creating a new specific sequence for ion-ion data taking (PbPb and Pb gas)
adding two new sequences to standalone.py to be used during PbPb data-taking: standalone_hlt2_light_reco_ion
(created from standalone_hlt2_light_reco
) and standalone_hlt2_light_reco_ion_without_UT
(created following standalone_hlt2_light_reco_without_UT
)
each of them call (in hlt2_tracking.py) make_hlt2_tracks_ion
and make_hlt2_tracks_ion_without_UT
(instead of make_hlt2_tracks
and make_hlt2_tracks_UT
) to get the track information.
In them, Best tracks can be filtered with LoKi cut throught a LoKiTrackSelector
. The cut can be implemented through the bind fit_preselection
, to be set to None
for no cut and directly to the selection string. Only best tracks are filtered (BestLong in reconstruction without UT and both BestLong and BestDownstream in reconstruction with UT). The cut must be tuned to a loose ghost probability cut (like TrGHOSTPROB<0.9999) to reduce the amount of fake tracks in high occupancy events to greatly reduce the time spent in RICH reconstruction.
adding 8 new option files in Moore/Hlt/RecoConf/options for testing, with and without MC checking: hlt2_lead_lead_light_reco_pr_kf_without_UT_gec_40000.py, hlt2_lead_lead_light_reco_pr_kf_without_UT_gec_60000.py, hlt2_lead_lead_light_reco_pr_kf_gec_40000.py, hlt2_lead_lead_light_reco_pr_kf_gec_60000.py.
In the test option file for the sequence with UT, we set a cut of TrGHOSTPROB<0.9999 to reduce the time spent at RICH (following this presentation). For the test of the sequence without UT, we use TrGHOSTPROB<0.8. This is tighter with respect to what was shown in the presentation since the ghost probability performance is now worse in the reconstruction without UT (see M!1560 for details).
future tunings of the track reconstruction algorithms (specially seeding and forward) to improve the performance in high-occupancy events will be called only in those sequences to avoid interference with pp conditions.