Notes from CAT-top physics group meeting 22/9/06 ================================================ Present: Pamela, Richard, Ewa, Maria, Alan, Pippa, Christian, Grant, David, Martin Kinematic fitting in ttbar events - Christian Schmitt --------------------------------- See slides at: http://indico.cern.ch/conferenceDisplay.py?confId=6951 Christian presented the kinematic fit HitFit, originally developed by Scott Snyder and extensively used in D0. It has been ported to the ATLAS environment by Christian, and can be used in Athena, though is not yet integrated as an AlgTool, interfaced to AOD or EventView data models etc. Typically in semileptonic events, the fit has 17 input variables, and 2 (W masses) or 3 (W masses + equal top masses) overconstraints. Normally, only 4 jets are treated, to reduce the number of combinations to be considered (assignment of jets to W and direct b decays, assignment of Ws to leptonic and hadronic top decays). An initial calculation is done to find the z component of the neutrino momentum, which does not appear in the chisquare, but which can lead to two possible local solutions with a quadratic ambiguity. The fit requires parameterisation of the jet energy resolution, and an energy scale correction to map the reconstructed jet energies to parton energies. Christian has derived both for 11.0.41 reconstruction (with cone jet R=0.4), and these are currently read in through a text file. The code is available in CVS, and is being used by Grant for ttbar studies, and by M Siebel for H->ttbar. Other people (Pamela etc) are interested in using this code, and perhaps interfacing it to EventView. Once it is fully integrated into ATLAS software, it could be advertised to the whole top group, and included in the 'official' analysis code, where it may well find applications outside ttbar events. Pippa asked about the mass constraints - at the moment, these (for both W and top, if used) are used without uncertainties. It is not clear if it would be easily possible to implement e.g. Gaussian constraints to represent the W width or top mass uncertainty. CAT top group contributions to CSC notes ---------------------------------------- The status of the CERN group contributions to CSC notes was reviewed: - T9 (mass). CERN (Tancredi and Maria have expressed interest) is expected to contribute to the measurement of the top mass via the dependence on the cross section. Tancredi and Maria met this week to discuss this. One major issue is the scale dependence, where unfortunately the W and top cross-sections seem to be anti-correlated, with a large dependence on the calculations used. Other variables (e.g. Et) will be considered, and they hope to have a contribution to the discussion of this note on Tuesday 17th October. - T6 (cross-section). At present, there is no explicit CERN group contribution to the cross-section note (see https://uimon.cern.ch/twiki/bin/view/Atlas/T6CrossSection ) although several people have expressed interest. If we want to make a concrete contribution here, we need to start getting involved. - T5 (trigger). David reported that work is continuing, focusing on technical analysis issues at present. The group has adapted the SUSY group's SFrame analysis framework package to be able to analyse ntuples with multiple EventViews per event - this may be of wider use beyond the trigger note. Documentation is in progress, and pointers will be circulated when something is available to try out. - T3 (b-jets). Contributions are expected on the derivation of b-tagging efficiencies from data, using kinematic fit-based and kinematic-selection based methods (Grant, Christian, Richard and maybe others). The Wuppertal group are also looking at jet properties in ttbar events in a more general way (which jet finders and b-tagging algorithms work best in the ttbar multijet environment). This work may go into the T3 top note or flavour tagging note (or both). Next events: - Tuesday 17th October: T9 CSC note discussion - Thursday 19th October: CAT physics meeting - 30th October - 2nd November: ATLAS trigger/physics week, including top meeting on 1st November pm.