Muon Offline Meeting -- Minutes ================================= 24 - July - 2002 ---------------------------------------------------------------------- Dave Dagenhart -- Reconstruction Efficiency ========================= Dave presented the work he did with Ken Bloom on measuring the reconstruction efficiency for muons in the CMU, CMP and CMX. For a detailed description, see cdf note 6029. They use the muon+SVT track sample for the Feb-June data, and looked for J/psi's. They loop over tracks which are well within (10cm) the edges of chambers, using Dave's fiducial tool. They form the invariant mass distributions for tracks which are found as muons, and those which aren't. Clean distributions are obtained when requiring one subsystem when testing the other. Side-band subtracted yields are used for the efficiency calculation. Ken scanned the CMU and CMP failures, and categorized most of them. Out of 28 CMP failures, eight have zero hits, which cannot be ascribed to chamber inefficiency. These are used to set a systematic error. Comments and suggestions during the talk: - For the future it would be best to apply cuts on the track (such as Nax>24, Nst>24) which match those used in analysis, so that these efficiency measurements apply directly to the analysis. We need to find out what the tracking group recommends and will use in defTracks. - The requirement of CMU when testing CMP may eliminate the events with a large angle scatter in the calorimeter, and lead to a larger efficiency number. - Run on Monte Carlo to see what the impact of small non-fiducial regions is. ---------------------------------------------------------------------- Ashutosh Kotwal -- Identifying Single-Legged Cosmics with the COT ============================================== In a cosmic ray event, one track is backward-moving. This track can be detected with the COT, which is a precision timing device. Specifically, one can check the beta=-1 hypothesis. Also, it provides a T0 for the track which can be compared to the T0's for other tracks. Strategy: - Find the first muon track and refit with a floating T0 and the two hypotheses beta=+1 and -1. Choose the right hypothesis on the basis of chi-squared. - Using this fit, open a road of a few mm to look for hits on the other side of the COT. Energy loss and multiple scattering are assumed to be unimportant, which is accurate for cosmic ray muons above 10 GeV or so. - Fit the second track, for both beta=+1 and -1, and choose the best fit. - Compare the compatibility of the T0's and the beta's for the two tracks. - Do a fit to a single helix, with a single T0, but allowing for independent beta's in the two legs. The best fit corresponds to a pair of opposite-sign beta's for cosmic rays, and +1 for Z->mumu at rest. The expected error on T0 is about 0.5ns. Ashutosh showed how well this works with a number of case studies. The results are very convincing. A hand scan revealed no failure of the method in 120 events. The fitting code is based on CT_ReFit and will appear for 4.7.0. Ashutosh will work with Anyes to add the output of his module to the cosmic ray object. Later, the CR Tagger will be modified to take this information into account. (The goal is to have this ready for 4.8.0, ie, 4-5 weeks.) Comments from Henry: > The CR Tagger should be run in Production on Z->ee candidates by default; this requires a modification to the CR tagger. > The muon group should make official samples of CR events for study. ---------------------------------------------------------------------- Un-Ki Yang -- Badly Reconstructed Tracks in the Muon Sample ============================================= Un-Ki was checking the muons with zero hadronic energy. A fraction of W->munu candidates (25%) have zero hadronic energy, which is a large increase over previous samples. The fraction decreases to 6% when he imposes a requirement of at least 20 stereo hits. He and Bruce Knutson examined these events in detail. The `missing' energy is seen in the isolation sum, and the event display confirms that the muon track is not pointing to the correct calorimeter tower. These tracks have bad chi-squared, and do not point to the primary vertex: dZ is on the order of 20 or 30 cm. (As a consequence, they are eliminated from the W and Z candidates in the cross section and R measurements. ) It is evident that they are not cosmic rays (this is confirmed by looking at the phi distribution of these muons). They noticed that these high-pT muons often have another track very close by in phi, which has a good dZ and which points to the correct tower. They conclude that the high-pT track is a ghost and the low-pT track the real muon. The pT distribution of the real muon peaks around 5.8 GeV. It seems very clear that these muons represent tracking failures. Un-Ki and Bruce are meeting with experts from the tracking group to understand what is wrong. It would be possible to take track and fit quality into account when matching tracks with muon stubs, and for this class of events, the problem would be solved. Until now, however, it was the policy of the muon and tracking groups that muon reconstruction should not apply such criteria; this issue can be further discussed after more study. ---------------------------------------------------------------------- Michael Gold -- Simulation Progress =================== Michael has performed some ad hoc tuning of the muon chamber simulation on order to better reproduce the residual distributions in the data. He has added some noise to the CMU, which helps reproduce the up-turn of the drift distance distribution at small distances, but is not yet sure this is the correct thing to do. He has matched the resolution in Z for the CMU, so the dZ distributions now agree with the data. The CMP residuals are very puzzling, as they are much larger than expected. Michael does not favor inflating the smearing that much -- more work is needed here. Similarly, the chi-squared distributions are hard to understand. Michael also looked at the hit efficiency and also its behavior near the edges of the chambers. He finds 95% to be a good value for the CMU, and will look at other systems later. ---------------------------------------------------------------------- M.S.