Muon Offline Meeting -- Minutes ================================= 16 - Oct - 2002 ---------------------------------------------------------------------- Slava Krutelyov -- Retuning the MExtrapolator ========================== Recently, we discovered that the track-stub matching in the CMX was rather poor for low-momentum muons, suggesting that the extrapolator needed to be better tuned in that region. Slava, the original author of our extrapolator, has taken on this work. (This is the sort of thing that G3X would have done for us naturally, but we can't use it now.) He uses single-muon events for his tuning, and will compare the results to J/psi data in the next few days. Slava considers the problem to be an energy-loss effect, and that only has an effect in regions where there is a B field. Thus, in the CMU and CMP, energy loss has very little effect, as the B field is small in the CHA, and there is no sizeable correction to the trajectory. But there are effects in the CMX and IMU, because there is large flux, and the field is not parallel to z, so z component of the muon momentum is important. Slava had to tune pretty hard; there are non-trivial corrections. But as a result, he gets significant improvement in CMX -- what was an offset in the dx mean of about 1.5 cm at 4 GeV/c is corrected to few mm level. Things still look pretty bad below 2 GeV/c, but no one would use muons there anyway. Looking at dx as a function of pT, there appears to be a discontinuity around 3.5 GeV/c, which led to some questions. Phil and Tony both expressed concerns about peculiar regions of the CMX, especially near the CMP corner regions, and Slava was asked to produce plots of dx as a function of eta and phi. He still needs to look at the data, and might have to do some detuning. Getting the tuning more precise does not sound easy. Anyes Taffard -- Cosmic-ray update ================= As has been presented before, Anyes has made many improvements to the cosmic tagger. The most important is Ashutosh's work to refit tracks to a single helix, and to find incoming tracks that weren't found in the standard tracking. This refit helps other aspects of the tagging work better, including some new TOF code. Anyes showed her proposed tagging criteria, based on studying W and Z events, using 4.5.3 data and selection cuts similar to those for the ICHEP analysis. She has tried to avoid cuts on funny-looking distributions, particularly a strange d0 vs. phi distribution that supposedly arises from one of the COT algorithms and is supposed to be under control in 4.8.4. As usual, Anyes breaks the situation down into one-leg and two-leg cases, but the Ashutosh refit is done before this classification, which allows many one-leg events to become two-leg events. To estimate performance, she looks at the delta-z distribution between the muon tracks and the event vertex, which should be flat for cosmics and peaked for physics. From this, she can calculate an overefficiency leading to the rejection of real W and Z events, and impurity rate in those samples. Her proposed cuts are 0.5% overefficient for W, and the resulting impurity is 4.5%. For the Z, she can't see measurable overefficiency, and the impurity is 5.4%. Upcoming features will be a measurement of out-of-time energy that can be an indication of cosmics that brem, or of beam halo, but these will not be used in the default tag, at least for now. Anyes also gave a short user guide. The default cuts are implemented as a talk-to to CosmicFinderModule, which the user can change if desired. A new stripping module allows the user to keep or not keep cosmics, and to change the tag cuts if desired. Our intent is to get all of this into 4.9.x, and there were no objections voiced. Henry asked how the code can be run on electrons, and the answer was "not easily." We should change that, as the electrons are a good control sample. Victoria Martin -- Weird CMU stub-z positions ========================== Slava had looked at a sample of low-pT muons, and found CMU stubs with z coordinate outside the chambers! That doesn't make any sense, so Victoria tried to understand what is going on. Slava sees about 1% of muons are affected by this problem, at least in the run range he looked at. Victoria looked at high-pT CMU and CMUP muons over a larger range with 4.8.3 reco, and finds that muons passing standard ID cuts have wacky z positions at most 0.4% of the time. She looked at just a few stubs in detail. The events in question tend to have many chamber hits, and therefore many stubs, one of which has the wacky z. Some of the failure modes include stubs with hits with overflow width, which are not used in the fit. If a stub has zero or only one hit with a legitimate width, it's essentially impossible to get a good fit of the z position, and crazy answers result. There is also trouble with low-width hits, which are very close to the expected minimum value. We probably shouldn't worry too much about this, as these are probably not useful muons, but Victoria will continue to explore the problem. Volker Drollinger -- Low-pT muon ID ============== Volker is trying to come up with a set of muon ID cuts that have a constant efficiency as a function of pT. He selects muons based on the energy deposition, the d0 and z0 of the track, and a chisquared-like quantity based on the dx and dphi values for the track-stub match(es). This chisquared is normalized by pT-dependent errors that are determined by fitting the distributions as a function of pT. (Andreas noted that the cuts being used in the J/psi group use a looser chisquared.) Volker determines the efficiencies of these cuts with J/psi muons, doing a sideband subtraction to get a pure sample. The efficiency as a function of pT is not flat, but it's a start; presumably the tails of distributions become a problem. Tom was surprised by the loss of so many events on the z0 cut. This could be because the J/psi do not come from the primary vertex, or that maybe the primary vertex is not trustworthy. More work to come. Awards Ceremony -- Ken announced that he owed Tom a beer for his work on studying the opening angle of muons from J/psi. ---------------------------------------------------------------------- K. Bloom