Measurement of ttbar production cross section using dilepton events in 4.47 fb-1

This analysis has been documented in public CDF note 9890 by Alexei Varganov and edited by Dan Amidei


We report on a measurement of the top cross section using candidates events selected in dilepton channel with each lepton identified as electron or muon. We evaluate backgrounds using normalized simulation samples. For selection with b-jet identification we calculate backgrounds relying on tagging matrix prediction applied to normalized pretag samples. In a sample of about 4.47 fb-1 of Run II data collected with CDF we obtain

σpre = 6.56 ± 0.65stat ± 0.41syst ± 0.38lumi pb.

for 215 candidate events before applying loose SecVtx tagging, and

σtag = 7.27 ± 0.71stat ± 0.46syst ± 0.42lumi pb.

for 119 candidate events containing at least one tagged b-jet.

Analysis Highlights

The samples used in this analysis were collected from high-PT central electron and muon data. The event selection requires a pair of oppositely charged isolated leptons passing ET ≥ 20 GeV kinematic cut. Each lepton is reconstructed electron or muon object from central or plug region. Selected jets must pass |η| < 2.5 geometric and ET ≥ 15 GeV kinematic cuts. Finally, events passing Drell-Yan veto should have MET ≥ 25 GeV and at least on jet must have ET ≥ 30 GeV

Jet Multiplicity Spectrum
for candidate events
Number of SecVtx tags in
Njet ≥ 2 candidate sample

Lepton selection validation is carried out using very pure high statistics sample of dielectron and dimuon events near Z pole. Measuring Drell-Yan cross section validates leptonic efficiencies and scale factors. In addition, checking electron-muon channel provides a thorough check for our physics model. In particular, we test the QCD and fakes background which is described using same charge dilepton events.

Dilepton invariant mass spectrum
for electron-muon events with Njet ≤ 1
Transverse missing energy
for electron-muon events with Njet ≤ 1
Sample composition for pre-tag signal events.

To get the signal sample composition we apply the tagging model to the high MET pretag background events with Njet ≥ 2. We count tags directly for top, ZZ, and fake sample and we use tagging matrix prediction for Drell-Yan, WW, and WZ events.

Sample composition for tagged signal events.

Systematic Uncertainties

Summary of systematic uncertainties when using pretag selectionSummary of systematic uncertainties when using selection with b-tag

These measurements became public CDF result on August 6, 2009.