1. Missing B-tracks in L1 trigger
13 April 2019
Federica Legger

2. Summary Motivations Method used L1 flow and main failures
Breakdown for some representative channels
Results
A case study: Bd  p+ p-
L1 failures correlation
Origin of the decision tracks
Status and plans
13 April 2019

3. Motivations Post-TDR study to better understand L1 behaviour
LHCb trigger TDR (September 2003)
TDR L1 efficiencies…
or inefficiencies: why do we lose 37.3% of the events for Bd  p+ p- (easy channel)??
Some bugs corrected after TDR
Post-TDR study to better
understand L1 behaviour
13 April 2019

4. What I did…(I)
Take some representative channels, and select events which pass L0 and the offline selection ( reconstructible), but that are rejected by L1
Hadronic
Muon
Electro
magnetic
13 April 2019

5. …what I did… (II) Look for signal decay tree (MC Truth);
Consider B daughter which is more likely to trigger:
In the cases where there are more than one particle with the same probability to trigger (Bd  p+ p-), I consider the one with the highest Pt;
Bd  p+ p-
Bs  Ds k
Bd  F Ks
Bs  J/ (m+ m-) F
Bs  J/ (e+ e-) F
Bd  K* g
13 April 2019

6. …what I did (III) Associate MC particle to L1 2D and 3D tracks;
Look what happened to signal track and see why it didn’t trigger;
Classify events according to trigger failures;
Study of L1 failures correlations in Bd  p+ p- (both pions taken in account).
13 April 2019

7. L1 flow (I) r z (r,z) plane Match L0 m to 2D tracks (2<16)
VELO hits
(r,z) plane z r NO
2D Reco
YES
2D PV,
IP2D
Match L0 m to 2D tracks (2<16)
IP2D cut
.15 <IP2D< 3 mm NO NO
Muon
YES
YES
3D Reco NO
YES
Track
rejected
3D PV,
IP3D
13 April 2019

8. L1 flow (II)
m match to retrieve tracks for dimuon mass bonus (out of scope)
Backward,
Weird IP (-2<IP3D<5mm),
IP3D cut (.15<IP3D<2mm) NO
YES
Match to TT or m (Pt)
If not assign default 400 MeV/c
PT, IPS
One of the two
highest Pt track
(PT1 and PT2)
Decision track NO
YES
Track
rejected
L1 Plane cut
D > L1 equivalent
4% min. bias ret. D
YES NO
L1 YES
L1 NO
13 April 2019

9. L1 code TDR baseline version + some bugs fixed (PVZ @ 100mm)
No bonus system (dimuon mass, electron and g transverse energies)
No multiple PV treatment (implemented in new L1 code)
TDR
now
PVZ
PVZ
13 April 2019

10. L1 flow (I) Main failures Lost@2D Lost@3D 1 2 VELO hits 2D Reco 2D PV,NO
2D Reco
YES
2D PV,
IP2D 1
IP2D cut
.15 <IP2D< 3 mm NO NO
Muon
YES
YES
3D Reco NO
YES 2
Track
rejected
3D PV,
IP3D
13 April 2019

11. L1 flow (II) Main failures Lost@3D Low PT Small D 3 4 Backward,
Weird IP (-2<IP3D<5mm),
IP3D cut (.15<IP3D<2mm) NO
YES
Main failures
PT, IPS
Low PT 3
Decision track NO
YES
Track
rejected
L1 Plane cut
D > L1 equivalent
4% min. bias ret.
Small D 4
YES NO
L1 YES
L1 NO
13 April 2019

12. Main failures To be treated separately Main loss (~60%) ~10% ~10% ~15%
Bd  p+ p-
Bs  Ds K
Bd  F Ks
Bs  J/ (e+e-) F
Bd  K* g
13 April 2019

13. 2. Tracks lost at 2D B daughter MC truth Lost@2D IP2D cut Or Muon
2D track
Correct
Wrong
2. Tracks lost at 2D
13 April 2019

14. Tracks lost at 2D Absolute Relative Main loss ~20% OK Bd  p+ p-
Bs  Ds K
Bd  F Ks
Bs  J/ (e+e-) F
Bd  K* g
13 April 2019

15. 3. Tracks lost at 3D B daughter MC truth Lost@2D IP2D cut Or Muon
2D track
Correct
Wrong
IP3D cut
3D track
3. Tracks lost at 3D
Correct
Wrong
13 April 2019

