Mazurov ferrara2014

High Level Trigger software performance profiling andχb production study at the LHCb experiment

Alexander MazurovTutor: Concezio Bozzi

University of Ferrara (Italy) & CERN

7 March 2014

1/46

Plan

1 LHCb experiment2 Measuring Υ(NS) (N=1, 2, 3) fraction originating from χb decays andχb(3P ) mass:

MotivationDatasetsDetermination of Υ yieldsDetermination of χb yields in the following decays:

χb(1, 2, 3P )→ Υ (1S)γχb(2, 3P )→ Υ (2S)γχb(3P )→ Υ (3S)γ

Monte-Carlo efficienciesSystematic uncertaintiesResults

3 Summary4 Backup: HLT profiler

2/46

The LHC Accelerator System

The main LHC design parameters:Circumference 27 km

Center-of-mass energy 14 TeVInjection energy 450 GeV

Field at 2 × 450 GeV 0.535 TField at 2 × 7 TeV 8 T

Helium temperature 2KLuminosity 1034 cm−2s−1

Bunch spacing 25 nsLuminosity lifetime 10 hrTime between 2 fills 7 hr

The LHC operated mainly at center-of-mass energies of 7 TeV and 8 TeV in2011 and 2012, respectively. In 2015, after the long shutdown, LHC is plannedto reach energy of 13 TeV.

3/46

The LHCb experiment (1)

Forward spectrometer with planardetectors:

LHCb uniqueness:

Tracking, RICH and calorimeterscover the full detector acceptance(2.0 < η < 5.0); tracking coveragealso in the backward region(−4.0 < η < −1.5)Covers just 4% of the solid anglebut captures 25% of heavy quarkpairs produced at the LHC.Ability to study low-pT processesat large η

Heavy quark pair production at the LHC:

Fraction of bb̄ pairs in the acceptance:√s ATLAS/CMS LHCb

7 TeV 44% 25%14 TeV 41% 24%

4/46


Excellent tracking performance

High quality particle identification: robust hadron ID +photon/lepton/hadron separation

Selective and flexible trigger system

5/46


LHCb 2010–2012 operation:

Integrated luminosities:0.038 fb−1 in 2010

1.107 fb−1 in 2011

2.082 fb−1 in 2012

Visible interactions:

2× 1014 visible pp interactions

6× 1012 visible cc̄ interactions

3× 1011 visible bb̄ interactions∼93% data taking efficiency∼99% of accumulated data are useful for physics analysesLuminosity leveling: constant and moderate interaction rate throughoutthe data taking periodsSmooth data taking in 2011–2012 regardless high luminosity running

6/46

Study of χb production

7/46

Motivation

Bound bb̄ states, which can be produced in different spin configurations, are anideal laboratory for QCD tests. It’s like a hydrogen atom in QCD.

Measured mass

Mass from theory

States with parallel quark spins (S=1):S-wave Υ state.P-wave χb states, composed by 3 spin statesχb(0,1,2).Υ can be readily produced in the radiative decaysof χb.χb(3P ) state recently observed by ATLAS, D0 andLHCb.

This thesis:1 Measurement of Υ(NS) (N=1, 2, 3) fraction

originating from χb decays as function of pT (Υ).2 Measurement of χb(3P ) mass.

8/46

Previous analysis at LHCb

”Measurement of the fraction of Υ(1S) originating from χb(1P ) in ppcollisions at

√s =7 TeV ”, arXiv:1209.0282,

∫L = 32 pb−1

”Observation of the χb(3P ) state at LHCb in pp collisions at√s =7 TeV

”, LHCb-CONF-2012-020,∫L = 0.9 fb−1.

9/46

In this thesis

The results in this study extend the statistical precision of previous LHCbmeasurements and add considerably more decays and higher transversemomentum regions. The measurement of Υ (3S) fraction in radiative χb(3P )decay was performed for the first time.

In each pT (Υ) bin calculate:

σ(pp→χb(mP )X)×Br(χb(mP )→Υ(nS)γ)σ(pp→Υ(nS)X) =

Nχb(mP )→Υ(nS)γ

NΥ(nS)× εΥ(nS)

εχb(mP )→Υ(nS)γ=

Nχb(mP )→Υ(nS)γ

NΥ(nS)× 1

εrecoγ

Calculate for each Υ(nS), n = 1, 2, 3 and χb(mP ),m = 1, 2, 3

Get N from fits: NΥ from m(µ+µ−) and Nχb→Υγ from[m(µ+µ−γ)−m(µ+µ−)] (for better resolution)

Compute efficiency εrecoγ from Monte-Carlo simulation

10/46

Analysis Plan

1 Datasets2 Determination of Υ yields3 Determination of χb yields in the following decays:

χb(1, 2, 3P )→ Υ (1S)γχb(2, 3P )→ Υ (2S)γχb(3P )→ Υ (3S)γ

4 Monte-Carlo efficiencies5 Systematic uncertainties6 Results

11/46

Datasets

Full 2011 dataset at√s =7 TeV.

