SOBRE EL MODO DE SONARSE PARTE I: TEXTOS

In the SM three production modes are available for single top events, distinguished by the virtuality of the W boson coupled to the top (see fig. 6.72)

Just recently, the D0 experiment gave evidence of single top events [36], but LHC will provide much higher statistics for all the three channels (the production of single top quarks will account for a third of the top pair production), allowing the observation also of the Wt production mode, and a more precise study of the single top phenomenology. The study of single top production provides a unique possibility to investigate some aspects of top quark physics that cannot be studied in t¯t production. In particular, the only way to measure directly Vtb(CKM matrix element), to investigate the tWb vertex structure and the FCNC

coupling directly in the production processes, and to search for possible manifestation of New Physics beyond SM such as anomalous couplings and s-channel resonances. Moreover, the single top quark production presents an irreducible background to several searches for SM and New Physics signals (for example Higgs boson searches in the associated production channel) and may provide additional measurements of Mtop and of the top

Figure 6.72: Main graphs corresponding to the three production mechanisms of single-top events: (a) t-channel (b) Wt associated production (c) s-channel.

rate at the LHC is also calculated in the SM to the NLO level of accuracy for all three production mechanisms. The computed NLO cross-sections for the t-channel, the tW and the s-channel are respectively 240, 60 and 10 pb. The three single-top processes result in quite distinct final states, leading to the definition of specific analyses in each case, making use of differences in jet multiplicity, number of b-tagged jets required, as well as angular distributions between lepton and/or jets present in the final states. Besides, important differences subsist in the level of backgrounds that are faced in the various analyses, leading to the development of tools dedicated to the rejection of specific backgrounds.

Similarly to the situation at the Tevatron, the selection of single top events will suffer from the presence of both W+jets andtt¯background, which are produced at much higher rates. Thus, careful approaches devoted to the understanding of these backgrounds in terms of shape and normalization performed directly from data will have to be defined. Besides, single top analyses will be very early dominated by the systematic uncertainties, and will require a good control of b-tagging tools and a reliable determination of the jet energy scale.

t-channel

CMS has performed a study for 10 fb−1 _{of integrated luminosity, with the pile-up expected}

for a luminosity of 1033_{. They assume to extract only the cross-section, with a simple}

counting experiment and without the use of any multivariate methods [37]. The generators which have been used for the signal are: SingleTop [38] and TopRex [39]. ATLAS explored the case of 1fb−1 _{of integrated luminosity, with no pileup [40]. A cut-and-count analysis}

consitutes a baseline; more complex multivariate methods have been developed in addition to get a better background rejection. The AcerMC Monte Carlo has been employed to generate the signal events. ATLAS made use of the fact that there are similar features in the three channels: a common pre-selection is therefore possible to reduce backgrounds. This pre-selection requires exactly one isolated high PT lepton, from 2 to 4 jets, one of

9-10% (10-12 %) for electron (for muons). With these cuts, the rejection of W+jets is of order O(104_{), while for t}¯_{t is O(20). As shown in fig. 6.72 (left graph) for the t-channel,}

the final partons (b-quark from top-quark decay, the charged lepton and light quark) have relatively large transverse momenta. However, an additional b-quark is produced with small transverse momentum. This will make very difficult to identify the low PT jet

originating from this quark and tag it as b-jet. Another specific feature of the t-channel single top events is the production of a light jet in the forward/backward direction. A cut on b-tagged jet pT >50 GeV reduces the W+jets significantly, while a cut on the hardest

light jet _|η_| > 2.5 can reject t¯t events. With this simple cut and count analysis a S/B value of 0.37 is reached. The statistical error on the cross-section measurement is around 5%, while the systematics (b-tagging, JES scale, ISR/FSR) reach 44.7%. The left plot in fig. 6.73 shows the number of jets for single top candidates in the t-channel and for the relevant backgrounds. By applying a more sophisticated multivariate analysis (Boosted Decision Tree), this last one can be reduced by a factor of 2 about.

Wt channel

From the theoretical point of view the definition of the Wt signal is not trivial, since at NLO it mixes with tt¯. The final state is very similar to t¯t production, except for the presence of one less b-jet: jet counting is therefore critical. Since it is not possible to achieve a good S/B, a correct background normalization from data will be important, to avoid large systematic uncertainties. CMS selects the events by requiring exactly one lepton (e or µ), one b-jet and two light quark jets, and missing ET. The correct (Wb)

pairing is obtained from a Fisher discriminant using variables like the PT(b+W), ∆R(W,b)

and the product of the b-quark and W charges. s-channel

The identification of s-channel events will be much harder at LHC than at Tevatron, as the relative cross-section is much smaller. The CMS selection requires one isolated lepton (e orµ), exactly two jets, both b-tagged, missing ET, and cuts on the transverse mass of

the reconstructed W, on Mtop, on PT(top), on ΣT and on HT. The uncertainty which can

be reached on a cross-section measurement with 10 fb−1 _{of data, is of 18% (statistical)}

and 31% (systematics), not including the error coming from the luminosity measurement. The right plot in figure 6.73 shows the reconstructed mass for single top candidates in the Wt channel and for the relevant backgrounds (CMS)

In a context of low S/B, the use of sophisticated tools like likelihoods and Boosted Decision Trees, appears very useful if one wants to reach evidence of the signal with the early data or to determine precisely their cross-section. These techniques, which are now of common use at the Tevatron, will require the use of reliable event samples for modeling signal and backgrounds, that will presumably be produced from the data. The analyses should also be optimized with respect to the total level of systematic uncertainty, which will be the main limiting factor for 30 fb−1 _{measurements. Finally, a precise determination}

of single top cross-sections can be achieved for a few fb−1 _{in the t-channel and the Wt-}

channel , while for the s-channel, higher statistics will be required. Their interpretation in terms of new physics should thus come at a later stage, once the systematic effects are under control.