Minutes of the muon software meeting April 2 2003, taken by Victoria Slava Krutelyov: Progress of a chisq variable for the CMX. ======================================================================== Slava is developing a x-position based chisq variable for CMX muons. He has parameterized dx for the CMX as a function of pT of the muon and eta. Slava noticed some misalignments in the North Arches of the CMX. Tom LeCompte asked if this misalignment was consistent with the expected "bowing" of the CMX. Tom also pointed out that we will probably only cut on the chisq for low pT muons low-pT, so it is not so important if the chisq values aren't as reliable at high pT due to the misalignments. Tony Liss asked Slava if he had looked at the eta-phi dependence of dx. Slava showed plots showing that the pull of the chisq is the same for all values of phi. Tony pointed out that the phi dependence is probably only relevant at low eta and he asked Slava to look at the phi-dependence for different regions of detector eta, or equivalently Z of the stub or track. For high-pT muons Slava commented that non-isolated muon have a pull on the chisq of 3, therefore a cut on chisq is very effective at rejecting Everyone asked Slava to please use less plots per transparencies and larger fonts so we can better read the scales and statistics boxes! Lucio Cerrito: Update on soft muon ID ========================================================================= Lucio has been working with Anyes Taffard with the aim of constructing a soft lepton ID variable for low-pT muons. As part of this working they have constructed parameterizations for dx, dphi and dz in terms of pT, eta and phi. They are using the CMU, CMP and CMX subsystems for this work. For the CMU dx no eta-phi dependence is observed and the parameterization found is consistent to what Tom uses for the CMU chisq. For the CMP dx there is a noticeable phi dependence. To calculate the pT- dependence of dx muons that are normally incident to the CMP walls (ie around 0, 90, 180 and 270 degrees) are used. The phi dependence is described by a sinusoidal factor. The amplitude of the sinusoid varies with pT. Ken asked if the sinusoidal parameterization was good enough, given the CMP is only nearly a box shape. Lucio said he hadn't tried to fit the phi dependence to any other function yet. For the CMX dx no eta-phi dependence has been investigated yet. For CMU dz a plot of the mean dz vs z shows two distinct slopes with an offset of up to 2cm! This effect is understood as a feature of the current reconstruction algorithm. Lucio has corrected for this effect in this analysis. Ken said the easiest way to correct for this effect would be to change the DtoE. Tom said there should be an elliptical relation between the multiple scattering in the x and Z directions. So given a value for the expected multiple scattering in the x direction and some formula dependent on the geometry of the CMU, the multiple scattering in the z-direction could be predicted. Tony said that he expected this wouldn't actually work, and a parameterization directly from the data was better. Tom asked Tony how much money he would be willing to bet on this ($1) and to this extent Victoria is now holding $1 from Tom in safekeeping for outcome of this wager. The soft lepton tag is formed from a combination of the chisq for all the matching variables considered (dx, dphi, dz) taking into account any correlations. Ken asked if the correlation terms vary with pT. Lucio said that the correlations seem not to show any pT dependence. The overall efficiency of the soft lepton tagger can be tuned by cutting on the likelihood variables. Tony noted the signal to background ratio for this soft lepton tagger can be dialed-in with a choice of cut value. Lucio agreed to work with Slava to look at the phi and eta dependence for the dx in the CMX. Discussion on the chisq ========================================================================== Victoria announced that yesterday she had added values for chsqXPostion() for CMP and CMX stubs into the CdfMuon object at the request of Tom, Pierre and Avi, to ensure that they appear in the 4.10.2 release. This is based only on dx, and the Lucio-Anyes parameterizations have been used. These values are just a placeholder until better parameterization are produced. Everyone is encouraged to keep working on the chisq! Tom asked about the other chisq variables in CdfMuon, for example the combined X-Z chisq variable. Ken said that these names were chosen a few years ago and that we should not restrict ourselves to those variables if we find better ones for analysis. Victoria Martin: CMX trigger inefficiency ======================================================================== Victoria reported on her work to investigate the data-MC mismatch in CMX W->mu nu candidate events that Michael Schmitt had presented on 01/08. By looking at Z->mu+mu- events, the effect is seen to be due to the track trigger (XFT/XTRP) which requires that the COT track leave at least 10 or 11 hits in _each_ of the axial SL (depending on the run number). The inefficient muon legs do exit the COT before they have traveled through the outermost COT SL. The trigger emulation program in the MC describes this effect well at the 98% level. Lee Pondrom and Dan Cyr said that similar problems have been noticed with BMU triggers. The BMU XTRP trigger requires that the tracks only leave hits in the inner 3 axial SL. If there are hits in the outermost SL this track is not considered for the BMU trigger. However this is removing some BMU muons from the acceptance of the trigger. Dan will be making a rate test for the BMU tomorrow allowing both 3 and 4 layer tracks.