∫L = 1 fb−1

Full 2012 dataset at√s =8 TeV.

∫L = 2 fb−1

Monte-Carlo simulation of χb inclusive decays, generated 62× 106

events.

12/46

The Υ selection

Description Requirement

Υ rapidity1 2.0 < yΥ < 4.5Track fit quality χ2/ndf < 4

Track pT > 1 GeV/cPrimary vertex probability > 0.5%

Luminous region |zPV | < 0.5m and x2PV + y2

PV < 100mm2

Kullback-Leibler distance > 5000

Muon and hadron hypotheses ∆ logLµ−h > 0Muon probability ProbNN > 0.5

Trigger lines:L0 L0DiMuon

HLT1 Hlt1DiMuonHighMassHLT2 HLT2DiMuonB

1y = 12

ln(E+pLE−pL

), where pL is the component of the momentum along the beam axis13/46

The Υ fit model

9 10 110

5000

10000

15000

20000

25000

30000

Can

dida

tes/

(40

MeV/c

2)

mµ+µ−[

GeV/c2]

√s = 7 TeV

6 < pµ+µ−

T < 40 GeV/c

Υ (1S)

Υ (2S)

Υ (3S)

3 Double Crystal Ball functionsfor signal yields.

Exponential function forcombinatorial background.

µ+µ− transverse momentum intervals, GeV/c6 – 40

√s = 7 TeV

√s = 8 TeV

NΥ (1S) 283,300 ± 600 659,600 ± 900NΥ (2S) 87,500 ± 400 203,300 ± 600NΥ (3S) 50,420 ± 290 115,300 ± 400

Background 296,400 ± 700 721,300 ± 1100

µΥ (1S), MeV/c2 9457.02 ± 0.10 9455.58 ± 0.07σΥ (1S), MeV/c2 42.86 ± 0.10 43.04 ± 0.06

µΥ (2S), MeV/c2 10,019.03 ± 0.21 10,018.05 ± 0.14σΥ (2S), MeV/c2 46.38 ± 0.20 46.45 ± 0.14

µΥ (3S), MeV/c2 10,351.16 ± 0.32 10,349.41 ± 0.16σΥ (3S), MeV/c2 48.63 ± 0.31 48.24 ± 0.11τ -0.3887 ± 0.0023 -0.3819 ± 0.0015

Υ (1S) mass is about 5 MeV/c2

lower than PDG value9460.30± 0.26 MeV/c2

14/46

Υ yields as function of pT

0 10 20 30 40 50

210

310

410

510

610

0 10 20 30 40 50

210

310

410

510

610

0 10 20 30 40 50

210

310

410

510

610

Eve

nts

pT (Υ) [ GeV/c]

Υ (3S)

Eve

nts

pT (Υ) [ GeV/c]

Υ (1S)

Eve

nts

pT (Υ) [ GeV/c]

Υ (2S)

√s =7 TeV√s =8 TeV



Yields normalized by bin width and luminosity.

The small difference between 7 and 8 TeV data is due to the productioncross-sections, which are expected to be about 10% larger.

15/46

χb selection

In this study photons reconstructed using the calorimeter information.Another approach uses photon conversions in e+e− pairs — this methodhas better invariant mass resolution, but requires more statistics.

Cuts on γ:

Transverse momentum of γ pT (γ) > 600 MeV/cPolar angle of γ in the µ+µ−γ rest frame cos θγ > 0Confidence level of γ cl(γ) > 0.01

Dimuon mass windows:Decay Cut Description

χb(1, 2, 3P )→ Υ(1S) 9310 < µ+µ− < 9600 MeV/c 3σΥ (1S) < µ+µ− < 2.5σΥ (1S) MeV/c

χb(2, 3P )→ Υ(2S) 9870 < µ+µ− < 10090 MeV/c 3σΥ (2S) < µ+µ− < σΥ (2S) MeV/c

χb(3P )→ Υ(3S) 10300 < µ+µ− < 10526 MeV/c σΥ (3S) < µ+µ− < 3σΥ (3S) MeV/c

16/46

χb1,2(1, 2, 3P )→ Υ (1S)γ fit model

10 10.50

500

1000

1500

2000

2500

-4

-2

0

2

4

10 10.50

200

400

600

800

1000

-4

-2

0

2

4

Can

dida

tes/

(20

MeV/c2

)

mµ+µ−γ −mµ+µ− +mPDGΥ (1S)

[GeV/c2

]

√s = 8 TeV

14 < pΥ (1S)T

< 40 GeV/c

Can

dida

tes/

(20

MeV/c2

)


[GeV/c2

]

√s = 7 TeV

14 < pΥ (1S)T

< 40 GeV/c

χb(1P )

χb(2P )

χb(3P )

χb(1P )

χb(2P )

χb(3P )

One Crystal Ball (CB) for eachχb1,2(1P, 2P, 3P ) state: 6 CB intotal

Exclude the study of χb0 due to itslow radiative branching ratio.

