Method
This analysis presents a measurement of the top quark mass in the dilepton channel using 2.8 fb-1 of data. The top quark mass is measured using only the lepton's PT information. The leptons' PT is a variable that can be measured very well in the tracker and the calorimeter and can be accurately calibrated against Z→dilepton decays. Jets have a minimal involvement in this analysis, i.e. only in the criteria used for the event selection. Therefore the top quark mass as extracted through this method, is associated with a low JES uncertainty.
Similar measurements have been presented and approved by the CDF for the top-antitop lepton+jets and dilepton decay channels. It was first implemented for the lepton+jets channel at the low luminosity of 340 pb-1 and again at 2.7 fb-1 using b-tagging, and improving significantly both the statistical and the systematic uncertainties to Mtop = 172.1±7.9(stat)±3.0 (syst) GeV/c2. The first measurement in the dilepton channel used 1.8 fb-1 of data and no b-tagging. This top mass measurement gave Mtop = 156.2 ±20(stat)±4.6 (syst) GeV/c2.
This analysis has been based upon the observation that the leptons' transverse momentum PT is sensitive to the top mass and the dependence is linear.
In Figure 1 the lepton mean PT vs the top mass is shown. The leptons' mean PT is derived from the PT distributions of the mass signal templates generated for different input top masses. In Figure 2 each signal template has been combined to the total background template taking into account that the purity of the total sample, as calculated for Mtop =175 GeV/c2, is ρ = signal/(signal+background) = 0.94.
FIG 1: Lepton mean PT sensitivity to the top mass form signal only PT distributions, where b-tagged dilepton selection was applied. The sensitivity of the PT of the signal leptons to the top mass is 15 ± 0.5 %.
FIG2: Lepton mean PT sensitivity to the top mass form combined signal and background PT distributions,where b-tagged dilepton selection was applied. The sensitivity of the PT of the signal & background leptons to the top mass is 14 ± 0.4 %
We model the leptons' PT distribution with an analytical function to examine how this function depends on the top mass. We found that such a function can be the product of a Gamma times a Fermi function.
This function models successfully the shape of both the signal and the background PT distributions (Figures 3 and 4). It has two free parameters p, q, where p is related to the expected rate of leptons with the average PT and q can be interpreted as the expected average PT per lepton.
FIG 4: (left) Fit of the Gamma x Fermi function to the PT distribution of the signal MC leptons generated for Mtop = 175 GeV/c2
FIG 5: (right) Fit of the Gamma x Fermi function to the PT distribution of the total background
The top mass is measured by employing the likelihood minimization procedure. The Gamma x Fermi function is used as a probability density function (p.d.f).