BandQ sprucing tuning targetting on April deadline
This MR is for the BandQ sprucing rate tuning before April deadline.
Description of the MR circulated in lhcb-onia mailing list
Detached Jpsi2MuMu, psi2S2MuMu sprucing line with persistReco
Update: tighten PIDmu cuts, loosen fight distance and pT cuts
These lines are expect to behave as "StrippingFullDSTDiMuonJpsi(Psi)2MuMuDetachedLine", mainly for b->psiX and other detached psiX studies. We update the selections in Run3 framework and make them closer to the corresponding lines in Stripping34: https://lhcbdoc.web.cern.ch/lhcbdoc/stripping/config/stripping34/dimuon/strippingfulldstdimuonjpsi2mumudetachedline.html
This includes:
*Tighten the PIDmu cuts from -5 to 0
*Loosen the DecayLengthSignificance cuts from 4 to 3
*Loosen the pT(mu) cuts from 500MeV to 400MeV
Comments:
Estimated impact on rate: reduce 30%
Estimated impact on sprucing efficiency: reduce 3 to 5% for typical b->jpsi X decays. But flight distance and pT are usually used in offline MVA, a looser cut on these variables in sprucing may improve the MVA performance and partially recover the efficiency lost. The updated selection is almost the same as the corresponding Run2 stripping, with the pT(mu) cut slightly looser.
Prompt Jpsi2MuMu, psi2S2MuMu sprucing line with persistReco & with tight kinematic cuts
Update: tighten PIDmu cuts
These lines are expected to act as the "StrippingFullDSTDiMuonJpsi(Psi)2MuMuTOSLine" in Run1-2, for prompt psi+X studies. We update the selections in Run3 framework and make them closer to the corresponding selection in stripping34: https://lhcbdoc.web.cern.ch/lhcbdoc/stripping/config/stripping34/dimuon/strippingfulldstdimuonjpsi2mumutosline.html. This includes:
*Tighten the PIDmu cuts from -5 to 0
Comments:
Estimated impact on rate: reduce 10%
The updated selection is almost the same as the corresponding Run2 stripping.
Double open-charm lines
Update: require two open-charm hadrons originating from the same reconstructed PV, or from two reconstructed PVs that are very close to each other
These lines will be used mainly for spectroscopy studies with two open-charm hadrons in the final states. The master-branch requirement is only to ask a event to have two open-charm candidates reconstructed and selected. Now we would like to also require:
*The z-component of the distance between the associated PVs of these two charm hadrons is smaller than 10mm
Comments:
Estimated impact on rate: reduce O(10%)
Estimated impact on efficiency, tested using prompt D0D0, D0D+ samples: no visible variation using O(10k) signals as input
Estimated impact on efficiency, tested using D0D0bar from B-decays: reduce 2% relative efficiency.
Several inclusive dimuon triggers without a quarkonia mass window cut
Update: remove the persistReco feature
We noticed a few inclusive dimuon triggers with persistReco=True, but cannot think out a usecase of the full event using the output of these triggers. Some of these lines have a Run-2 version, in Leptonic stream of Stripping34, and are saved in MDST format. For now, we would like to set persistReco=False (do not store the whole reconstructed event, only store the dimuon signals) for the following lines:
BandQ_DiMuonInc
BandQ_DiMuonSameSignInc
BandQ_DiMuonIncHighPT
BandQ_DiMuonSameSignIncHighPTComment:
Some of the lines have a sizable rate, even comparable with the inclusive detached Jpsi2MuMu persistReco sprucing line. Setting persistReco=False for these lines can significantly reduce the bandwidth cost.
But if anyone knows any potential physics projects using these lines to build dimuon+X combinations, where persistReco is needed, please get in touch with us, and we will try to find out alternative ways to make the bandwidth cost under control.
detached Etac->4h combiner
Update: Add the Combination12, Combination123 cuts to speed up the combination. Add a SumPT cut and SumIPCHI2 cut to reduce the rate.
Require SumPT of all tracks > 2500MeV, SumIPCHI2 of all tracks > 30. These values are taken from the Run2 Lb->etac(4h) p K stripping line: https://lhcbdoc.web.cern.ch/lhcbdoc/stripping/config/stripping34/bhadron/strippinglb2etackp_4hline.html
Comment:
2022 data shows that these combiners have a large "multiple candidate" issue. See the discussion here: https://mattermost.web.cern.ch/lhcb/pl/fpdnfbd7oibkme45743t6p75ty . The updates proposed here is part of the effort of resolving this issue.
B->etac(4h)Kpi, Lb->etac(4h)pK lines
Update: tighten the selection using Run2 stripping as a reference
Rates of these lines are too high, at O(1kHz) level. Using the Run2 Lb2etac(4h)pK stripping line as a reference could significantly reduce the rate. This include:
*Add SUMPT cut of KPi, pK combinations
*Tighten the b-candidate vertex quality cut (help reject BKGs from b->dX decays with large branching fractions and the same final states)
*Tighten the b-candidate flight distance cut
Comment:
2022 data shows that these lines have a large "multiple candidate" issue. See the discussion here: https://mattermost.web.cern.ch/lhcb/pl/fpdnfbd7oibkme45743t6p75ty . The update proposed here is not only for reducing the rate, but also part of the effort of resolving the multiple-candidate issue.
Impact on bandwidth cost based on local test using Moore_hlt2_and_spruce_bandwidth.sh. Compare this MR and the master associated with this MR.
Hlt2
No/negligible variation expected.
Master
This MR
Sprucing
Expect a visible bandwidth reduction for BandQ, no impact on other WGs.
Master
This MR
Summary
Negligible impact on Hlt2 (a small fluctuation ?). Reduce sprucing bandwidth cost about 0.1 GB/s.