Product of exponential and linearcombination of polynomials forcombinatorial background.

17/46

χb1,2(1, 2, 3P )→ Υ (1S)γ fit model (2)

Free parameters: µχb1(1P ), yields andbackground parameters.Linked parameters for χb1 and χb2 signals:

µχb2(jP ) = µχb1(jP ) + ∆mPDGχb2(jP ),

j=1,2µχb2(3P ) = µχb1(3P ) + ∆mtheory

χb2(3P )

Nχb = λNχb1 + (1− λ)Nχb2σχb2 = σχb1

Other linked parameters:µχb1(2P ) = µχb1(1P ) + ∆mPDG

χb1(2P )

µχb1(3P ) = µχb1(1P ) + ∆mχb1(3P )

(∆mχb1(3P ) measured in this study)Fixed parameters from MC study:

σχb1(1P ) scaled by 1.17,σχb1(2P )

σχb1(1P ),σχb1(3P )

σχb1(1P )

α and n parameters of CB.

Υ(1S) transverse momentum intervals, GeV/c14 – 40

√s = 7 TeV

√s = 8 TeV

Nχb(1P ) 2090 ± 80 5070 ± 130Nχb(2P ) 450 ± 50 1010 ± 80Nχb(3P ) 150 ± 40 220 ± 60

Background 8830 ± 130 23,910 ± 210

µχb1(1P ), MeV/c2 9889.7 ± 1.0 9890.3 ± 0.7σχb1(1P ), MeV/c2 22.0 22.5σχb1(2P )/σχb1(1P ) 1.5 1.5σχb1(3P )/σχb1(1P ) 1.86 1.86

τ -2.6 ± 0.5 -3.27 ± 0.30c0 -0.08 ± 0.12 0.07 ± 0.06c1 1.33 ± 0.04 0.29 ± 0.04

χ2/n.d.f 1.03 1.24

χb1(1P ) mass agrees with the PDG value

9892.78± 0.26 (stat)± 0.31 (syst) MeV/c2 .

18/46

Mass of χb1(1P ) in χb → Υ (1S)γ decay

10 20 30 40

9.882

9.884

9.886

9.888

9.89

9.892

9.894

9.896

√s =7 TeV,

√s =8 TeV

χb1

(1P

)m

ass

GeV/c

2

pΥ (1S)T [ GeV/c]

The major cause of χb1(1P ) mass floating in 10 MeV/c2 range can be theunknown fraction between Nχb1 and Nχb2 yields (λ parameter). We have onlytheoretical prediction for λ value . In this study λ is fixed to 0.5 for all χbdecays.

In the following, this mass was fixed to 9.887 GeV/c2 which is the valuemeasured on the combined 2011 and 2012 datasets. A systematic uncertaintydue to this assumption has been assigned.

19/46

χb yields in χb → Υ (1S)γ decays

10 20 30 400

10

20

30

40

50

60

70

80

90

100

10 20 30 400

500

1000

1500

2000

2500

10 20 30 400

100

200

300

400

500

600

700

800

900

1000

Eve

nts

pΥ (1S)T [ GeV/c]

χb(3P )

Eve

nts

pΥ (1S)T [ GeV/c]

χb(1P )

Eve

nts

pΥ (1S)T [ GeV/c]

χb(2P )





The small difference between 7 and 8 TeV data is due to the productioncross-sections, which are expected to be about 10% larger..

