Muon Offline Meeting -- Minutes ================================= 2 - Oct - 2002 ---------------------------------------------------------------------- Bruce Knuteson -- Local Energy Flow and Muon ID ============================= This work is documented in cdf note 6083. Bruce is seeking to improve muon ID in the sense of separating prompt high-pT muons from hadrons and muons in jets. He prepares a sample of "mu+MET" events using standard kinematic cuts and a veto of events with a jet opposite the muon, and a "mu+jet" sample demanding a jet opposite the muon. He is careful to avoid any cuts on the energy or isolation of the muon. He uses the electromagnetic and hadronic energies associated to the muon, the calorimeter isolation, and a track-based isolation. He showed that the latter two are only partially correlated, so in principle some discrimination can be obtained by combining them. In essence he is working on the isolation of the muon. Bruce defines a log likelihood ratio as his estimator. (He takes the logarithm to obtain a manageable distribution.) The discrimination power is good. He compares to the old "box" cut as follows eff contamination box 74 % 5.0% log ratio 75% 4.4% A significant reduction in the contamination is observed, for essentially the same muon efficiency. Furthermore, he estimates from these studies that the probability to find a muon in a jet is about 5 x 10**-5. There was a discussion of the tails in the mu+MET sample. Henry suggested that they might come from Bremsstrahlung which could be identified as a special case. Bruce plans to look into this. ---------------------------------------------------------------------- Camille Ginsburg -- IMU Offline Update ================== Camille et al. used data from recent special BMU runs, and used the latest version of Jim's code. Since the last muon offline meeting, Jim has fixed the bug giving spurious (X,Y) coordinates for some stubs. Together with fixes to the handling of the Z coordinate, Camille reported a doubling of the number of muons reconstructed! She shows an eta distribution which demonstrated muon identification out to eta of +/- 1.6 ! This was greeted with enthusiasm by the group. The phi distribution is reasonably flat within the fiducial range of the chambers. Furthermore, Dan Cyr produced plots of the calorimeter energy associated with these muons. The distributions show a very clear min-I peak with little indication of contamination (high energy values) or fake tracks (zero or little calorimeter energy). It is clear that there are roughly 40k BMU muons in these runs. The Wisconsin group wants to investigate some apparent holes in the hit coverage and some anomalies in the stub chi-squared distribution. The group felt that a patch for L3, if needed, should be pushed vigorously so that the best possible BMU muons can be triggered on. ---------------------------------------------------------------------- Victoria Martin -- dX-dphi correlations for Matching ================================= Victoria prepared two samples, one which is muon-like and one which is hadron-like, by cutting hard on the calorimeter energy. She examined the correlation between dX and dphi for muons, and sees a weak but definite correlation. (dX is dRphi, ie, the matching distance in the azimuthal direction, while dphi is the difference in the track/stub slope in the transverse plane.) She finds the major axis semi- empirically by looking at the width of projections while changing an angle of rotation. The side-by-side comparison of muons and hadrons indicates that the discrimination could be improved by including a cut on dphi, possibly as an ellipse in the (dX,dphi) plane. However, for muons there are a significant fraction which show large dphi for small dX. A cut applied now would eliminate several percent of the muons and this is unacceptable for most analyses. Victoria investigated the events which a large dphi and small dX. She showed that these typically have fewer hits on the muon stub. This is true for CMU and especially for CMP. On the basis of other distributions she is sure they are not cosmics. It is possible that the stub reconstruction is impacted by delta rays or something like that. Victoria will study this further. ---------------------------------------------------------------------- M.Schmitt / NWU