We measure the forward-backward asymmetry of the top quark--antiquark pair events (\(\afbtt\)) in the dilepton final state with the full
CDF run II data, corresponding to an integrated luminosity of \(9.1~\fbm\). The inclusive \(\afbtt\) is \(\afbtt = 0.12 \pm 0.11 (\stat) \pm 0.07 (\syst) = 0.12 \pm 0.13\).
This result is consistent with the NNLO standard model expectation of \(\afbtt = 0.095 \pm 0.007\) and the CDF measurement in the lepton + jets
final state of \(\afbtt = 0.164 \pm 0.047\). We also measure the \(\afbtt\) as a function of the rapidity difference of the top pairs (\(\dy\)).
The results are \(\afbtt(|\dy|<0.5)=0.12\pm0.33(\stat)\pm0.20(\syst) = 0.12\pm0.39\) and \(\afbtt(|\dy|>0.5)=0.13\pm0.13(\stat)\pm0.11(\syst)=0.13\pm0.17\),
which can be compared to the predictions from the Powheg Monte Carlo at 0.017 and 0.081, respectively.
The slope of the \(\afbtt\) vs. \(|\dy|\) is estimated to be \(\alpha=0.14\pm0.15\), consistent with the NNLO SM prediction.
Optimization
Fig. 1: Optimization point chosen based on expected statistical + background systematic uncertainty.
Top reconstruction performance derived from Powheg MC
Fig. 2: Distribution of \(\dy\)(reconstructed)-\(\dy\)(generated). Fig. 3: \(\dy\) smearing matrix. Fig. 4: Efficiency of the four \(\dy\) bins, as a function of the parton-level \(\afbtt\).
Unfolding validation
Fig. 5: \(\afbtt\) measured vs. \(\afbtt\) generated with reweighted Powheg. All points are correlated. Good agreement. Fig. 6: \(\afbtt\) measured vs. \(\afbtt\) generated with reweighted Powheg, LO SM and benchmark BSM samples. All the reweighted Powheg points are correlated.
Data
Table 1: Expected number of events in data along with the observed number of events, passing all top dilepton event selections and top reconstruction quality selections. Fig. 7: Distributions of \(\pttt\), \(\pztt\), and \(\Mtt\) from data compared with SM expectations. The agreement is good.Fig. 8: \(\dy\) distribution from data compared with SM expectations. Fig. 9: Posterior of \(\afbtt\) from data. A Gaussian fit is performed to extract the result. Table 2: Table of uncertainties for the \(\afbtt\) measurement. Result
Fig. 10: A comparison of all top \(\afb\) results from the Tevatron with the NLO and NNLO SM predictions.
\(\afbtt\) vs. \(\dy\)
Fig. 11: \(\afbtt\) measured vs. \(\afbtt\) generated in two \(\dy\) bins with reweighted Powheg. Fig. 12: Posterior probability distribution of \(\afbtt(|\dy|<0.5)\) and \(\afbtt(|\dy|>0.5)\). Gaussian fits are performed to extract the results. Fig. 13: Two-dimensional posterior probability distribution of \(\afbtt(|\dy|>0.5)\) vs. \(\afbtt(|\dy|<0.5)\), with statistical uncertainties only. The correlation between the two observables (statistical uncertainty only) is estimated to be -0.44.Table 3: Table of uncertainties for the \(\afbtt(|\dy|<0.5)\) and \(\afbtt(|\dy|>0.5)\) measurement. Fig. 14: \(\afbtt\) vs. \(\dy\). The inner error bars represent the statistical uncertainties, and the whole error bars represent the total uncertainties.
The data points are placed at the bin centroids predicted by the Powheg MC. The blue line with corresponding filled band represents a linear fit to the two data points with
zero intercepts. The slope \(\alpha\) is estimated to be \(0.14\pm0.15\). Result