20/46

χb1,2(2, 3P )→ Υ (2S)γ fit model

10.2 10.4 10.6 10.8 110

100

200

300

400

500

600

-4

-2

0

2

4

10.2 10.4 10.6 10.8 110

20406080

100120140160180200220

-4

-2

0

2

4

Can

dida

tes/

(20

MeV/c2

)


[GeV/c2

]

√s = 8 TeV

18 < pΥ (2S)T

< 40 GeV/c

Can

dida

tes/

(20

MeV/c2

)


[GeV/c2

]

√s = 7 TeV

18 < pΥ (2S)T

< 40 GeV/c

χb(2P )

χb(3P )

χb(2P )

χb(3P )

One Crystal Ball (CB) for eachχb1,2(2P, 3P ) state: 4 CB in total



21/46

χb1,2(2, 3P )→ Υ (2S)γ fit model (2)

Free parameters: µχb1(2P ), yields andbackground parameters.Linked parameters for χb1 and χb2 signals:

µχb2(2P ) = µχb1(2P ) + ∆mPDGχb2(2P )

µχb2(3P ) = µχb1(3P ) + ∆mtheoryχb2(3P )


Other linked parameters:µχb1(3P ) = µχb1(1P ) + ∆mχb1(3P )

(∆mχb1(3P ) measured in this study)Fixed parameters from MC study:

σχb1(2P ),σχb1(3P )

σχb1(2P )

α and n parameters of CB.


√s = 7 TeV

√s = 8 TeV

Nχb(2P ) 237 ± 29 650 ± 50Nχb(3P ) 50 ± 17 78 ± 26

Background 1830 ± 50 4600 ± 80

µχb1(2P ), MeV/c2 10,249.1 ± 2.2 10,249.9 ± 1.3σχb1(2P ), MeV/c2 13.0 13.3σχb1(3P )/σχb1(2P ) 1.65 1.65

τ -7.5 ± 0.8 -7.7 ± 0.5c0 0.431 ± 0.027 0.435 ± 0.016c1 -2.07 ± 0.09 -2.12 ± 0.05c2 0.79 ± 0.36 0.79 ± 0.17

χ2/n.d.f 0.98 1.35

The mass of χb1(2P ) is about 6 MeV/c2 less than

value in PDG (10.25546 GeV/c2).

22/46

Mass of χb1(2P ) in χb → Υ (2S)γ decay

20 25 30 35 4010.235

10.24

10.245

10.25

10.255

10.26

10.265

√s =7 TeV,

√s =8 TeV

χb1

(2P

)m

ass

GeV/c

2

pΥ (2S)T [ GeV/c]

In the following, this mass was fixed to 10.249 GeV/c2 which is the valuemeasured on the combined 2011 and 2012 datasets. A systematic uncertaintydue to this assumption has been assigned.

23/46

χb yields in χb → Υ (2S)γ decays

0 10 20 30 40 500

10

20

30

40

50

60

0 10 20 30 40 500

1

2

3

4

5

6

7

8

9

10

Eve

nts

pΥ (2S)T [ GeV/c]

χb(2P )

Eve

nts

pΥ (2S)T [ GeV/c]

χb(3P )√s =7 TeV√s =8 TeV



24/46

χb1,2(3P )→ Υ (3S)γ fit model

10.5 10.6 10.70

5

10

15

20

25

30

35

40

45

-4

-2

0

2

4

10.5 10.6 10.70

2

4

6

8

10

12

14

16

18

-4

-2

0

2

4

Can

dida

tes/

(20

MeV/c2

)


[GeV/c2

]

√s = 8 GeV

27 < pΥ (3S)T

< 40 GeV/c

Can

dida

tes/

(20

MeV/c2

)


[GeV/c2

]

√s = 7 GeV

27 < pΥ (3S)T

< 40 GeV/c

χb(3P )

χb(3P )

One Crystal Ball (CB) for eachχb1,2(3P ) state: 2 CB in total



25/46

χb1,2(3P )→ Υ (3S)γ fit model (2)

Free parameters: µχb1(3P ),σχb1(3P ), yields, backgroundparameters.Linked parameters for χb1 and χb2signals:

µχb2(3P ) =

µχb1(3P ) + ∆mtheoryχb2(3P )


Fixed parameters from MC study:α and n parameters of CB.


√s = 7 TeV

√s = 8 TeV

Nχb(3P ) 31 ± 12 72 ± 16

Background 97 ± 14 283 ± 21

µχb1(3P ), MeV/c2 10,517 ± 4 10,504.0 ± 2.5σχb1(3P ), MeV/c2 9 ± 6 8.3 ± 2.7

c0 0.52 ± 0.18 0.52 ± 0.09c1 -0.42 ± 0.19 -0.36 ± 0.10c2 1.3 ± 0.8 -1.23 ± 0.18

χ2/n.d.f 0.38 1.09

In this study the mass of χb1(3P ) wasfixed to 10.508 GeV/c2 which is thevalue measured on the combined 2011and 2012 datasets.