16. Tracks lost at 3D Absolute Relative Main loss ~2% Bd  p+ p- Bs  Ds K
Bd  F Ks
Bs  J/ (e+e-) F
Bd  K* g
13 April 2019

17. 4. Tracks lost because of low Pt
MC truth
B daughter
Low PT
IP2D cut
Or Muon
Default Pt
400 MeV/c
Measured
low Pt
2D track
Correct
Wrong
Correct
Wrong
4. Tracks lost because of low Pt
IP3D cut
3D track
Correct
Wrong
13 April 2019

18. Tracks lost because of low PT
Absolute
Relative
Main loss (~10%)
insignificant
Bd  p+ p-
Bs  Ds K
Bd  F Ks
Bs  J/ (e+e-) F
Bd  K* g
13 April 2019

19. 5. Tracks lost because of small D
MC truth
5. Tracks lost because of small D
B daughter
Low PT
IP2D cut
Or Muon
Default Pt
400 MeV/c
Measured
low Pt
2D track
Correct
Wrong
Correct
Wrong
Small D
IP3D cut
3D track
PT1
PT2
Correct
Wrong
Correct
Wrong
Correct
Wrong
13 April 2019

20. Tracks lost because of small D
Correct > Wrong
Correct < Wrong
Almost 100%
no measured Pt
13 April 2019

21. Tracks lost because of small D
B track is more likely to be PT1
Bd  p+ p-
Bs  Ds K
Bd  F Ks
Bs  J/ (e+e-) F
Bd  K* g
13 April 2019

22. Summary for all channels but muons
MC truth
Summary for all channels but muons
60%
B daughter
13%
Low PT
12% 3%
IP2D cut
Or Muon
Default Pt
400 MeV/c
Measured
low Pt
2D track
37%
20% 0% 1%
12%
Correct
Wrong
15%
Correct
Wrong
Small D
10%
IP3D cut
3D track
PT1
PT2 1% 1%
10% 2% 2% 1%
Correct
Wrong
Correct
Wrong
Correct
Wrong
13 April 2019

23. L1 Design vs. Resolution failures
(correct boxes)
(wrong+failed)
Best channel
50%, 50%
Enhanced
by bonus...
Bd  p+ p-
Bs  Ds K
Bd  F Ks
Bs  J/ (m+m-) F
Bs  J/ (e+e-) F
Bd  K* g
13 April 2019

24. Results (I) Main failure: Main loss at 3D: IP2D p+ IP2D e+
IP2D cut ~60%, wrong (40%) + correct (~20%)
Main loss at 3D:
failed 3D reco (~8%), worse than failed 2D reco (~3%), could also be the associator (MCParticle2L1Track)
IP cut
IP cut
IP2D p+
IP2D e+
13 April 2019

25. Results (II) IP3D cut not very significative (~2%)
No decision track because Pt is too low (~10%):
no measured Pt (~100%) failed TT reco
Rejected decision tracks (~15%) :
if PTMC we would gain ~5% of the evts.
IP3D p+
IP3D e+
IP cut
13 April 2019

26. The muon channel
L0 Muons (63%) go to 3D reco even if they do not pass the IP2D cut (73%)
10% evts lost at 2D (3% wrong)
Main loss at 3D: IP3D cut (61%)
99% were L0 muons
13% evts wrongly rejected
These tracks could be later use for dimuon mass bonus (out of scope)
Low Pt tracks: 3%
No decision track: 18%
Worst design (71% vs. 29%) BONUS system
13 April 2019

27. A case study: Bd  p+ p- ! L1 failures
Consider the two pions to study failures correlation !
L1 failures
NO2D No 2D track
NO3D No 3D track
IP2D doesn’t pass IP2D cut and is not a m
PT1 Leading track
PT2 2nd leading track
LOWPT Not one of the two leading tracks
IP3D doesn’t pass IP3D cut
13 April 2019

28. A case study: Bd  p+ p- IP2D cut  Main loss is Even if other
track is PT1!!
13 April 2019

29. Origin of the decision tracks
Bd  p+ p-
L0 && !L1 && SEL
L0 && L1 && SEL
13 April 2019

30. PT1
PT2
L0 && !L1 && SEL
L0 && L1 && SEL
13 April 2019

31. Status and plans Study completed; Note with full results almost ready;
Write a Tool to automatically flag trigger failures and include it into new L1 code.
13 April 2019