Measurement of the Jet Energy Scale in the CMS experiment ... - IIHE

CHAPTER 5: Estimating of the Jet Energy Scale Calibration Factor 89combinatorial backgrounds. In order to reduce the contribution of the combinatorialbackgrounds, additional event selection cuts are applied.In the topology of the e+jets t¯t event, there are four quarks produced in the final statewhich subsequently contribute to at least four reconstructed jets in the detector. Inaddition to the four jets which represent the quarks generated in the hard scatteringprocess, any other jet which is reconstructed in the event is the consequence of theradiation of a parton in the initial or final state, generally known as ISR or FSR effect,respectively. As already discussed in Section 5.1.1, most of the e+jets t¯t events, containat least an ISR jet which is among the four leading reconstructed jets. Then thesekind of events, where the four leading jets can not be matched to the four hard-scatterpartons, would contribute to the combinatorial backgrounds. Since there is no truejet-parton matching in such events, the MVA method would always return a wrongjet combination. Therefore, any event with at least one extra jet is vetoed and is notprocessed in the rest of analysis. In addition to the four leading jets in the selectedevents, any other jet with p T > 30 GeV within the tracker acceptance is defined as aradiation jet and the events containing at least one radiation are rejected.In order to further reduce the multiplicity of badly reconstructed events in the finalselected sample, the events are constrained to fulfill some kinematic requirements. Allevents which pass the radiation veto cut, are fed to the kinematic fit package. Thethree reconstructed jets in the e+jets t¯t event, which are now associated to the partonsin the hadronic branch t → Wb → q¯qb by means of the MVA method, are forced tofulfill the mass constraints as described in Section 5.2. The masses of the W bosonand top quark used in the fit package are taken to be equal to those used in the simulationstep, being 80.4 GeV and 172.5 GeV respectively. The energy of the three jetsassociated to the hadronic top quark are shifted in certain steps. A three-dimensionalspace, which is spanned by the changes in the energies of the three reconstructed jets(∆E l1 , ∆E l2 , ∆E b ), is obtained, where ∆E l1 and ∆E l2 represent the correction factorson the light jets from the hadronic W boson decay and ∆E b corresponds to the changesin the energy of the b jet arrising from the hadronic top quark decay. In each pointin the three-dimensional space, the kinematic fit is performed. The fit is applied on agrid in this space in steps of 2% within a window of ±40% around the reconstructedenergies of both light as well as b jets, individually. As a result, the energy of a genericjet with, for example, E=100 GeV is shifted from E=60 GeV to E=140 GeV in steps of2 GeV and in each step the fit is performed, hence 41 times per jet per event. Since thelight jets arrising from the W boson are indistinguishable, their correction factors canbe assumed to be equal ∆E l1 = ∆E l2 = ∆E l . Therefore the kinematic fit is performedin a two-dimensional space which is obtained by shifting in the energy of the light jets,for which the shifts are taken to be identical, and in the energy of the b jet (∆E l , ∆E b ).As a result, the kinematic fit is performed 41×41 times per event.In each point of the two-dimensional grid, the kinematic fit returns a probabilityP KinFit (∆E l , ∆E b ), which represents how likely, for that particular point of the grid,the hypothesis of originating the three jets in e+jets t¯t event from the decay of the topquark in the topology t → Wb → q¯qb, is correct. At the same time, the fit probabilitystates how much the mass hypotheses are correct for the three jets in e+jets t¯t event