Leptonic asymmetry in ttbar production at CDF

The leptonic asymmetry in semileptonic ttbar decays is measured with the CDF detector using the full Tevatron Run II dataset, which corresponds to 9.4 fb^-1 of integrated luminosity. The measured asymmetry is extrapolated to the full kinematic range and the measured value of A_FB^lep = 0.094^+0.032_-0.029 is compared to the NLO prediction A_FB^lep = 0.038 +/- 0.003.


Introduction
The CDF and D0 experiments have measured the forwardbackward asymmetry A FB for tt production in pp collisions [1,2], the CDF measurement reports A FB = 0.164 ± 0.045, the D0 measurement A FB = 0.196 ± 0.065. Both results are higher than the prediction A FB = 0.088 ± 0.006 [3], which includes both electroweak and QCD next-to-leading order (NLO) corrections. Much effort has been invested to improve the theoretical calculation of the asymmetry in the Standard Model, through the estimation of beyond NLO corrections and related uncertainties. Soft gluon resummation was found to give a negligible contribution [4], electroweak corrections are of the order of 25% and are included in the current predictions [3,5], a calculation based on the Principle of Maximum Conformality (PMC) for the scale setting reports a 40% enhancement, and finally a more realistic estimate of scale uncertainties at NLO of the order 30% should be considered [3,6,7].
The leptonic asymmetry, defined as where q is the lepton charge and y l the lepton rapidity in the laboratory frame, is an observable related to A FB which can provide complementary information. The measurement of A lep FB depends only on lepton charge and direction, and therefore can be measured very precisely. The D0 experiment reported measurements of the leptonic asymmetry in tt production both in the semileptonic and in the dilepton channels with about half of the full Tevatron Run II dataset, the combined result is A lep FB = 0.118 ± 0.032 [8]. There are two physical origins of leptonic asymmetry A lep FB , the A FB asymmetry, and the polarisation of the tt system. Leptons partially inherit the asymmetry of the parent tops, and in addition the V-A coupling of the weak interaction connects the direction of the top decay products to the polarisation of the top quarks. Top pairs are a e-mail: stefano.camarda@desy.de produced unpolarised in the Standard Model, an excess of right handed top pairs would enhance A lep FB , while lefthanded pairs would induce a negative contribution.
The relationship between the top asymmetry, the tt polarisation and the leptonic asymmetry has been the subject of recent theoretical work in the context of possible explanations of the top asymmetry A FB [9,10].

Physics models and expected asymmetry
In the measurement of A lep FB several reference models and corresponding Monte Carlo samples are used, they are listed in table 1.
All the Monte Carlo samples are showered with pythia [11] and processed with the full CDF detector simulation. alpgen [12] is a LO matrix-element matched to PS generator which predicts no asymmetry, powheg [13] is a NLO generator which predicts a small asymmetry, OCTET A, L and R are axigluon models simulated with madgraph [14] which predict A FB comparable to the measured values, but different tt polarisation and different values of A lep FB . The two polarised models, Octet L and Octet R, are light (200 GeV/c 2 ) and wide (50 GeV/c 2 ) axigluons. Octet L has a left-handed coupling and negative polarisation, while Octet R has a right-handed coupling and positive polarisation. Octet A is a massive (2.0 TeV/c 2 ) and narrow axigluon with unpolarised couplings.
A Standard Model NLO QCD fixed order calculation of A lep FB including electroweak corrections reports 0.038 ± 0.003 [3]. When comparing the NLO fixed order result to the prediction from a NLO generator interfaced to parton shower, an important difference has to be considered. The first non trivial orders for the numerator and denominator of equation (1)

Event selection ad sample composition
The full CDF Run II dataset, corresponding to an integrated luminosity of 9.4 fb −1 is used to measure A lep FB . The tt semileptonic events are selected with high-p T electron or muon and large missing E T triggers. Jets are reconstructed with the jetclu cone algorithm in a radius R = 0.4. Events are selected with exactly one lepton with p T > 20 GeV/c and |y l | < 1.25, missing E T > 20 GeV, at least 4 jets with |η| < 2.0, at least 3 jets with E T > 20 GeV, at least one jet with E T > 12 GeV, at least 1 b-tagged jet, and H T > 220 GeV. After the event selection the sample is mainly compose by tt events, with the main background coming from W+jets events.
Background processes are expected to contribute a nonzero asymmetry, in particular the largest background, namely W+jets, is asymmetric for a combination of electroweak and PDF effects. In order to validate the modelling of the leptonic asymmetry in the background simulation a background-enhanced control region is defined requiring that none of the jets is identified as a b-tagged jet. Figure 1 shows the signed rapidity distribution qy l in the control region, where the W+jets background is simulated with the alpgen+pythia Monte Carlo. The observed leptonic asymmetry of 0.076 in the background-enhanced region is in good agreement with the expected value of 0.062.

Extrapolation to the full kinematic region
The signed rapidity distribution qy l is measured in the limited range |y l | < 1.25, which corresponds to the detector acceptance. In order to extrapolate the measurement to the full kinematic space, N(qy l ) is decomposed into symmetric and asymmetric components:

Results
The largest systematic uncertainty on A lep FB is associated to the background subtraction. The background uncertainty is evaluated with a pseudo-experiment technique which accounts simultaneously for the uncertainty on the normalisation of the backgrounds and for the uncertainty on the shape due to limited statistics of the Monte Carlo samples.
Another important source of uncertainty comes from the modelling of the tt recoil due to QCD radiation. The presence of radiated jets is strongly correlated with both A FB and the p T of the tt system. Colour predominantly flows from an initiating light quark to an outgoing topquark and from an anti-quark to an anti-top. As a consequence, events with larger difference between initial state quark and top directions are associated with harder QCD  radiation. Events with more radiation have a larger acceptance because can more easily pass the high p T selection requirements. The uncertainty on the recoil modelling is estimated comparing the acceptance of the nominal powheg Monte Carlo with two other models, namely pythia and alpgen+pythia. The recoil spectra of both pythia and alpgen+pythia are harder than powheg and give larger results for A lep FB , the uncertainty is therefore single-sided. An additional uncertainty related to the recoil model may arise from the initial-state radiation model in the pythia parton shower. The uncertainty is estimated performing variations of the initial and final state radiation parameters (IFSR), the effect is found to be small.
Other QCD and jets related sources of uncertainties like colour reconnection, parton shower model, and jetenergy-scale, have been estimated. They give a small contribution to the A lep FB unertainty because hadronic jets are used in the measurement only to select the event sample. PDF uncertainties largely cancel between the numerator and denominator in the definition of A  Figure 3 shows the parton level unfolded A(qy l ) measured in the data, compared to the powheg prediction, and the result of fits to both data and Monte Carlo with equation (4). After the convolution of equation (5) Figure 3. Asymmetric component A(qy l ) of the signed rapidity distribution qy l as measured in the data (black points) and compared to the powheg prediction (green). A hyperbolic tangent fit to the data and to the prediction is shown as a smooth curve of same colours. The dark (light) grey bands shows the statistical (total) uncertainty on the fit result. The measured value of A lep FB is in good agreement with the D0 measurement A lep FB = 0.118 ± 0.032, and can be compared to the fixed order NLO QCD+EW prediction 0.038 ± 0.003 [3].