Measurement of &sigmatt in the all-hadronic channel using 2.9fb-1 of data

Luca Brigliadori1   Andrea Castro1   Fabrizio Margaroli2  [Contact]

1Bologna University & INFN    2 Purdue University 

The CDF Collaboration 

Abstract

We present here the measurement of the t-tbar production cross section in the all-hadronic channel, i.e. where both W's decay into q-qbar pairs. The measurement is performed using the signal yields derived from a previous top quark mass measurement with in situ measurement of the Jet Energy Scale (JES), referring to about 2.9 fb-1 of ppbar collisions collected with a multijet trigger. The candidate events were selected using a neural network followed by the request for 1 or more b-tags. We run pseudo-experiments on simulated samples to evaluate possible biases and to estimate the main systematic uncertainties. The measured cross section is evaluated for two choices of top quark mass and JES. The first choice refers to Mtop=174.8 +2.7 -2.8 GeV/c2 and ΔJES=-0.3 (the values measured in the all-hadronic channel) and corresponds to σtt=7.2±0.5(stat)±1.0(syst)±0.4(lumi) pb. The second choice refers to Mtop=172.5 GeV/c2 and ΔJES=0 (the reference values for a CDF average over all channels) and corresponds to σtt=7.2±0.5(stat)±1.1(syst)±0.4(lumi) pb.

Introduction

At the Tevatron, top quarks are mainly pair produced in ppar collisions via q-qbar annihilation (85%) and gluon-gluon fusion (15%). According to the Standard Model, the top quarks decay into W bosons and b quarks with BR~1. In this analysis we search for events in which both W bosons decay into quark pairs, leading to an all-hadronic final state. This channel has the advantage of the largest branching ratio, about 44%, and of the fully reconstructed kinematics. The major downside is the huge background from QCD multijet production which dominates the signal by three orders of magnitude even after the application of the specific top multijet trigger. A sophisticated event selection based on kinematical and topological variables, followed by the request of identified b-jets is thus needed in order to further improve the signal to background ratio (S:B). In this document we recur to the results of the 2-dimensional Template Method (TMT2D) technique which we use for the measurement of the top quark mass using about 2.9fb-1 of data, which provided not only a measurement of the top quark mass and of the JES, but also the amount of t-tbar signal seen in the data. The displacement, ΔJES, of the jet energy scale from its nominal value is measured in units of total energy uncertainty σc. We refer to this web page and to this document for a detailed description of the method and of the results.
Using the signal yields from the TMT2D analysis, and accounting for the signal efficiency of the corresponding selection, we derive here a measurement of the top-antitop production cross section. Since the efficiency depends strongly on the value assumed for the top quark mass and JES we present here the measurement for two points of interes: one corresponding to the all-hadronic measurements themselves (Mtop=174.8 +2.7 -2.8 GeV/c2 and ΔJES=-0.3) and one useful for a CDF combination with the other channels (Mtop=172.5 GeV/c2 and ΔJES=0).
As done for the mass measurement, the estimate of biases and of the systematic uncertainties are performed running simulated experiments (pseudo-experiments).
A detailed description of this cross section measurement can be found in this document .

Summary of the mass results

We summarize here all the relevant numbers obtained while measuring the top quark mass in the all-hadronic channel.

Top reconstructed invariant masses for signal events with exactly one b-tag (left) or at least 2 b-tags (right).

Cross section likelihood

The measurement of the t-tbar production cross section is performed through a likelihood fit using the following likelihood:
Excl. SecVTX
where each term is written as:
Excl. SecVTX
The likelihood terms for the integrated luminosity, LL, and for the efficiencies, Lε, are Gaussian. The term corresponding to the signal yield, Lns, is also Gaussian but can be written in terms of the cross section as :
Excl. SecVTX

Calibration

We investigate for possible biases in our measurement using pseudo-experiments where we start from an assumed cross section σttin, obtain the coresponding expected signal yields, feed them into the pseudo-experiment generation, perform the mass fit and obtain fitted values for the signal yields and derive finally a fitted cross section, σttfit. The two values for the cross section do not differ by much but we introduce, however, a calibration constant, which amounts to about 0.982, to multiply our results by, in order to obtain an unbiased measurement.

