Minutes of the Muon Software meeting 14 May 2003, reported by Victoria. 0. news ----------------------------------------- No one had any news! 1. Victoria - CMP Geometry ---------------------------------------- Un-Ki had noticed some funkiness with the CMUP muon eta-phi distribution at phi=135 degrees. David and Phil had suggested this made be due to an overlap in the CMP geometry. Victoria has looked again at the geometry, and nothing overlaps. Some of the CMP containment volumes do touch, but in many places, not just near phi=135 degrees. 2. Slava - CMX chisq ----------------------------------------- Slava has been working towards a chisq variable for CMX muons based on dxCMX. He has been using both 4.9.1hpt1 and 4.10.0 with the improved extrapolator. Using sideband-subtracted J/psi->mu+mu- candidates, Slava has parameterized the mean and the width of the distributions as a logarithmic functions of charge, pT and eta. Using his parameterizations, he finds that the pull on the chisq is almost perfect: mean of 0 and sigma of 1. If he doesn't use the corrections on the mean, he also finds a pull with a mean of 0 and a sigma of just over 1. However this is the pull averaged over all pT bins and hides the variations for different values of pT. Comments: Lucio asked what was the motivation for the logarithmic functions. He said physically there was no interpretation. A more physical function could be of the form a/pT (to account for multiple scattering) plus b/pT^2 which is important at very low pT, where a 1/pT term does not describe the multiple scattering correctly. 3 Anyes - CMX chisq ----------------------------------------------- Anyes has also been working towards a chisq variable for CMX muons. Using Slava's parameterization of the mean, Anyes has also made a parameterization for the width of the dxCMX distribution. Her parameterization takes account of variations in eta as a function of pT, plus multiple scattering effects using a a/pT + b/pT^2 distribution. 4 Lucio : Correlation and pulls for CMUP muons ------------------------------------------------ As part of his work on the soft muon tagger for CMUP muons, Lucio has been looking at the correlation between dxCMU, dphiCMU, dzCMU, dxCMP and dphiCMP. The aim is to make a soft muon tag variable accounting for all of these correlations. Comments: Ken asked why dphiCMU and dxCMP are highly correlated. Phil said that because of the distance between the CMU is large any different in phi in the CMU will project into a difference in x in the CMP. dphiCMU is actually a better predictor of the position of the muon in the CMP. 5 Michael Gold : Residuals ------------------------------------ Michael has looked at the residuals for the CMU, CMP and CMX hits from data. He has fitted them to the sum of two Gaussians. The CMX doesn't really warrant a two Gaussian fit, so for this only the inner Gaussian is used. This parameterizations of the residuals is now in the code and can be turned on via a-talk to. If there are used the agreement between the matching variables for data and MC is much better. Comments: Lucio asked why the CMP residuals appear to be so large, ~1mm as compared to only about 350 um in run 1. Ken commented that this could be due a a bad drift model.