Measurement of the rate of top quarks decaying to tau leptons

Authors:


S. Demers, K. McFarland, A. Vaiciulis
University of Rochester



Abstract:

We measure the rate of top quarks decaying to a W boson plus a b quark with the W decaying to a hadronically decaying tau lepton and compare that rate to what is predicted by the Standard Model. We predict 2.3 events (including signal plus background) in 193 /pb of Run 2 data and we see 2 events. We place a 95% confidence level upper limit of 5.0 on the ratio of top quark to tau lepton decays observed over what is predicted by the Standard Model.  Our result is

r(tau) = (observed BR top->tau) / (SM BR top->tau) < 5.0
at 95% confidence level

If the top to tau branching ratio is predicted correctly by the Standard Model then r(tau) would equal 1.0. Our result is therefore consistent with the Standard Model.

The analysis and results were blessed at the 28-APRIL-2004 top meeting.



Using W->taunu to understand tau ID:


We looked in 58 /pb of Run2 data for a pure sample of taus using the physics process W->taunu. We compared the number of W->taunu events in the data with the number predicted by Pythia W->taunu Monte Carlo. The following plot shows the track multiplicity distribution for the tau candidates. The tau signature can be seen in the 1 and 3 track bins, which correspond to 1 and 3 prong tau decays:

Track Multiplicity Distribution
(left-click to bring up the EPS version of the image)
(right-click to save the GIF version of the image)


We also compared the tau candidate pt spectrum in the data with the tau candidate pt spectrum in the Monte Carlo:

Tau Candidate Pt Spectrum
(left-click to bring up the EPS version of the image)
(right-click to save the GIF version of the image)





Systematic Uncertainties:

Our systematic uncertainty is dominated by the uncertainty on the tt acceptance shown below:

Source
Uncertainty (%)
Electron/Muon ID
5.0
Tau ID
10.5
Jet Energy Scale
5.8
ISR
7
FSR
7
PDFs
1
MC Generator
7
Total
17

Backgrounds:

The largest background in the analysis is due to jets faking taus. We measure the rate of jets to fake taus in several different data sample and show that they agree with each other to the level of 26%. Shown is the relative fake rate that jets matching to tau candidates passing a subset of tau ID cuts will pass all tau ID cuts:

Jet to Tau Relative Fake Rate
(left-click to bring up the EPS version of the image)
(right-click to save the GIF version of the image)



We show the Background Summary and Expected Signal in the e,tau channel (top table) and muon, tau channel (lower table):

Background Summary Tables
(left-click to bring up the EPS version of the image)
(right-click to save the GIF version of the image)



As a check of our analysis we remove some of our final analysis cuts and look at the number of events we predict and expect as a function of the number of jets in the event (0jet, 1 jet, 2 or more jets) and the channel (electron,tau or muon,tau channel) and the charge distribution (OS = the electron or muon is the opposite charge of the tau candidate, SS = the electron or muon is the same charge as the tau candidate.):

Category
Predicted
Seen
0jet, e, OS
24.3 +/- 3.6
17
1jet, e, OS
4.8 +/- 0.9
5
2jets+, e, OS
2.7 +/- 0.4
8
0jet, muon, OS
21.7 +/- 3.3
11
1jet, muon, OS
2.8 +/- 0.6
4
2jets+, muon, OS
2.2 +/- 0.5
1
0jet, e, SS
7.3 +/- 1.8
8
1jet, e, SS
1.9 +/- 0.6
3
2jets+, e, SS
1.1 +/- 0.3
4
0jet, muon, SS
5.6 +/- 1.5
3
1jet, muon, SS
0.8 +/- 0.3
0
2jets+, muon, SS
0.6 +/- 0.2
0


Results:

Separating the result into the electron and muon channels we show our observed number of events:

Sample
electron
muon
Background
0.77 +/-0.12 +/-0.13
0.53 +/-0.08 +/-0.08
tt
0.59 +/-0.05 +/-0.10
0.47 +/-0.04 +/-0.07
Data
2
0


We show one of our two events in a lego plot below, where the calorimeter energies are plotted by calorimeter tower in the CDF Detector. The objects (electron, tau and jets) are labled:

probability of rtau
(left-click to bring up the EPS version of the image)
(right-click to save the GIF version of the image)



Our Probability distribution for the value r(tau) (defined above) is shown. The most likely r(tau) value is 0.8.
We exclude r(tau) > 5.0 at the 95% confidence level.


probability of rtau
(left-click to bring up the EPS version of the image)
(right-click to save the GIF version of the image)




Further details:

More information can be found at the documentation/Q&A page for this analysis (restricted to CDF), or by contacting any one of the good people listed at the top of this page.



Last updated: 06-MAY-2004