Update on LHCb Level1 trigger 1102022 Federica Legger
Update on LHCb Level-1 trigger 1/10/2022 Federica Legger 1
Summary n n n n 1/10/2022 TDR L 1 Post TDR needs L 1 redesign L 1 bandwidth division L 1 efficiencies for some representative channels Long-standing L 1 open questions Status and plans 2
TDR times LHCb trigger TDR (September 2003) Aim for highest possible efficiencies for a list of benchmark channels (tagging not taken into consideration due to the lack of statistics…) 1/10/2022 3
TDR L 1 code n L 1 Generic (B meson decay products) q n Massive high Pt Long-lived high IP Selects events with two high PT tracks in IP window [0. 15, 3. ] mm. ; Bonus system (specific) q q Dimuon n |mmm – m. J/ | < 500 Me. V n mmm > m. B – 500 Me. V Electron n q Etmax > 3 Ge. V Photon n Etmax > 3 Ge. V Courtesy of Thomas Schietinger 1/10/2022 4
Advantages: n Bonus values are tunable for optimal efficiencies/bandwidth; q n Good for TDR! Trigger summary in just one variable: q q Convenient for comparisons Easy for users Disadvantages: n Correlation among various subtriggers; q n n n Hard to understand why an event has triggered; An event with some momentum, some dimuon mass, a little bit of photon and electron can trigger…(? ? ? ) Not easy to implement parallel subtriggers q 1/10/2022 Cut on PT depends on bonus values example: Thomas’ single muon hack 5
Post TDR n Some bug corrections and code improvements q n n tracking and multiple PVs Need to better understand L 1 behaviour & systematics Unbiased data samples q q trigger on tag Indipendent trigger line Redesign of L 1 code 1/10/2022 6
DC 04 L 1 code n SPECIFIC Parallel (and overlapping) subtriggers q q q n GENERIC Generic Single muon Dimuon (J/Psi) Electron Photon Final decision is OR of single subtrigger decisions 1/10/2022 Courtesy of Thomas Schietinger some PT cut still needed to reduce bandwidth 7
n n Generic (p. T): q For tracks with IP > 0. 15 mm q Pile-up veto (IP < 0. 15 mm), up to 2 PVs Single-muon: q ln(PT 1) + ln(PT 2) > 13. 925 IP > 0. 15 mm, PT > 2. 32 Ge. V Pile-up veto (IP < 0. 1 mm), up to 3 PVs n Dimuon general: mmm > 500 Me. V, IPmm > 0. 05 n Dimuon J/ : |mmm – m. J/ | < 500 Me. V OR n q includes B, Z, H, X mm q No IP cut Electron: mmm > m. B – 500 Me. V max. ET(e) > 3. 6 Ge. V AND ln(PT 1) + ln(PT 2) > 13. 0 n Photon: max. ET(g) > 3. 1 Ge. V AND ln(PT 1) + ln(PT 2) > 13. 0 1/10/2022 8
Bandwidth division Before: Massive improvements in generic algorithm (multiple PVs!) result in larger bandwidth for other triggers (in particular dimuons) Now: 1/10/2022 Generic Single-muon Dimuon, general Dimuon, J/Psi Electron Photon Overlaps are absorbed in this direction Bandwidth (k. Hz) 29. 4 (74. 1%) 6. 8 (17. 1%) 1. 5 ( 3. 8%) 1. 8 ( 4. 6%) 3. 7 ( 9. 4%) 4. 3 (10. 8%) Courtesy of Thomas Schietinger Adjusted for overlap 29. 4 (74. 1%) 3. 2 ( 8. 0%) 1. 2 ( 3. 1%) 1. 2 ( 2. 9%) 2. 2 ( 5. 5%) 2. 5 ( 6. 4%) 9
L 1 efficiencies Offline selected Reconstructible 1/10/2022 10
L 1 efficiencies why different? 10 -20% improvement!!! Offline selected Reconstructible Hadrons trigger e and g! 1/10/2022 4 -prong give best generic efficiency! 11
Implementation: n Da. Vinci (LHCb analysis program) q n Trg/L 1 Decision q n v 12 r 2; v 3 r 1; Activating L 1 decision via option file generates q N-tuple with L 1 information n 1/10/2022 Root macro coming with L 1 package creates all kinds of plots to analyze L 1 performances 12
To check L 1 performances… L 1 summary Min Bias fit Efficiency vs. Retention plots for each subtrigger 1/10/2022 13
More complicated case… Choose second cut first: q Electron q Photon 1/10/2022 14
1/10/2022 15
Advantages of new L 1 code: n Easy to impose bandwidth division q n Single trigger bits accessible to users q n 1/10/2022 user-steerable; Easy to understand why an event triggered; Allows clean implementation of new trigger lines. 16
L 1 open questions for generic subtrigger: n Multiple PVs treatment q n L 1 Generic decision function q q q 1/10/2022 Is current solution the best one? S (log PT) log (S PT) Weighted PT 2? PT 3? Is 400 Me. V/C a good value for tracks with no measured PT? 17
Control channels: 1/10/2022 n Bd p+ p- 2 high PT tracks n Bs Ds K 1 high PT track n Bd D 0(Kp)K* no high PT track 18
Multiple PVs n No big differences q NOW n # PVs > 2 q q q Veto on # PVs? 1/10/2022 current solution seems to be the best Not enough statistics at current luminosity Higher number of PVS Need some high luminosity data!! Is PV 1 = highest multiplicity vertex enough? 19
Log IPS 1 + Log IPS 2 TDR L 1 Generic Signal Min Bias D Log PT 1 + Log PT 2 L 1 Variable = D (2 d distance from L 1 line) 1/10/2022 20
DC 04 L 1 Generic n Improved vertexing q n Simpler cut: q 1/10/2022 No need for IPS; Log PT 1 + Log PT 2 21
• Log (PT 1+PT 2) • Log PT 1 + Log PT 2 = Log(PT 1*PT 2) • Do we need PT 3? GENERIC • Do we need a weighted PT 2? GENERIC NOW Weight 1/10/2022 22
Default PT for tracks with no measured PT NOW 1/10/2022 23
Current solution is OK, but do we really understand it? n n Are default Pt = 400 Me. V/c and L 1 = S log disentangled? PT 1, PT 2 dependence not so clear q 1/10/2022 Try two separate thresholds for PT 1 and PT 2 and play with it 24
Status and plans 1/10/2022 n L 1 shows good (and unexpected) performances; n Still some open issues; n Start thinking about systematics (high luminosity, beam background) and online issues 25
- Slides: 25