Draft: Hlt2 and Sprucing strategy for PbPb and PbSMOG2 2024
MR to prepare the 2024 heavy ion data taking
For PbPb:
- Keep hadronic PbPb + UPC lines as they were in 2023
- Add jpsi/upsi line to get ready to use data
- Prepare Hlt2 Monitoring (Ks L0)
For PbSMOG2:
- With the addition of the ghost reduction thks to UT we can have a proper hlt2 strategy
- The physics case almost identical to the pSMOG2 -> tune the lines from pSMOG2
To do:
- Finish tuning the PBSMOG2 rate
- add a PbPb tag for the pSMOG builder
Note: From first tests the average event size seems much more bigger than in 2023 (~1Mb compared to 130Kb in 2023). Might be due to Upstream and downstream tracks plus UT rawbank, has to be investigated and possibly write a new persistreco version.
Merge request reports
Activity
added RTA label
added RTA label
added 2 commits
- 425f5818 - reverse changes
- 18843bd8 - Merge branch 'sbelin_PbPb_2024' of ssh://gitlab.cern.ch:7999/lhcb/Moore into sbelin_PbPb_2024
added 135 commits
Toggle commit listrequested review from @mveghel
88 # ion=ion_lines, 89 # ionraw=ionraw_lines, 90 # lumi=lumi_lines, 91 #) 92 return Streams(streams=[ 93 Stream("full", lines=full_lines, detectors=[]), 94 #Stream("turbo", lines=turbo_lines, detectors=[]), 95 Stream("turcal", lines=turcal_lines, detectors=DETECTORS), 96 #Stream("lumi", lines=lumi_lines, detectors=[]), 97 #Stream("no_bias", lines=no_bias_lines, detectors=[]), 98 #Stream("hlt2calib", lines=hlt2calib_lines, detectors=[]), 99 ]) 100 101 from typing import Callable 102 103 def make_streams(real_make_streams: Callable = _make_streams) -> Streams: -
please have a look at the way we collect binds these days, see https://gitlab.cern.ch/lhcb/Moore/-/blob/2024-patches/Hlt/Hlt2Conf/python/Hlt2Conf/settings/hlt2_binds.py?ref_type=heads#L65
103 def make_streams(real_make_streams: Callable = _make_streams) -> Streams: 104 from PyConf.Algorithms import PrHybridSeeding 105 from RecoConf.legacy_rec_hlt1_tracking import (make_reco_pvs, 106 make_PatPV3DFuture_pvs) 107 from RecoConf.hlt2_tracking import ( 108 make_PrKalmanFilter_noUT_tracks, make_PrKalmanFilter_Seed_tracks, 109 make_PrKalmanFilter_Velo_tracks, make_TrackBestTrackCreator_tracks) 110 from RecoConf.hlt2_global_reco import (reconstruction, 111 make_fastest_reconstruction) 112 113 with make_fastest_reconstruction.bind(skipUT=False),\ 114 reconstruction.bind(make_reconstruction=make_fastest_reconstruction),\ 115 PrHybridSeeding.bind(RemoveBeamHole=True, RemoveClones_forLead=True),\ 116 make_reco_pvs.bind(make_pvs_from_velo_tracks=make_PatPV3DFuture_pvs),\ 117 make_PrKalmanFilter_Velo_tracks.bind(max_chi2ndof=6.),\ 118 make_PrKalmanFilter_noUT_tracks.bind(max_chi2ndof=8.),\ changed this line in version 15 of the diff
101 from typing import Callable 102 103 def make_streams(real_make_streams: Callable = _make_streams) -> Streams: 104 from PyConf.Algorithms import PrHybridSeeding 105 from RecoConf.legacy_rec_hlt1_tracking import (make_reco_pvs, 106 make_PatPV3DFuture_pvs) 107 from RecoConf.hlt2_tracking import ( 108 make_PrKalmanFilter_noUT_tracks, make_PrKalmanFilter_Seed_tracks, 109 make_PrKalmanFilter_Velo_tracks, make_TrackBestTrackCreator_tracks) 110 from RecoConf.hlt2_global_reco import (reconstruction, 111 make_fastest_reconstruction) 112 113 with make_fastest_reconstruction.bind(skipUT=False),\ 114 reconstruction.bind(make_reconstruction=make_fastest_reconstruction),\ 115 PrHybridSeeding.bind(RemoveBeamHole=True, RemoveClones_forLead=True),\ 116 make_reco_pvs.bind(make_pvs_from_velo_tracks=make_PatPV3DFuture_pvs),\ changed this line in version 15 of the diff
I ask @jcambonb to test the reco performance for both and there is a much higher efficiency with the default one at high multiplicity so I ll prepare two throughput test with each reco
