Search for Single Top-Quarks using Neural Networks (public web page)

Abstract

We report on a search for non-standard-model single top-quark production in the channel u(c)+g ➞ t with CDF II data corresponding to 2.2 fb^-1 of integrated luminosity.

We apply neural networks to construct a discriminant that distinguishes between anomalous single-top-quark and background events and observe no significant excess in collision data. We then set an upper limit of 1.8 pb. Using cross section calculations at next-to-leading order we obtain upper limits on the anomalous coupling constants:

$\text{[math]}$ $\text{[math]}$

Theory graphs and equations

Results
NN output without ano-top		NN output with σ_ano=1.8 pb

The predicted and measured distributions for the case no anomalous top-quark exists. Background templates are normalized to the SM prediction.		The predicted and measured distributions for the case that anomalous top-quark has a cross section of 1.8 pb. Background and signal templates are normalized to prediction.

Upper limit on the anomalous cross section (in red). The black line is the theoretical trend in NLO calculation. As a result, the colored region is excluded from the measurement, which sets an upper limit of 0.025 TeV^-1 for the anomalous coupling constant (gtu-coupling).	The posterior probability density returns the observed limit on the cross section of the anomalous top-quark @ 95% C.L . The limit is 1.8 pb.	Upper limit on the anomalous cross section (in red). The black line is the theoretical trend in NLO calculation. As a result, the colored region is excluded from the measurement, which sets an upper limit of 0.105 TeV^-1 for the anomalous coupling constant (gtc-coupling).

Event Selection

The CDF event selection exploits the kinematic features of the signal final state, which contains a top quark and a bottom quark. To reduce multijet backgrounds, the W boson originating from the top quark is required to decay leptonically. One therefore demands a single high-energetic electron or muon (E_T(e) > 20 GeV, or P_T(μ) > 20 GeV/c) and large missing transverse energy (MET) from the undetected neutrino MET > 25 GeV.
The backgrounds belong to the following categories: Wbb, Wcc, Wc, mistags (light quarks misidentified as heavy flavor jets), top pair production (tt) and SM single top production events (s-channel, t-channel), non-W (QCD multijet events where a jet is erroneously identified as a lepton), Z→ll and Diboson WW, WZ, and ZZ. We remove a large fraction of the backgrounds by demanding the jet present in the event to have E_T> 20 GeV and \|η\| < 2.8. This jet has to be tagged as a b-quark jet by using displaced vertex information from the silicon vertex detector (SVX). The non-W content of the selected electron dataset is further reduced by several requirements to MET, MET significance, transverse W boson mass, and several angles between the MET vector, lepton vectors and jet vectors. The numbers of expected and observed events are listed in the tables below.

Neural Network Input Variables

Using neural networks kinematic or event shape variables are combined to a powerful discriminant.
One of the variables is the output of the KIT flavor separator. The KIT flavor separator gives an additional handle to reduce the large background components where no real b quark is contained, mistags and charm-backgrounds. Both of them amount to about 60% in the W+1 jet data sample even after imposing the requirement that one jet is identified by the secondary vertex tagger of CDF. The following plots show the 14 input variables for the anomalous top-quark network. The plots in the third column show the variables in the "zero-tag" sample (for cross-check).


MC distributions: the mass of the reconstructed top-quark	data - MC comparison: the mass of the reconstructed top-quark	data - MC comparison: the mass of the reconstructed top-quark

MC distributions: the neural network output of the KIT flavor separator for the b-tagged jet	data - MC comparison: the neural network output of the KIT flavor separator for the b-tagged jet

MC distributions: the transverse mass of the reconstructed top-quark	data - MC comparison: the transverse mass of the reconstructed top-quark	data - MC comparison: the transverse mass of the reconstructed top-quark

MC distributions: the transverse mass of the reconstructed W-boson	data - MC comparison: the transverse mass of the reconstructed W-boson	data - MC comparison: the transverse mass of the reconstructed W-boson

MC distributions: the cone built by the charged lepton and the jet	data - MC comparison: the cone built by the charged lepton and the jet	data - MC comparison: the cone built by the charged lepton and the jet

MC distributions: the product of the charge of the lepton and the pseudorapidity of the reconstructed top-quark	data - MC comparison: the product of the charge of the lepton and the pseudorapidity of the reconstructed top-quark	data - MC comparison: the product of the charge of the lepton and the pseudorapidity of the reconstructed top-quark

MC distributions: Δφ between the most energetic jet and the direction of the missing transverse energy	data - MC comparison: Δφ between the most energetic jet and the direction of the missing transverse energy	data - MC comparison: Δφ between the most energetic jet and the direction of the missing transverse energy

MC distributions: the transverse momentum of the jet	data - MC comparison: the transverse momentum of the jet	data - MC comparison: the transverse momentum of the jet

