Skip to content

Porting of Run 2 DiMuon lines to Run 3 framework

Alessio Piucci requested to merge apiucci-dimuon into master

Porting of the prompt and exclusive DiMuon lines (+ related filters) from Run 2 to the Run 3 framework. Implementation of basic DiMuon, Ks LL and Ks DD combiners in the standard_particle module (EDIT: the two Ks combiners will have to be removed, they are being implemented by MR !232 (merged)). The structure of the filters and lines (aka the requirements on the particles/candidates) is the same of Run 2.

The make_massconstrained_jpsi2mumu combiner now takes as input the loosest dimuons used for the dimuon lines. Therefore it has been converted to a filter.

Added an option to test the DiMuon lines on a variety of MC samples.

Lines which are implemented:

  • Hlt2DiMuonJPsiLine
  • Hlt2DiMuonJPsiHighPTLine
  • Hlt2DiMuonPsi2SLine
  • Hlt2DiMuonPsi2SLowPTLine
  • Hlt2DiMuonUpsilonLine
  • Hlt2DiMuonSoftLine
  • Hlt2DiMuonZLine
  • Hlt2DiMuonChicJpsiDiMuonLine
  • Hlt2DiMuonB2KSMuMuLLLine
  • Hlt2DiMuonB2KSMuMuDDLine

Other lines are implemented, but commented:

  • Hlt2DiMuonBLine
  • Hlt2DiMuonBUBLine
  • Hlt2DiMuonBPromptLine

These lines have been used for B(s) --> mu mu analyses, and the authors are implementing their own exclusive lines. I am not aware of other analyses which were using the same lines: in case, they can be restored in the future. EDIT: actually there are two MRs opened already for B(s) --> mu mu analyses (MR !207 (closed) and MR !210 (closed)).

Performances of the dimuon lines

Two sets of requirements are investigated:

  • à la Run 2;
  • tighter mass windows around the Jpsi and the Psi2s (120 MeV --> 100 MeV), tighter ProbNNmu on the muons of the Jpsi and Psi2s lines (0.1 --> 0.2).

The rates of the lines are measured from a HLT1-filtered MinBias sample used for the Upgrade selection studies, Sim09cUp2, about 60k filtered events. Assuming a HLT1 output of 1 MHz, the rates of the HLT2DiMuon lines implemented are given by their efficiency over MinBias events (number of events passing the line / number of input events). The measured rates are (first column: Run 2 requirements; second: tighter selection) :

Hlt2DiMuonJPsiLine       = 1240 Hz +- 140 Hz | 930 Hz +- 130 Hz  # the elephant in the room, as usual
Hlt2DiMuonJPsiHighPTLine = 707 Hz +- 110 Hz  | 589 Hz +- 100 Hz
Hlt2DiMuonPsi2SLine      = 320 Hz +- 73 Hz   | 168 Hz +- 53 Hz
Hlt2DiMuonPsi2SLowPTLine = 185 Hz +- 56 Hz   | 84 Hz +- 38 Hz
Hlt2DiMuonUpsilonLine    = 101 Hz +- 41 Hz
Hlt2DiMuonSoftLine       = 185 Hz +- 56 Hz
Hlt2DiMuonZLine              < 15 Hz
Hlt2DiMuonChicJpsiDiMuonLine < 15 Hz
Hlt2DiMuonB2KSMuMuLLLine     = 185 Hz +- 56 Hz
Hlt2DiMuonB2KSMuMuDDLine     = 51 Hz +- 29 Hz

A 0.2 prescale was applied to Hlt2DiMuonJPsiLine and 0.1 prescale to Hlt2DiMuonPsi2SLine in Run 2.

The signal efficiency of the lines are computed over locally-produced signal Upgrade MC, Sim09cUp2, 10k-12k events per mode. The inclusive Upsilon sample (18112001) is taken from the bookkeping. The efficiencies are computed as: number of events passing the line / number of events passing the GEC (&& PV requirement).

HLT2 signal efficiencies without HLT1 filtering:

eff Hlt2DiMuonJPsiLine (24142001)       = (43.38 +- 0.49) % | (43.20 +- 0.49) %
eff Hlt2DiMuonJPsiHighPTLine (24142001) = (30.40 +- 0.45) % | (30.17 +- 0.45) % # a pt cut should be put in the "reconstructible" candidates
eff Hlt2DiMuonPsi2SLine (28142001)      = (44.70 +- 0.55) % | (44.31 +- 0.54) %
eff Hlt2DiMuonPsi2SLowPTLine (28142001) = (16.44 +- 0.41) % | (16.23 +- 0.40) % # a pt cut should be put in the "reconstructible" candidates
eff Hlt2DiMuonUpsilonLine (18112001)        = (52.17 +- 0.52) %
eff Hlt2DiMuonChicJpsiDiMuonLine (28144040) = (4.13 +- 0.24) %
eff Hlt2DiMuonB2KSMuMuLLLine (11114100)     = (7.37 +- 0.31) %
eff Hlt2DiMuonB2KSMuMuDDLine (11114100)     = (3.45 +- 0.21) %

HLT2 signal efficiencies over HLT1-filtered events (caveat: the HLT1 as implemented for the Upgrade selection studies has been used):

eff Hlt2DiMuonJPsiLine (24142001) | HLT1       = (71.86 +- 0.63) % | (71.39 +- 0.63) %
eff Hlt2DiMuonJPsiHighPTLine (24142001)LT1     = (50.22 +- 0.70) % | (49.90 +- 0.70) % # a pt cut should be put in the "reconstructible" candidates
eff Hlt2DiMuonPsi2SLine (28142001) | HLT1      = (65.71 +- 0.67) % | (65.14 +- 0.67) %
eff Hlt2DiMuonPsi2SLowPTLine (28142001) | HLT1 = (23.93 +- 0.60) % | (23.64 +- 0.60) % # a pt cut should be put in the "reconstructible" candidates
eff Hlt2DiMuonUpsilonLine (18112001) | HLT1        = (85.05 +- 0.23) %
eff Hlt2DiMuonChicJpsiDiMuonLine (28144040) | HLT1 = (14.19 +- 0.88) %
eff Hlt2DiMuonB2KSMuMuLLLine (11114100) | HLT1     = (10.35 +- 0.43) %
eff Hlt2DiMuonB2KSMuMuDDLine (11114100) | HLT1     = (4.94 +- 0.31) %

Performances of the Hlt2BsToJpsiPhiLine exclusive line

The Hlt2BsToJpsiPhiLine line was using the make_massconstrained_jpsi2mumu combiner. Its performances using the old-fashion combiner and the new filter are compared (first column: old version, second column: new version).

rate Hlt2BsToJpsiPhiLine = 12970 Hz +- 700 Hz | 2100 Hz +- 190 Hz
eff Hlt2BsToJpsiPhiLine  = (28.99 +- 0.37) % | (27.55 +- 0.37) %
eff Hlt2BsToJpsiPhiLine | HLT1 filtered = (34.63 +- 0.30) % | (33.15 +- 0.30) %
Edited by Alessio Piucci

Merge request reports