THE LHCB TRACKS RECONSTRUCTION INRUN 2: STRATEGY AND PERFORMANCELaurent Dufour1, Renata Kope�ná2, Alex Pearce3, Maarten van Veghel1on behalf of the LHCb collaboration

1Nikhef National Institute for Subatomic Physics, the Netherlands2Physikalisches Institut, Heidelberg University, Germany3CERN, Switzerland

39th ICHEP, Seoul, July 2018

LHCB DETECTOR1 [1] [2]

RECONSTRUCTION IMPROVEMENTS IN RUN 2⇤ Major improvements in the pattern recognition algorithms⇤ Factor of 2 speed-up, without loss in performance⇤ Extensive use of machine-learning techniques⇤ Example: Improved fake-track reduction⇤ Fake tracks: not corresponding to a real particle’s trajectory⇤ Sped up by O(90%): allow the use in the online event selection⇤ Ghost rate in online event selection reduced from 22% to 14%

2

]2 [MeV/cπKm1800 1850 1900

)2ca

ndid

ates

/ (4

MeV

/c

0

500

1000

1500

2000

2500

3000

3500

LHCbpreliminary

π K→0all Dfakes

[4]

MEASUREMENT OF TRACKING EFFICIENCY⇤ All data selected and stored in software trigger: “TurCal” [5]⇤ O�ine-quality event reconstruction available online⇤ Detector calibration ran online: o�ine-quality data⇤ Data immediately available for analyses

⇤ E�ciency from data using tag-and-probe method⇤ Track parameter inference independent of standard track re-

construction⇤ Exploit two-particle decays⇤ First track from the decay is fully reconstructed (tag track)⇤ Second track is reconstructed excluding the probed subdetec-

tor (probe track)

⇤ Tracking e�ciencies easily accessible using “TrackCalib” tool⇤ Transparent and easy e�ciency estimation⇤ User-defined track-quality cuts, binning and variables⇤ Available for all users

MUON TRACKING EFFICIENCY⇤ J/ ! µ+µ� decay is used [3]⇤ µµ pair ideal to probe the whole tracking system⇤ J/ gives a clear signal in the detector

⇤ J/ coming from B+ ! J/ X (data) and B+ ! J/ K (MC)⇤ Detached J/ for clean signal⇤ Unbiased selection by dedicated trigger

Long Method T-station method Velo method

Final method = Long method + (VELO method � T-station method)

3

η2 2.5 3 3.5 4 4.5

ε

0.86

0.88

0.9

0.92

0.94

0.96

0.98

1

1.02

1.04 LHCb preliminary

Data (50ns) 2015

Data (50ns) 2012

]c [GeV/p20 40 60 80 100 120 140 160 180 200

ε

0.86

0.88

0.9

0.92

0.94

0.96

0.98

1

1.02

1.04 LHCb preliminary

Data (50ns) 2015

Data (50ns) 2012

ELECTRON TRACKING EFFICIENCYElectrons have a considerably di�erent behaviour compared to muons⇤ Significant energy loss (bremsstrahlung) along trajectory⇤ Decreases reconstruction e�ciency downstream of VELO

New method for measuring reconstruction e�ciencies for electrons:Kinematically constrained VELO tracks originating fromB+ ! J/ (! e+e�)K+

⇤ E�cient VELO reconstruction(⇡98%)

⇤ Applicable also to muons &hadrons

)2 c

Can

dida

tes /

( 5

MeV

/

LHCb simulation preliminary

100

200

300

400

500

600

700

800

900

]2c[MeV/±K�/Jm5200 5400 5600

Kinematics⇤ Direction inferred from VELO

segment⇤ Probe momentum inferred

from J/ mass constraint⇤ Use B+ mass with J/ mass

constraint to distinguishbetween signal andbackground

REFERENCES[1] A. A. Alves Jr. et al. [LHCb collaboration], JINST 3 (2008) S08005[2] R. Aaij et al. [LHCb collaboration], Int. J. Mod. Phys. A30 (2015) 1530022[3] R. Aaij et al. [LHCb Collaboration], JINST 10, no. 02, P02007 (2015)[4] M. De Cian et al., LHCb-PUB-2017-011 (2017)[5] S. Benson et al., J. Phys.: Conf. Ser. 664 (2015) 082004