MC distributions: transverse momentum of the charged lepton	data - MC comparison: transverse momentum of the charged lepton	data - MC comparison: transverse momentum of the charged lepton

MC distributions: the rapidity of the reconstructed top-quark	data - MC comparison: the rapidity of the reconstructed top-quark	data - MC comparison: the rapidity of the reconstructed top-quark

MC distributions: the pseudorapidity of the reconstructed W boson	data - MC comparison: the pseudorapidity of the reconstructed W boson	data - MC comparison: the pseudorapidity of the reconstructed W boson

MC distributions: the pseudorapidity of the charged lepton	data - MC comparison: the pseudorapidity of the charged lepton	data - MC comparison: the pseudorapidity of the charged lepton

MC distributions: the aplanarity	data - MC comparison: the aplanarity	data - MC comparison: the aplanarity

MC distributions: Δφ between the charged lepton and the direction of the missing transverse energy	data - MC comparison: Δφ between the charged lepton and the direction of the missing transverse energy	data - MC comparison: Δφ between the charged lepton and the direction of the missing transverse energy

Systematic Uncertainties

Systematic uncertainties can cause a shift in the event detection efficiency for events of different physics processes, but can also cause a change in the shape of the template distributions. The rate uncertainties are summarized in the tables. Below you find six examples of systematic shape uncertainties: jet energy scale (JES), the initial and final state radiation (ISR, FSR) for the single top-quark template, factorization and renormalization scale (Q²) for Wbb events, the modeling uncertainty on the KIT flavor separator output (KIT opt.) and the influence of the nonW flavor composition.


Systematic rate uncertainties for 1 jet and 1 b tag.	Systematic JES down/up rate uncertainties.

Expected Limit

To compute the upper limit on the anomalous production cross section, ensemble tests are used. In this context, an ensemble test consists of a set of pseudo experiments. For each pseudo experiment, first the number of events N_j of each event category is determined by drawing a random number from a Poisson distribution of a mean ν_j. As a result, the pseudo experiment features a total number of Σ N_j events. Systematic uncertainties on the expected background rates are considered by fluctuating the Poisson means ν_j within their uncertainties Δ_j. In a second step, N_j random numbers are drawn from the template distributions of the neural network output for all considered event categories. Those random numbers are filled in a histogram which constitutes the neural network output distribution of a particular pseudo experiment. To obtain a measure for our a-priori sensitivity we perform Monte Carlo experiments without anomalous top-quark events. For each experiment we calculate the upper limit at 95% C.L. (confidence level). We define the median of all upper limits as our sensitivity. We obtain: σ⁹⁵_apriori=1.4 pb. The 16% quantile is 0.9 pb, while the 84% quantile is 2.3 pb. This effect is expected and increases the probability of our observed result (under the assumption there is no anomalous top-quark production).

A-priori sensitivity: the expected limit on the cross section is 1.4 pb (median), the 16% quantile: is 0.9 pb, the 84% quantile is 2.3 pb.

Binned Likelihood Fit to Data


After the expected sensitivity has been determined, the neural network is applied to observed events. At first, the output distributions of observed events are compared to the expected distributions. Finally, the templates are fitted to the observed distributions to determine an upper limit on the anomalous top-quark cross section as shown in Observed Limit.

The predicted and measured distributions for the case no anomalous top-quark exists. Background templates are normalized to the SM prediction.		The predicted and measured distributions for the case that anomalous top-quark has a cross section of 1.8 pb. Background and signal templates are normalized to prediction.

Templates

Fit templates in the pretag sample.	Predicted distribution in the pretag sample.

Fit templates in the tagged sample.

Dominic Hirschbühl, Jan Lück, Thomas Müller, Adonis Papaikonomou, Thomas Peiffer, Manuel Renz, Svenja Richter, Irja Schall, Jeannine Wagner-Kuhr, Wolfgang Wagner
KIT, Universität Karlsruhe

Abstract
Results
Event Selection
Neural Network Input Variables
Templates
Systematic Uncertainties
Expected Limit
Binned Likelihood Fit to Data
Observed Limit
Public Note (pdf)

To view a plot with full resolution in .gif or .jpg format, right-click and select "View Image."


Systematic shape uncertainties: jet energy scale (JES) for the anomalous top-quark template.	Systematic shape uncertainties: initial state radiation (ISR) for the anomalous top-quark template.	Systematic shape uncertainties: final state radiation (FSR) for the anomalous top-quark template.


Systematic shape uncertainties: factorization and renormalization scale (Q²) for Wbb events.	Systematic shape uncertainties: uncertainty due to the influence of the mistag model.	Systematic shape uncertainties: uncertainty due to the influence of the nonW flavor composition.

Observed Limit