector at

√s = 13 TeV, ATL-PHYS-PUB-2015-015, July 2015, url: https:

//cds.cern.ch/record/2037613.

[64] ATLAS Collaboration, Monte Carlo Calibration and Combination of In-situ Measurements of JetEnergy Scale, Jet Energy Resolution and Jet Mass in ATLAS, (2015), url: https://cds.cern.ch/record/2044941.

[65] ATLAS Collaboration, Commissioning of the ATLAS b-tagging algorithms using tt̄ events in earlyRun-2 data, (Aug. 2015), url: https://cds.cern.ch/record/2047871.

[66] ATLAS Collaboration, Performance of b-Jet Identification in the ATLAS Experiment, JINST 11(2016) P04008, arXiv: 1512.01094 [hep-ex].

[67] ATLAS Collaboration, Performance of missing transverse momentum reconstruction for the AT-LAS detector in the first proton-proton collisions at at

√s= 13 TeV, ATL-PHYS-PUB-2015-027,

July 2015, url: https://cds.cern.ch/record/2037904.

22

[68] D0 Collaboration, Extraction of the width of the W boson from measurements of σ(pp̄→ W + X )×B(W → eν) and σ(pp̄ → Z + X ) × B(Z → ee) and their ratio, Phys. Rev. D 61 (2000) 072001,arXiv: hep-ex/9906025 [hep-ex].

[69] A. L. Read, Presentation of search results: The CL(s) technique, J. Phys. G 28 (2002) 2693.

[70] G. Cowan et al., Asymptotic formulae for likelihood-based tests of new physics, Eur. Phys. J. C 71(2011) 1554, arXiv: 1007.1727 [physics.data-an].

[71] G. Cowan et al., Erratum to: Asymptotic formulae for likelihood-based tests of new physics, Eur.Phys. J. C 73 (2013) 2501.

23

Appendix

(a)

Figure 12: e+e−µ+ event display passing the SR7 selection (HT = 0.9 TeV). Electron tracks are in blue and themuon track is in red. Blue cones represent b-tagged jets.

24

(a)

Figure 13: e−µ− event display passing the SR7 selection (HT = 1.1 TeV). The electron track is in blue and the muontrack is in red. Blue cones represent b-tagged jets.

25