High-granularity measurements of the detector are combined to form reconstructed particle candi- dates. This analysis requires the reconstruction of electrons, muons, jets, photons, and hadronic tau decays to characterize each of the recorded events and test the SM and BSM hypotheses. The algorithms used to reconstruct individual objects were presented in Section 4.4. Here we discuss
analysis-specific choices, balancing the efficiency of reconstructing the targeted objects with the purity of the sample obtained. These reconstructed objects are used to build more complicated full-event observables.
Generally two sets of objects are defined. Baseline objects make loose requirements on identifica- tion criteria and kinematics and are used for the purposes of overlap removal. This procedure ensures that detector signatures are not reconstructed as multiple objects and energies double-counted. A tighter ‘signal’ selection is applied to baseline quantities which satisfy the overlap removal criteria, providing the ultimate objects used for the final analysis selections.
6.5.1
Electrons
Electron candidates are selected which satisfy selections on the identification and isolation param- eters described in Section 4.4.2. Two classes of signal electrons are defined to be used in separate analyses. Loose leptons are used by default and use relaxed identification and isolation criteria to maximize acceptance to high-pTsignals where other selections efficiently reduce backgrounds from non-prompt and mis-identified leptons. Tight leptons are used in specialized search regions targeting signatures with low-pTleptons where tighter identification is useful. Due to the lower requirement onpT, tight signal leptons are not a subset of loose leptons. The selection criteria for baseline and signal leptons is summarized in Table6.2.
The likelihood operating points used are denoted ‘VeryLooseLH’, ‘LooseAndBLayerLH’, and ‘TightLLH’. The VeryLooseLH and LooseAndBLayerLH requirements are variations of the Loose working point which is found to be 93% efficient for prompt electrons with ET = 40 GeV. The former requires only a single hit in the pixel detector, while the latter requires two, one of which must be in the innermost pixel layer. The tight operating point is determined to be 80% efficient for 40 GeV electrons.
Isolation variables are defined requiring that tracks (calorimeter deposits) that are not associated to the reconstructed electron but are within ∆R= 0.2 haveEiso
T,cone (pisoT,cone) less than a particular value. The ‘LooseTrackOnly’ isolation criterion does not enforce the calorimeter requirement and achieves an efficiency of 99% independent of electron ET and η. The ‘FixedCutTight’ criterion requires that bothEiso
6. A Search for Top Squarks 102
6.5.2
Muons
Baseline muons are identified using the Loose working point, which accepts all four varieties of muon described in Section4.4.3. The majority of reconstructed muons are combined muons (approximately 97.5% of muons satisfying |η| < 2.5), with extrapolated muons are allowed only in the region beyond the ID coverage 2.5<|η|<2.7, and calo-tagged (1.5%) and segment-tagged (1.0%) muons accepted only in the region|η|<0.1. Additional requirements on the number of MS segments and the significance of the muonpT are enforced. Signal muons are required to additionally pass the Medium identification criteria, which rejects calo-tagged and segment-tagged muon candidates. A comparison of the efficiency to reconstruct muons with the Loose and Medium working points, as measured inZ→µ+µ− events is shown in Figure 6.9.
Signal muons are further required to the pass the ‘FixedCutTightTrackOnly’ isolation require- ment which requires thatpiso
T,cone be less than 6% of the muonpT, where the isolation variable in this case is computed from tracks within a cone of ∆R= 0.3. The selection criteria for baseline and signal muons is summarized in Table6.3.
In the following description of the stop analysis, leptons are collectively referred to as passing the ‘loose’ or ’tight’ identification criteria; while variable criteria are used for electrons, no change is made for muons. The only distinctions is that lowerpTmuons (to 4 GeV) are included in analyses using tight leptons.
6.5.3
Hadronic tau decays
ttbar decays with hadronic taus comprise a significant source of background for some signal se- lections, motivating the use of loose tau identification criterion. Only a single category of tau (reconstructed as described in Section4.4.5) is used in the analysis, with those selected satisfying Table 6.2: Selection criteria for electrons is summarized. In addition to baseline objects defined for overlap removal, ‘loose’ and ‘tight’ categories of signal leptons are defined. Tight electrons are defined at lowerETand use tighter identification and isolation criteria.