26/46

Mass of χb1(3P ) in χb → Υ (3S)γ decay (1)

Systematic uncertanties due to unknown ratio and mass difference between χb1and χb2 states:

0 0.5 1

10.5

10.502

10.504

10.506

10.508

10.51

10.512

10.514

10.516

Mas

sofχb1(3P

)(G

eV/c2

)

λ

∆mχb1,2(3P ) = 13 MeV/c2

∆mχb1,2(3P ) = 10 MeV/c2

The mass is measured with different ratios (λ) and mass difference (∆mχb1,2(3P )) between χb1(3P )

and χb2(3P ) states. The measurement is performed on the combined 2011 and 2012 datasets.

27/46

Mass of χb1(3P ) in χb → Υ (3S)γ decay (2)

The measured mχb1(3P )=10,508± 2 (stat)± 8 (syst) MeV/c2 is consistentwith the mass measured in another study with converted photons —10,510± 3.0 (stat)+4.4

−3.4 (syst) MeV/c2 (very preliminary results).

ATLAS measured χb1 and χb2 mass barycenter formχb2 −mχb1 = 12 MeV/c2 and λ = 0.5:mχb(3P ) = 10,530± 5 (stat)± 9 (syst) MeV/c2

D0: mχb(3P ) = 10,551± 14 (stat)± 17 (syst) MeV/c2

28/46

MC efficiency (1)

MC true events χb(3P )→ Υ (1S)γ (other decays have the same shape)

0 0.5 1 1.5 20

500

1000

1500

2000

2500

Can

dida

tes

mµ+µ−γ −mµ+µ−[

GeV/c2]

The flat left tails are due to photons which, although being correctly associated to the χb decay,are poorly reconstructed in the calorimeter. In principle, these tails could be modeled in our fit tosignal, but in practice they will not be distinguishable from background. Therefore, the numberof χb events for efficiency calculations is not determined by simple event counting but froma fit where the tails are considered as background. Υ events are obtained by counting mc-trueevents.

29/46

Monte-Carlo photon reconstruction (2)

Example of fits:

9.8 9.9 10 10.10

1000

2000

3000

4000

5000

6000

-4

-2

0

2

4

χb1(1P )→ Υ (1S)γ

6 < pΥ (1S)T < 40 GeV/c


[GeV/c2

]

Can

dida

tes/

(10

MeV/c2

)

10 10.2 10.4 10.60

500

1000

1500

2000

2500

-4

-2

0

2

4

χb1(2P )→ Υ (1S)γ

6 < pΥ (1S)T < 40 GeV/c


[GeV/c2

]C

andi

date

s/(1

0M

eV/c2

)

10.2 10.4 10.6 10.80

200

400

600

800

1000

1200

1400

-4

-2

0

2

4

χb1(3P )→ Υ (1S)γ

6 < pΥ (1S)T < 40 GeV/c


[GeV/c2

]

Can

dida

tes/

(10

MeV/c2

)

10.2 10.25 10.30

10

20

30

40

50

-4

-2

0

2

4

χb1(2P )→ Υ (2S)γ

18 < pΥ (2S)T < 40 GeV/c


[GeV/c2

]

Can

dida

tes/

(10

MeV/c2

)

10.4 10.5 10.60

20

40

60

80

100

120

-4

-2

0

2

4

χb1(3P )→ Υ (2S)γ

18 < pΥ (2S)T < 40 GeV/c


[GeV/c2

]

Can

dida

tes/

(10

MeV/c2

)

10.45 10.5 10.55 10.6 10.650

10

20

30

40

50

60

70

80

-4

-2

0

2

4

χb1(3P )→ Υ (3S)γ

27 < pΥ (3S)T < 40 GeV/c


[GeV/c2

]

Can

dida

tes/

(10

MeV/c2

)

30/46

Data — Monte Carlo comparison

A comparison of the distribution of the relevant observables used in thisanalysis was performed on real and simulated data, in order to assess thereliability of Monte Carlo in computing efficiencies

0 0.2 0.4 0.6 0.8 10

0.01

0.02

0.03

0.04

0.05

0.06

0 0.2 0.4 0.6 0.8 10

0.01

0.02

0.03

0.04

0.05

0.06

0 0.2 0.4 0.6 0.8 1

-0.02

0

0.02

0.04

0.06

0.08

0.1

0 2 4

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0 2 4

0

0.02

0.04

0.06

0.08

0 2 4-0.02

0

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0 10 20 300

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0 10 20 30

0

0.02

0.04

0.06

0.08

0.1

0 10 20 30

-0.02

0

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0.16

15 20 25 30 350

0.01

0.02

0.03

0.04

0.05

0.06

15 20 25 30 35

0

0.02

0.04

0.06

0.08

0.1

15 20 25 30 35

0

0.05

0.1

0.15

0.2

γ confidence level γ confidence level γ confidence level

χ2 of decay tree fitter χ2 of decay tree fitter χ2 of decay tree fitter

pT [χb(1P )][

GeV/c2]

pT [χb(2P )][

GeV/c2]

pT [χb(3P )][

GeV/c2]

pT [Υ (1S)][

GeV/c2]

pT [Υ (1S)][

GeV/c2]

pT [Υ (1S)][

GeV/c2]

