Tool Status Reports

• Readiness of global tracking for online running: Daniel Whiteson.
  Daniel has completed basic offline checks of the reliability of the L3 global tracking with the test trigger list global_CalMuon-6.20.

  Robert Illingworth has updated the cft stereo map and smt geometry files. It is hoped that the cft stereo map will fix the main problem observed so far, which is a poor and phi asymmetric stereo track-finding efficiency. Daniel will try out these new files with high priority.

  At the Trigger Board today (22nd May) a special run to try out L3 tracking was approved. We shall add a stand-alone track filter with 100% mark&pass fraction to the following triggers: CEM(1,10), CEM(2,5), CEM(1,5), mu1ptxatxx_fz and mu1ptxatxx_CJT(1,3). Tracking in the tau tool will be switched on for the e_tau and mu_tau triggers. All other triggers will be left out of the list or prescaled away at L1. Tracking tool and filter p_t cut will be set to 3 GeV. Daniel and Terry will sit down with Elizabeth on Thursday to get the appropriate trigger file made.

  In order for this to work the following changes need to be made on the L3 farm (running p11.06):

  • New tools.rcp file.
  • Code modification to widen roads.
  • smt/cft geometry file and smt/cft calibration file needed in the local directory.
  Daniel will provide Gustaaf with the appropriate files.
• Studies of online performance of L3 Muon: Martin Wegner, Martijn Mulders.
  Comparisons of L3 and reco local muons show a number of alarming features. In particular:
  • Muons that match well in phi have very poor matching in eta.
  • Local muon momenta agree poorly between L3 and reco.
  Perhaps even more disturbing is that by comparing with muons that have matched central tracks, L2 appears to have a better local muon momentum resolution than either L3 or reco. Next steps:
  • Try reconstructing L3 with the (Monte Carlo!) geometry used by reco.
  • Compare momentum components: drift direction, wire direction, p_z, separately.
  • Try to persuade the muon software group to provide a reasonable approximation to the full survey geometry in the .rcp format used for L3 in p11.
  • In the long run convert L3 to use the EVPACK geometry format stored in the offline database.
  The reasons why the memory consumption of the muon reconstruction code exploded in events with large number of hits has now been fully understood. A fix was supposed to go in for p11.07, but will be there for p11.08.
• Status of L3 missing ET tool: Lee Sawyer.
  Comparisons of L3 and offline missing ET tools have been made using Monte Carlo samples: QCD p_t>10, QCD p_t>100 and ttbar. The performance looks pretty reasonable except for the scalar ET sum for L3, which showed a large (~80 GeV) offset. This is probably caused by using the wrong calorimeter weights. Lee will investigate. This is a trigger list parameter, but this should really be handled automatically! This is a long-standing problem. How can sufficient pressure be brought to bear on those (who should feel) responsible for fixing this?

L3 Algorithm Plans for p12

Scribe: Terry Wyatt.