Selection criteria Baseline Loose signal Tight signal
ET >5 GeV >25 GeV >5 GeV
|η| <2.47
Identification VeryLooseLH LooseAndBLayerLH TightLLH
Isolation – LooseTrackOnly FixedCutTight
d0 significance – <5
2.5 − −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5
Efficiency
0.96
0.98
1
2.5 − −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.50.6
0.65
ATLAS -1 = 13 TeV, 3.2 fb s Data MC µ µ → Zη
2.5
−
−2−1.5−1−0.5
0 0.5 1 1.5 2 2.5
Data / MC 0.98
1
1.02
Stat onlyStat only Sys Sys ⊕⊕ Stat Stat muons Medium | < 0.1) η muons (| LooseFigure 6.9: The reconstruction efficiencies for muons identified with the Loose and Medium working points are compared as a function of pseudorapidity. In the top panel, efficiencies for data and MC are shown with the corresponding ratio (“scale factors”) shown in the lower panel. Loose muons include calo-tagged and segment-tagged muons in the region |η| < 0.1, allowing for significantly enhanced efficiency in this region. Figure is reproduced from Ref. [99].
Table 6.3: Selection criteria is summarized for muons, with identification and isolation working points are defined in the text. Muons as low in pT as 4 GeV are utilized only for the soft lepton analysis targeting the Higgsino LSP signatures.
Selection criteria Baseline Signal
ET >5 GeV >4 or 25 GeV
|η| <2.7
Identification Loose Medium
Isolation – FixedCutTightTrackOnly
d0 significance – <3
6. A Search for Top Squarks 104
pT>20 GeV and |η|<2.5. Candidates are identified using the looses working point, found to be 60% (50%) efficient for one-prong (three-prong) decays. The total charge of the tracks associated with the tau is required to be opposite to that of the identified lepton.
6.5.4
Jets
Jets are clustered using the anti-k⊥ algorithm with radius parameter R = 0.4 as described in Section 4.4.4. Jets from SM backgrounds and SUSY signals are generally expected to be central, so jets are only considered in the region |η|<2.5 which is also well-covered by the inner tracker. For low-pTjets in the tracker acceptance, jets from pile-up can be efficiently rejected by requiring that the charged particle tracks within the jet cone originate from the primary vertex. This is accomplished with a requirement on the JVT discriminant that is found to be 92% efficient in signal jets from the hard-scatter.
Jets are considered b-tagged which pass a MV2c10 tagging requirement that was found to be 77.0% efficient for bjets in tt¯events. The corresponding rejection rates (inverse of efficiency) are 134 for light jets, 6.21 for charm and 22.0 for taus.
6.5.5
E
TmissETmiss is defined as described in Section 4.4.6, using baseline objects described above as inputs. Photons and hadronic taus are not included, but may enter as electrons, jets, or via the soft term.
6.5.6
Overlap removal
To avoid double-counting of momenta from physics objects reconstructed from the same detector outputs an overlap removal (OR) procedure is enforced. Categories of baseline objects are com- Table 6.4: A summary of selection criteria for calorimeter jets used in the analysis. JVT is a requirement on the ‘jet vertex tagger’ which removed pileup jets at low-pT.
Selection criteria Baseline jets Signal jets
ET >20 GeV >25 GeV
|η| <2.5 <2.5
JVT – forpT<60 GeV
pared to check for overlaps in an ordered procedure so that removed objects are not considered in subsequent steps. The comparisons considered are, in order of precedence:
1. Muon & electron. If ∆R < 0.01: the muon is removed only if it is calo-tagged, else the electron is removed.
2. Electron & non-b-tagged jet. If ∆R <0.2: the jet is removed. 3. Muon & non-b-tagged jet. If ∆R <0.2 and pT(muon)
pT(jet) >0.7: the jet is removed. 4. Non-b-tagged jet & lepton. If ∆R <min(0.4,0.04 +10 GeVpT(`) ): the lepton is removed. 5. Electron & tau. If ∆R <0.1: the tau is removed.
Surviving objects that also pass the signal selection criteria in each category are considered well- measured signal objects used to define the remainder of the analysis selection.