Kinematic fit of associated single top-quark and W boson production and its application in a neural-network based analysis at ATLAS

The Wt channel is predicted to give the second largest contribution to the total single top-quark production cross section at the LHC, while at the Tevatron its cross section is too small to be measured. The Wt channel is characterised by the associated production of a top-quark and an on-shell W boson. In the lepton plus jets decay channel this leads to one b-quark jet, two light-quark jets, one charged lepton and a neutrino in the final state. One general problem of Wt channel analyses is that it is very hard to find variables that provide good separation from background, in particular from top quark pair production. The analysis strategy of the Bonn single top group is to employ a neural network to extract the signal. In this talk a kinematic fit to the signal topology in order to construct some variables that are directly sensitive to how much an event looks like Wt signal is presented. The implementation of the fit using the KLFitter package is discussed. A neural network taking solely output from the kinematic fit as input variables is used to demonstrate that these are indeed useful for analysis.