Muon reconstruction in release 14.0.0.1

D. Adams
April 2008

Release 14.0.0.1 is for validation. Production is planned for 14.2.0.

For other releases, see the complete list of results.

Issues in 14.0.0

1. Hole counts are absent for muid and staco combined.
  • Hole counting is not enabled in 14.0.X reconstruction.

    2. Muid combined fit chi-square does not include the inner track. Staco includes inner, outer and match.
  • Users should add Muid inner and combined chi-squares (and DOF's) to get comparable values.

    3. MuGirl efficiency has dropped to zero.
  • Fixed in 14.1.0.1.

    4. CaloTag efficiency has dropped 10% with a slight increase in the fake rate.
  • Most fakes are at |eta|=1.8 and the barrel fakes are much lower. CaloTag will improve in 14.2.0.
  • Fixed in 14.2.0.1.
    5. CaloLR continues to have low efficiency.
  • CaloLR is not run to find muons in 14.0.X. Likelihoods are added to muons found with other algorithms.
  • Fixed in 14.2.0.1.

    6. The Muonboy T1 mean fit chi-square/DOF has increased to 1.37 from 1.02 in 13.0.40.
  • This is fixed in 14.1.0.

    7. The Moore/Muid T1 mean energy loss has increased to 4.7 GeV from 4.1 in 13.0.40.
  • Alan Poppleton reports that the FSR treatment was dropped temporarily in 14.0.0 due to EDM changes. Muons previously flagged as FSR and assigned parameterized energy loss are now flagged as TAIL and assigned (much larger) measured energy loss.

    8.The T1 direct calo isolation energy has increased to 2.4 GeV from 2.2 GeV in 13.0.40. This is a 10% increase compared to the 20% increase in the Muid energy loss.
  • There are been changes in the track extrapolation that are being checked.
  • The muon subtraction for the isolation calculation was previously made using the energy loss assigned to the muon. In release 14, a core value is used instead.
  • Most of problem appears to come from changes in isolation tool. See D. Adams talk at May 16 physics validation meeting.

    9. Calorimeter-tagged muons all have energy loss set to zero.

    10. The Muid efficiency is low in valid1.005200.T1_McAtNlo_Jimmy.recon.AOD.e322_s418_r398.
  • This is only seen in one file where no tracks are found in the last 800 events.

    11. The CaloTag and CaloLR fake rates drop by a factor of 10 in release 14 reconstruction of release 14 simulation.
  • Most of the fakes in the release 13 samples have no pixel hits. Reconstruction r398 from release 14 simulation does not have these presumably because backtracking was (inadvertently) left off.
  • The fake rate in release 14 reconstruction of release 13 simulation is artifically high. It can be reduced by this factor of 10 by introducing a cut on the number of pixel or SCT hits.
  • Fixed in 14.2.0.1.

    12. The mean number of Moore RPC and TGC hits fall by a factor of two in release 14.
  • Moore is now treating pairs of hits from the trigger doublet layers as single measurements. Muonboy does not do this (but should).

    Graphical summaries

    T1 history
    Z->mumu R13/14 geometry
    T1 R13/14 geometry
    R12/R13/14 history and geometry

    dladams@bnl.gov