Muon reconstruction in release 13.0.40.2

D. Adams
February 2008

Release 13.0.40.2 is a production cache.

For other releases, see the complete list of results.

Issues from 13.0.10

1. Hit counts are absent for muid and staco combined.
  • This was problem with muperf sum of inner and outer. Now fixed.

    2. Hole counts are absent for muid and staco combined.
  • Hole counting is not enabled in 13.0.X reconstruction.

    3. Staco combined fit quality now includes constributions from inner and outer tracks. Previously these were absent. Muid still excludes the inner track.

    4. Muonboy fit chi-square tail is still large. There is a slight improvement from 12.0.X. Moore is again much better.

    Issues from 13.0.20

    1. Calo muon container is present but has many fakes, 10/event in SU(3) and T1. The CaloLR likelihood is always 0.
  • The fakes are somewhat reduced (by about a factor of 4 in Zmumu) in 13.0.30. Remainder can be suppressed with cuts on isolation and pT.
  • CaloLR is fixed in 13.0.30 but efficiency is low (66% in T1).

    2. Endcap efficiency is low for data simulated with 13.0.20.
  • 29aug07 - Ahmimed reports that there is a problem with the TGC simulation.
  • Is this fixed in 13.0.30?

    3. Truth isolation missing for truth muons.
  • Apparently this was done intenally to save space. Tools can be used generate this from the AOD.

    4. It has been reported that muons all have positive charge. Presumably this is in the charge() method--there is not problem with truth matching which is presumably based on track parameters.
  • This should be fixed in 13.0.30.

    Issues in 13.0.30

    CaloLR efficiency is lower than that of CaloTag.
  • To be fixed in 14.2.0.

    Issues in 13.0.40

    1. The MuGirl T1 efficiency falls to about 72% in later builds of this release. It has this value in 13.0.40.4 but 90% in 13.0.40.2.
  • The problem comes from the geometry system and affects ATLAS-CSC-02-01-00 but not ATLAS-CSC-02-00-00.

    2. When the trigger is enabled, Muid energy loss is parameterized and thus loses its tail.
  • Apparently reco Muid is picking up configuration intended for trigger Muid.
  • To be fixed in 14.1.0.

    Results

    Efficiency and fakes are for pT>4 GeV/c, |eta|<2.8 and r0<2 cm.
    Cuts are very loose.

    Graphical summaries:
       T1 history
       SLC3/SLC4 comparison

    dladams@bnl.gov