Search For Single Top Quark Production
Optimized for s-channel Production
With 2.7 fb-1 CDF Run II Data
Authors : Koji Nakamura, Shinhong Kim for CDF Collaboration.
Result
Documentation
cdf public note 9460
Abstract
We present a s-channel-optimized search for single top quark production using 2.7 fb-1
of data accumulated with the CDF detector. This analysis reports the first result of the s-channel
optimized search. We are using events with one high-PT lepton,
large missing ET and two b-quark-identified jets using secondary vertex tagger(SecVtx).
In this analysis we have developed a kinematics fitter and a likelihood-based separator between signal and background.
We are still working to improve sensitivity using the same luminosity data.
So we didn't calculate s-channel sensitivity and didn't set a limit.
|
Event Selection and Expected Number of Events
Candidate events for this analysis are selected by requiring a W + 2jets events topology where W decays leptonically,
W→eνe or W→μνμ. Both of the two jets should be identified as a b-jet
using SecVtx or JetProb tagging algorithm. Following table lists the expected event yield for W + 2-btagged jets events
and W + 2-btagged jets + 1 jet events. We are using only W + 2-btagged events for this analysis.
|
Kinematics Fitter
There are two significant ambiguity in calculating the reconstructed Mlνb. One comes from the uncertainty on
assigning a b-jet correctly to the b-jet from top quark decay. The other is the neutrino energy. Transverse energy of the
neutrino can be measured indirectly as the missing transverse energy, but neutrino momentum z-component can not be measured.
The purpose of the kinematics fitter is to evaluate the neutrino momentum and find the right b-jet assignment.
|
Analysis Method
In order to discover single top production, and to measure its rate with a highest precision, we must take advantage of as many
differences between the signal and the background as possible. To this end, a variety of quantities which can be computed from
the reconstructed event variables have been investigated for their ability to separate the signal from the background.
No single variable encodes all conceivable separation, and so a likelihood is proposed to combine several variables together into
one discriminant to compute the cross section limits or to discover the signal.
The Likelihood is constructed by first forming template histogram of each variable, separately for the signal and for the several
background, denoted Pmifor variable i for the background event class m and signal. Since about 75% of the background
are W + HF and ttbar, we used only W + HF and ttbar for the background template. Likelihood is defined as :
We will list the input variables shape comparison between expected and observed as following :
|
|
|
Invariant mass of the reconstructed top-quark.
|
the scalar sum of the transverse energy for all final particles
|
|
|
Di-jet invariant mass.
|
Transverse energy of Di-jet system.
|
|
|
Transverse energy of leading jet.
|
Missing transverse energy.
|
|
Likelihood output of Kinematics fitter.
|
---|
Systematic Uncertainty
The estimation of systematic uncertainties on the signal and background predictions are key ingredients in procedure
to search for single top production and the measurement of the cross section. As the systematic uncertainty,
we consider uncertainties in the predicted rate of signal and background processes, and uncertainties in the shapes of
the histogram templates. Here shows summery of the systematic uncertainties.
|
Result
Likelihood discriminant distribution
Cross section Measurement
The result of the binned maximum likelihood fit is shown below. All sources of systematic uncertainties (normalization and shape) are included in the result.
To evaluate singletop cross section, we combined our analysis and another Likelihood analysis
which is optimized to t-channel.
We fit single top s-channel cross section and t-channel cross section simultaneously.
For the category which have W+2jet and both of 2 jets identified as b-jet, our Ls discriminant are used,
and for the others category, Lt discriminant are used.
Figure below shows best fit cross section and it's 1σ and 2σ uncertainty as a two dimensional distribution.
Best fit cross sections for the s-channel and t-channel are 0.9~pb and 1.2~pb respectively.
|
Best Fit Cross Section : σs = 0.9 pb, σt = 1.2 pb.
|