Arb

itrar

yun

its

Arb

itrar

yun

its

Arb

itrar

yun

its

Arb

itrar

yun

its

Arb

itrar

yun

its

Arb

itrar

yun

its

Arb

itrar

yun

its

Arb

itrar

yun

its

Arb

itrar

yun

its

Arb

itrar

yun

its

Arb

itrar

yun

its

Arb

itrar

yun

its

χb(1P ) χb(2P ) χb(3P )

χb(1P ) χb(2P ) χb(3P )

χb(1P ) χb(2P ) χb(3P )

χb(1P ) χb(2P ) χb(3P )

The agreement is generally very good.

31/46

Monte-Carlo photon reconstruction efficiency

20 30 400

5

10

15

20

25

30

Effi

cien

cy,%

pΥ (2S)T [ GeV/c]

χb(3P )→ Υ(2S)γ√s =7 TeV√s =8 TeV

20 25 30 35 400

2

4

6

8

10

12

14

16

18

20

22

Effi

cien

cy,%

pΥ (3S)T [ GeV/c]


10 20 30 400

5

10

15

20

25

Effi

cien

cy,%

pΥ (1S)T [ GeV/c]


20 30 400

5

10

15

20

25

Effi

cien

cy,%

pΥ (2S)T [ GeV/c]


10 20 30 400

5

10

15

20

25

30

Effi

cien

cy,%

pΥ (1S)T [ GeV/c]


10 20 30 400

5

10

15

20

25

30

Effi

cien

cy,%

pΥ (1S)T [ GeV/c]


Photon is more energetic as pT (Υ)increases so it is reconstructed moreefficiently.

32/46

Systematic uncertainties

Since this analysis measures the fraction of Υ(nS) particles originating fromχb decays, most systematic uncertainties cancel in the ratio and only residualeffects need to be taken into account.

Systematic uncertainties on the event yields are mostly due to the fit model of Υand χb invariant masses, while the ones on the efficiency are due to the photonreconstruction and the unknown initial polarization of χb and Υ particles.

33/46

χb yields systematic uncertainties. Unknown σχb2/σχb1ratio (1)

The σ(χb2)/σ(χb1) ratio prediction:

CDFLHCbΧb scaledΧb

Χc

CMS preliminary

5 10 15 20 25 300.0

0.5

1.0

1.5

2.0

2.5

3.0

pT, GeV

σ(χ

2)/σ

(χ1)

Transverse momentum distributions of the dσ [χ2] /dσ[χ1] ratio. Solid and dashed lines stand

for charmonium and bottomonium mesons. The dot-dashed line corresponds to the rescaled

bottomonium ratio: σb2/σb1(Mχc/MχbpT ). The experimental results for charmonium from

LHCb are shown with dots, CDF — with rectangles, and CMS — with triangles.

34/46

χb yields systematic uncertainties. Unknown σχb2/σχb1ratio (2)

χb yields systematic uncertainties (%) related to unknown λ = Nχb1/(Nχb1 +Nχb2) ratio inthe fit model for χb(1, 2, 3P )→ Υ (1S)γ decays.

Systematic uncertanties are measured in each pΥT bin and center-of-mass energy.

Υ(1S) transverse momentum intervals, GeV/c6 – 8 8 – 10

√s = 7 TeV

√s = 8 TeV

√s = 7 TeV

√s = 8 TeV

Nχb(1P ) Nχb(2P ) Nχb(3P ) Nχb(1P ) Nχb(2P ) Nχb(3P ) Nχb(1P ) Nχb(2P ) Nχb(3P ) Nχb(1P ) Nχb(2P ) Nχb(3P )

λ = 0.0 13.3 -8.3 — 10.1 -12.9 — 19.3 -8.7 — 15.9 -7.3 —λ = 0.1 9.0 -6.7 — 6.1 -8.9 — 13.9 -7.0 — 11.1 -5.9 —λ = 0.2 5.4 -5.2 — 2.9 -5.4 — 9.1 -5.2 — 6.9 -4.2 —

λ = 0.3 2.5 -3.4 — 0.8 -3.9 — 4.9 -3.2 — 3.6 -2.6 —λ = 0.4 0.7 -1.6 — -0.3 -1.5 — 1.9 -1.5 — 1.2 -1.1 —λ = 0.5 0.0 0.0 — 0.0 0.0 — 0.0 0.0 — 0.0 0.0 —λ = 0.6 0.8 1.5 — 1.3 0.7 — -0.7 1.2 — 0.1 0.8 —λ = 0.7 2.7 1.6 — 4.1 -2.2 — -0.1 1.9 — 1.4 1.0 —

