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

128 CHAPTER 6: Conclusionwhich are imposed on the system. In this thesis, the e+jets decay channel of t¯t eventsis chosen as the signal t¯t → W ¯Wb¯b → eν e q¯qb¯b, which contains an isolated electron thatcan be used to suppress the QCD multijets background and at least four reconstructedjets where two of them originate from heavy quarks. Several cuts are applied to rejectthe huge backgrounds and to select the signal events which were discussed in detailin Chapter 4. Having applied the event selection cuts listed in Table 4.6 on a sampleof simulated events corresponding to 100pb −1 of integrated luminosity, there are about308 e+jets t¯t events remained resulting in a signal over background ratio of about 2.5.As there are at least four reconstructed jets in the final state of e+jets t¯t events,which are indistinguishable if no tagging tools are used, a first task is to label thereconstructed jets and associate them to the hard scatter partons. This can be donewith the use of Multi-Variate Analysis techniques (MVA). A Likelihood Ratio methodis used to return a chosen jet combination based on the information which is obtainedduring the training of the MVA method. The chosen jet combination by the MVAmethod is not all the time the true combination which represents the combinationwhere jets are matched to the partons. The performance of the MVA method stronglydepends on the input variables which are chosen for the training. A full discription ofthe MVA method and a list of variables used for the training of the Likelihood Ratiomethod were given in Chapter 5. It was found that in around 17% of the selected e+jetst¯t events, the Likelihood Ratio method is able to return the correct combination wherethe three reconstructed jets match to the quarks from the hadronic top quark decayt → Wb → q¯qb. This is a great improvement compared to the random choice of thecorrect jet-parton combination which is equal to 8%.Considering the hadronic branch of the signal t → Wb → q¯qb, the masses of theW boson and the top quark can be used to constrain the system by requiring thefour-vectors of the reconstructed jets of the chosen jet combination to fulfill the massconstraints. The procedure of applying constraints is performed with the use of akinematic fit, which was explained in detail in Chapter 5. The t¯t events are excellentcandidates for the jet calibration purpose because, in addition to the high rate of toppair production, they contain both the light and the b jets in their final state whichyields to obtain calibration factors for different flavours of the quarks. Starting withthe three jets in the hadronic branch of the selected event t → Wb → q¯qb, which arenow assigned to the hard scatter partons by means of the MVA method, the energyof the jets are altered within a window of ±40% around the reconstructed energies.Changes are made in steps of 2%. As a result, a two-dimensional space is spannedin the directions of the light jet energy scale calibration factor ∆E l and the b jetenergy scale calibration factor ∆E b . For each selected event and in each point of thetwo-dimensional grid, the kinematic fit returns a probability P KinFit (∆E l , ∆E b ), whichreflects how likely the hypothesis of the three jets to originate from the top quark in thechosen jet combination, is true. The probability of the kinematic fit can be transformedinto a χ 2 (∆E l , ∆E b ) space. The information of all the selected events is combined bythe sum of the χ 2 distribution of the individual events, which makes a two-dimensionalparabola. The resulted parabola can be projected in each of the two dimensions andthe minimum of each of the two projected parabolas, refers to an estimation of theresidual jet energy scale calibration factors for both light and b quark jets.