The SR definition is designed to selecttZq candidate events with a trileptonic final state and, at the same time, minimise as much as possible the number of events coming from background processes.
This is done by applying certain selection criteria (cuts) on the kinematic variables of the physics ob-jects corresponding to the final state particles, as well as requirements on the reconstructed intermediate particles (theZand the top quark). Complementary to the SR, by applying specific sets of cuts, valid-ation and control regions (CRs) are defined. CRs are used either for deriving normalisvalid-ation factors or for estimating the non-prompt lepton background. VRs are defined such that they have a small signal contamination and that they are enriched in events coming from the process that is being validated.
For the SR, the selected physics objects are three leptons, one b-tagged jet, one untagged jet and missing transverse momentum. The Z boson is reconstructed by selecting two opposite sign, same flavour (OSSF) leptons. For events in which two such pairs can be identified, the two leptons with the invariant mass closest to theZ boson mass are used for reconstructing theZ boson. The remaining lepton, missing transverse momentum andb-tagged jet are combined for reconstructing the top quark.
The untagged jet is then assumed to be the jet coming from the spectator quark in the hard scattering process. Since it is expected to be emitted at highη, it is referred to as theforward jet.
A summary of the SR definition is given in table 5.2. The first step in the selection is requiring exactly three leptons passing different pT thresholds. Studies have shown that a higher pT threshold for the softest lepton would further increase the signal to background ratio by reducing the contribution fromZ+jets events, but overall this would not bring an improvement in the final result. Out of the three selected leptons, at least one opposite sign, same flavour pair is required.
Table 5.2: Summary of the requirements applied for selecting events in the signal region.
3 leptons with|η|<2.5 andpT>15 GeV pT(`1)>28 GeV, pT(`2)>25 GeV,pT(`3)>15 GeV
≥1 OSSF pair 2 jets,|η|<4.5 1b-jet,|η|<2.5 pT(jet)>30 GeV
|m``−mZ|<10 GeV mT(`W, ν)>20 GeV
Exactly two high-pTjets are selected. These are allowed to be anywhere within the|η|<4.5 range in order to also select jets going in the forward direction. One of the two jets must beb-tagged by passing the 77 % tagging working point and should be in the central region of the detector (|η|<2.5).
Requiring the invariant mass of the two leptons associated to theZboson (OSSF pair) to be within a 20 GeV window aroundMZ =91.2 GeV reduces all background processes that do not have aZboson,
5.4 Signal region definition
such astt andtW.
In order to separate events in which no leptonically decaying W boson is produced, and because neutrino related information is only available in the transverse plane, one can use the reconstructed transverse mass of theWboson,mT(`W, ν). This is defined as:
mT(`W, ν)= q
2pT(`)pT(ν)
1−cos∆φ `, ν.
For processes that include aW →`νdecay, this distribution is expected to peak at values close to the mass of theW boson. This will, however, not be exactly at 80 GeV due to detector effects and the fact that only the transverse information is used.
Selecting events with mT(`W, ν) > 20 GeV, improves the signal to background ratio by rejecting events in which noWboson is produced, especially theZ+jets non-prompt lepton background.
Figure5.6shows how the SR composition changes after applying each of these cuts. Each pie-chart is constructed using the number of SR events after applying the cut listed above the diagram. TheZ+jets contribution is not included because it is estimated using a fully data-driven procedure and hence it is only defined in the region of phase space in which it was derived. This gives a representation of how the signal over background ratio evolves and improves after each cut is applied.
After preselection, the number of tt and diboson selected events is significantly larger compared to the signal. Requiring exactly two jets, one of which has to be b-tagged, helps with reducing the diboson background. The tt contribution is considerably reduced when selecting events with exactly three leptons. The signal contribution becomes visible after that, constituting a significant fraction of the selected events. The requirement for having an OSSF lepton pair with invariant mass close to theZ boson mass visibly enhances the fraction of events coming from processes that include aZboson (such astZqandWZ) by minimising thettcontamination.
Complementary to figure5.6, the efficiency of each cut on the signal and background processes is shown in figure5.7. This is calculated as the number of events passing the respective cut divided by the number of events before applying the cut. Again, the cuts are applied sequentially, e.g. the efficiency of requiring exactly three leptons is calculated relative to the number of events that have already passed the jet related selection criteria. As a rule of thumb, for every selection criteria the efficiency on the signal should be higher than the efficiency of the backgrounds. This would indicate that by applying the cut, a larger fraction of the background is discarded than signal, thus improving the overall S/B. The efficiency in data is shown in black. There are several cuts that have very low efficiency in data, meaning that a lot of the previously selected events will be discarded, however, as long that the signal fraction is improved, this is not a problem. The signal efficiency of the last four cuts is very hight, above 90 %. In the bins in which the signal line is not visible, it is because it overlaps with the highest entry in that bin (e.g. for the “OSOF pair” cut, the signal and diboson efficiencies are equal and very close to 1).
By applying a similar selection as for the SR, CRs are defined for deriving a normalisation factor for the diboson contribution and performing thett andZ+jets non-prompt lepton background estimation.
In addition to that, two dedicated VRs are defined in order to check signal and background modelling.
One of them is enriched intt events and the second one validates the modelling of the diboson back-ground, as well as theZ+jets estimation. The definitions for all VRs and CRs are described relative to the SR definition, in the following subsections.
5 Event selection and background estimation
Preselection N jets = 2 N b jets = 1 N leptons = 3
p T lepton OSSF pair M ll M T W
tZq Diboson ttV + ttH + tWZ tt + tW
Figure 5.6: SR composition evolution after applying selection criteria.
Njets=2
Nb jets=1
Nleptons=3 pleptTon OSSF pair Mll MWT
0.0 0.2 0.4 0.6 0.8 1.0
Cut efficiency
tZqtt + tW ttV + ttH + tWZ Diboson data
Figure 5.7: Efficiency of each SR selection requirement on the signal and background processes.