λ = 0.8 6.1 0.8 — 7.1 -8.3 — 1.6 2.2 — 3.7 0.9 —λ = 0.9 9.4 -1.3 — 10.9 -17.2 — 4.3 2.0 — 7.1 0.5 —λ = 1.0 13.9 -5.8 — 14.7 -24.4 — 7.9 1.7 — 11.2 -0.4 —

Υ(1S) transverse momentum intervals, GeV/c10 – 14 14 – 18

√s = 7 TeV

√s = 8 TeV

√s = 7 TeV

√s = 8 TeV

Nχb(1P ) Nχb(2P ) Nχb(3P ) Nχb(1P ) Nχb(2P ) Nχb(3P ) Nχb(1P ) Nχb(2P ) Nχb(3P ) Nχb(1P ) Nχb(2P ) Nχb(3P )

λ = 0.0 15.8 -3.1 36.1 13.7 -13.6 36.2 6.8 -0.9 5.3 7.3 -2.0 -9.5λ = 0.1 10.7 -2.6 25.9 5.5 -7.0 -0.6 4.2 -1.0 5.5 4.7 -1.9 -5.7λ = 0.2 6.3 -1.7 16.6 5.2 -6.3 16.8 2.2 -1.0 5.1 2.5 -1.6 -3.1

λ = 0.3 3.0 -0.9 9.0 2.6 -3.4 9.3 0.9 -0.9 3.9 1.1 -1.2 -1.2λ = 0.4 0.9 -0.2 3.6 0.2 -0.9 0.9 0.1 -0.5 2.1 0.2 -0.6 -0.0λ = 0.5 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0λ = 0.6 0.4 -0.2 -2.1 0.6 0.6 -1.5 0.7 0.3 -6.2 0.5 0.9 -0.7λ = 0.7 1.9 -1.0 -2.9 2.5 0.3 1.4 1.6 2.2 -4.8 1.6 2.0 -1.8

λ = 0.8 4.4 -2.0 -3.0 4.1 -0.2 -4.8 3.1 4.4 -3.5 3.2 3.4 -3.4λ = 0.9 7.8 -3.3 -1.8 9.0 -1.6 7.5 5.4 5.7 -9.6 5.5 5.2 -4.7λ = 1.0 12.0 -4.8 0.4 13.3 -3.1 9.4 8.1 8.1 -11.4 8.3 7.3 -5.9

Theory predicts that λ belongs to the range between 0.4 and 0.7.

35/46

Systematic uncertainties due to χb polarization

Efficiencies are evaluated on MC where χb particles are unpolarized. To evaluatesystematic effects due to the unknown polarization of χb, MC events are reweighted asdescribed in HERA-B Collaboration, I. Abt et al., Production of the CharmoniumStates χc1 and χc2 in Proton Nucleus Interactions at s = 41.6-GeV, arXiv:0807.2167and LHCb collaboration, R. Aaij et al.,Measurement of the cross-section ratioσ(χc2)/σ(χc1) for prompt χc production at

√s = 7 TeV, arXiv:1202.1080

For each simulated event in the unpolarised sample, a weight is calculated fromthe distribution of the following angles in the various polarisation hypothesescompared to the unpolarised distribution.

ΘΥ angle between the directions of the µ+ in the Υ rest frame and theΥ in the χb rest frame.

Θχb angle between the directions of the Υ in the χb rest frame and χbin the lab frame.

φ angle between the Υ decay plane in the χb rest frame and the planeformed by χb direction in the lab frame and the direction of the Υin the χb rest frame.

Two hypotheses for χb1 state (w0, w1) and three hypotheses for χb2 (w0, w1,w2).

36/46

http://arxiv.org/abs/0807.2167

http://arxiv.org/abs/1202.1080

Systematic uncertainties due to χb polarization

The ratio of efficiency for unpolarized χb to efficiency for polarized χb.

10 20 30 40

0.99

0.995

1

1.005

1.01

1.015

1.02

1.025

1.03

1.035

10 20 30 40

0.985

0.99

0.995

1

1.005

1.01

pΥ (1S)T [ GeV/c] p

Υ (1S)T [ GeV/c]p

Υ (1S)T [ GeV/c] p

Υ (1S)T [ GeV/c]

ε mχb1/εunpol

ε mχb1/ε unpol

χb1(1P )→ Υ (1S)γ

|mχb1 | = 0

χb1(1P )→ Υ (1S)γ

|mχb1 | = 1

The difference between ratio and 1 is considered as systematic uncertainty.

