update of HLT2 trigger lines for Long Lived Particles

Retuning of the selection cuts of Dark Higgs and HNL lines using T-Tracks, combining them with improved multiclass T-tracks MVA filter Rec!3885 (merged)

• B -> K H'(->mu mu)
• B -> K H'(-> K K)
• B -> K* H'(->mu mu)
• B -> K* H'(-> K K)
• B -> mu N'(-> mu pi)
• Bc -> mu N'(-> mu pi)

improving efficiency, rates and bandwidth and limiting multiple candidates.

We have also considered to implement a Dark Higgs di-muon line Hlt2QEE_HtoMuMu_TT (H'->mu mu) removing the dependence on B decays and combinations with long tracks, in order to test for a more inclusive approach in the future. Furthermore, since rates/BW change looks low, we can think of full persisting this line and include T-Tracks lines in Sprucing for the first time in future.

Rates:

HLT rates: 
--------------------------------------------------------------------------------------------------------------
Hlt2QEE_BctoMuN_PiMu_exclTTDecision      Incl: 0.023 +/- 0.016 kHz,     Excl: 0.000 +/- 0.000 kHz
Hlt2QEE_BtoEN_MuPi_InclTTDecision        Incl: 0.046 +/- 0.023 kHz,     Excl: 0.046 +/- 0.023 kHz
Hlt2QEE_BtoKH_KK_exclTTDecision          Incl: 0.034 +/- 0.020 kHz,     Excl: 0.034 +/- 0.020 kHz
Hlt2QEE_BtoKH_MuMu_exclTTDecision        Incl: 0.011 +/- 0.011 kHz,     Excl: 0.011 +/- 0.011 kHz
Hlt2QEE_BtoKstar0H_KK_exclTTDecision     Incl: 0.011 +/- 0.011 kHz,     Excl: 0.011 +/- 0.011 kHz
Hlt2QEE_BtoKstar0H_MuMu_exclTTDecision   Incl: 0.000 +/- 0.000 kHz,     Excl: 0.000 +/- 0.000 kHz
Hlt2QEE_BtoMuN_MuPi_InclTTDecision       Incl: 0.057 +/- 0.025 kHz,     Excl: 0.057 +/- 0.025 kHz
Hlt2QEE_BtoMuN_PiMu_exclTTDecision       Incl: 0.057 +/- 0.025 kHz,     Excl: 0.034 +/- 0.020 kHz
Hlt2QEE_HtoMuMu_TTDecision               Incl: 0.000 +/- 0.000 kHz,     Excl: 0.000 +/- 0.000 kHz
Total                                    Incl: 0.217 +/- 0.050 kHz,     Excl: 0.217 +/- 0.050 kHz
--------------------------------------------------------------------------------------------------------------

BW changes

Estimated BW variation for both all qee turbo and full lines wrt qee_upgrade branch, considering 10k events from 2024 minbias HLT1-filtered v2 sample, are

• full lines: +14,5 MB/s
• turbo lines: -0.2 MB/s (it decreases).

Efficiency signal was very poor mostly for all lines in the previous version, now it has increased order 10x.

Effects of multiple candidates, mainly show up when matching the LLP candidate with the long track, are visible running over early 2024 HLT1-filtered data (see docs by Ross ).

Implementing a cut on events with nPVs<12 in each line flow would exclude these patological events