Systematic uncertainties

Various sources of systematic uncertainty affect this measurement. The majority of them has been studied already for the mass measurement (see this web page ), while some new terms are specific to the cross section measurement.
We have considered in detail the effects of:

• the calibration procedure
• the Monte Carlo modeling of the hadronization (Alpgen+HERWIG vs Alpgen+PYTHIA)
• the uncertainty on the BR(W into hadrons)
• the radiation in the initial or final state
• the choice of parton distribution functions (CTEQ vs MRST)
• the difference in energy scale of b-jets w.r.t. generic jets
• uncertainties on the b-tagging efficiency
• different modeling of the event pileup in data and in Monte Carlo simulations
• different color reconnections schemes
• background normalization and shape of the background mass distributions
• reduced statistics of the samples used to derive the mass p.d.f.'s
• the dependence of the efficiency on the JES
• residual JES effects on the signal yields and on the efficiencies
• the primary vertex distribution simulation
• the trigger simulation

All the corresponding systematic uncertainties are presented in the following table, where all terms, but the last two, do not depend strongly on the chosen top quark mass or JES. For the JES term the difference lies in the JES range: -0.9÷0.3 for the AH-TMT measurement and -1÷1 for the CDF-COMB one.
The total relative systematic uncertainty amounts then to about 14% or 15% for the two cases.

Cross section measurement

We perform the likelihood fit on the signal yields, applying the corresponding signal efficiencies and calibration factor



and obtain the following values for the cross section:

σtt = 7.2 ± 0.5(stat) ± 1.0(syst) ± 0.4 (lumi) pb)

for an assumed Mtop=175 GeV/c2 and ΔJES=-0.3, and

σtt = 7.2 ± 0.5(stat) ± 1.1(syst) ± 0.4 (lumi) pb

for an assumed Mtop=172.5 GeV/c2 and ΔJES=0.

Conclusions

After the kinematic selection with a cut on the neural network output, Nout > 0.90 (Nout>0.88), and on the mass reconstruction fit, χ2 <6 (χ2 <5) for events with 1 tag (≥ 2 tags), we are left with 3452 (441) events with 1 b-tag (≥2 b-tags). The expected background, corrected for the contribution due to t-tbar events, amounts to 2785 ± 83 (201 ± 29) events with 1 tag (≥ 2 tags).
The mass measurement procedure gives an amount of signal events corresponding to 643±80 with 1 b-tag and 216 ±25 events with 2 or more b-tags. From these numbers we derive a t-tbar production cross section whose value depends on the top quark mass and the value of JES. For this reason we quote here two values:

σtt = 7.2 ± 0.5(stat) ± 1.0(syst) ± 0.4 (lumi) pb

for an assumed Mtop=175 GeV/c2 and ΔJES=-0.3, and

σtt = 7.2 ± 0.5(stat) ± 1.1(syst) ± 0.4 (lumi) pb

for an assumed for Mtop=172.5 GeV/c2 and ΔJES=0. These measurement can be compared to the published one, in the same channel, corresponding to 1.02 fb-1 based on the assumptions for Mtop=175 GeV/c2 and ΔJES=0:

σtt = 8.3 ± 1.0(stat) +2.0-1.5(syst) ± 0.5 (lumi) pb


With respect to the previous measurement, we have here improved the statistical uncertainty even more than what granted by the increase in integrated luminosity, while the systematic uncertainty has been reduced by a smaller amount. In fact the present measurement has not optimized as the outcome of a simple counting experiment but is the result of a joint (though not simultaneous) measurement of top quark mass and cross section. At the price of adding more sources of systematic uncertainty we have improved on the characterization of the signal candidates we base our cross section measurement upon.

Created on July 10th, 2009.