37/46

Summary of systematic uncertainties

Υ fraction uncertainties common to all χb decays (%)

Υ fit model ±0.7γ reconstruction ±3

Summary of Υ fraction systematic uncertainties (%)(maximum deviations that were found in pΥ

T bins):χb fit model χb polarization

χb(1P )→ Υ (1S)γ +4.3−5.8

+5.1−4.0

χb(2P )→ Υ (1S)γ +4.8−6.2

+5.8−6.8

χb(3P )→ Υ (1S)γ +19.6−16.6

+6.9−6.7

χb(2P )→ Υ (2S)γ +2.3−7.0

+8.7−7.8

χb(3P )→ Υ (2S)γ +19.7−19.9

+4.5−4.2

χb(3P )→ Υ (3S)γ +20.9−27.6

+6.4−7.5

38/46

Υ fractions in χb → Υγ decays

10 20 30 400

5

10

15

20

25

30

35

40

45

50

Υ(1S

)fr

actio

n,%

pΥ (1S)T [ GeV/c]

χb(1P )→ Υ (1S)γ√s =7 TeV√s =8 TeV

10 20 30 400

5

10

15

20

25

30

35

40

45

50

Υ(1S

)fr

actio

n,%

pΥ (1S)T [ GeV/c]


10 20 30 400

5

10

15

20

25

30

35

40

45

50

Υ(1S

)fr

actio

n,%

pΥ (1S)T [ GeV/c]


10 20 30 400

5

10

15

20

25

30

35

40

45

50

Υ(2S

)fr

actio

n,%

pΥ (2S)T [ GeV/c]


10 20 30 400

5

10

15

20

25

30

35

40

45

50

Υ(2S

)fr

actio

n,%

pΥ (2S)T [ GeV/c]


10 20 30 400

10

20

30

40

50

60

70

80

90

100

Υ(3S

)fr

actio

n,%

pΥ (3S)T [ GeV/c]


10 20 30 400

5

10

15

20

25

30

35

40

45

50

Υ(1S

)fr

actio

n,%

pΥ (1S)T [ GeV/c]

χb(1P )→ Υ (1S)γ√s =7 TeV√s =8 TeV√s =7 TeV (2010)

Outer error bars show statistical and systematics

errors, inner error bars — only statistical errors.

39/46

Summary

Measured fractions of Υ (1, 2, 3S) originated from χb decays. About 40%of Υ come from χb, with mild dependence on Υ transverse momentum.

This analysis improves significantly the statistical precision of theprevious work and adds more decays and transverse momentum regions.

The LHCb detector design allows to perform measurements in Υ rapidityand transverse momentum regions, which are complementary to the onesexploited by ATLAS and CMS.

Measured mass of χb(3P ) is 10, 508± 2 (stat)± 8 (stat) MeV/c2,consistent with another determination which uses converted photons.

A software profiling tool was developed.

The results on χb production were regularly presented at the LHCbbottomonium working group, an internal document was prepared by the author,is currently under review and will form the basis of a future LHCb publication.

40/46

High Level Trigger softwareperformance profiling

41/46

Optimizing the LHCb trigger is crucial in many measurement. In particular,decays with muons can be triggered on with unprecedented high efficiency(∼80%).

The dimuon trigger:

is not prescaled.

is used in more than 50% ofLHCb’s papers.

42/46

Trigger software profiling

The trigger needs fastalgorithms!

Most experiments require a trigger in order torecord interesting events at suitable rate.

L0 Hardware Trigger 40MHz → 1MHz.Search for high pt, µ, e, γ, hadroncandidates.High Level Software Trigger Farm

HLT1: Add Impact parameter cutsHLT2: Global event reconstruction.

100 man/years work that has only 20-30ms to process an average event.

29K CPUs or 1700 servers.

2× 1015 bytes stored in 2012.

43/46

CPU Profiler Auditor

CPU profiler tool is vital for trigger optimization.

C++ library.

Deployed into the core softwareframework in LHCb — Gaudi.

Based on Intel R©

VTuneTMAmplifier XE User API

Full and clear reports.

44/46

CHEP2012: Talk and Paper

This work was presented on “Computing in High Energy and Nuclear Physics”conference. The largest of the conferenes in this area. The conference is heldevery 18 months.

A. Mazurov and B. Couturier, “Advancedmodular software performancemonitoring”, Journal of Physics:Conference Series 396 (2012), no.5052054.

45/46

Thank you!

46/46

Mazurov ferrara2014

Education

Transcript of Mazurov ferrara2014