Quarkonia in Jets - University of Birmingham · 2018. 5. 9. · Quarkonia in Jets Philip Ilten May 9, 2018 Birmingham Seminar Ilten Quarkonia in Jets 1 / 33

Quarkonia in Jets

Philip Ilten

May 9, 2018

Birmingham Seminar

Ilten Quarkonia in Jets 1 / 33

Introduction

QCD at the LHC

g

q̄ q

q q̄

p

η

ρ

φ

ω

n

Λ

p

p

100 GeV 1 GeV14 TeV

energyIlten Quarkonia in Jets 2 / 33

Introduction

QCD at the LHC

g

q̄ q

q q̄

p

η

ρ

φ

ω

n

Λ

p

p100 GeV 1 GeV14 TeV

energy

final stateshowers

hadronisation


Introduction

QCD at the LHC

g

q̄ q

q q̄

p

η

ρ

φ

ω

n

Λ

p

p100 GeV 1 GeV14 TeV

energy

partondistributionfunctions

initial stateshowers


Introduction

Factorising QCD

z = Eg/Eqθ

q̄

q

g

γ

e−

e+

q̄

g

q

γ

e−

e+

q̄

g

q

dσ ≈ σ(2 dcos θ

sin2 θ

)(αs2π

)(N 2c − 12Nc

)(1 + (1− z)2

z

)dz

• factorise into general form given any splitting kernel Pi

dσ ≈ σ∑

i

dθ2

θ2 Pi (z, αs) dz

• diverges when collinear (θ → 0, π) or infrared (z → 0)


Introduction

Sudakovs and Splitting Kernels

∆(Q21 ,Q2) = exp

[−∫ Q2

1

Q2

1q2

∫ 1−Q20/q2

Q20/q2

Pi(z, αs) dz dq2]

1 pick a random number r ∈ [0, 1]2 solve ∆(Q2

1 ,Q2) = r for Q2

3 if Q > Q0 generate emission and repeat from 14 if Q ≤ Q0 terminate shower

g → gg q → qg g → qq

1− zz + z

1− z +z(1−z) 1− zz + z

2 − 2µ z2 + (1− z)2 + µ2


Introduction

Reverse Engineering with Jets

• unfold final state particles to initial hard partons1 collinear safe → collinear emission changes nothing2 infrared safe → soft emission changes nothing3 insensitive to non-perturbative effects4 applicable to both parton and hadron level

• inclusive sequential clustering is algorithm of choice at LHC

dĳ = min(pkTi , pk

Tj)∆R2

ĳR2 , diB = pk

Ti

1 select minimum d2 if dĳ , combine particle i and j3 if diB, consider particle as jet and remove from clustering4 terminate if no particles otherwise return to 1


Introduction

Flavours of Sequential Clustering

Cambridge/Aachen kt anti-kt

arXiv:1111.6097

k = 0 k = 2 k = −2

• Cambridge/Aachen considers only geometry• kt and anti-kt also consider momentum• anti-kt provides circular jets in R at high-pT


http://arxiv.org/abs/1111.6097v1

Tagging a Jet

Tagging a Jet

LHCb, JINST 10 (2015) P06013


https://arxiv.org/abs/1504.07670

Tagging a Jet

From Theory to Detector

?

primary vertex

secondary vertex

tag

theory detector

• jet properties depend on initiating parton

c-hadron b-hadronmass 2 GeV 5 GeVlifetime (cτ) 0.1 mm 0.5 mmmultiplicity ≈ 2 ≈ 3


Tagging a Jet

Enter LHCb

VELO

RICH1

TT

magnet

OT

IT

RICH2

HCAL

ECALmuon system5 m

5 m 15 m

SPD/PRS

collisionpoint

1 good momentum and massresolution

2 excellent secondary vertexresolution


Tagging a Jet

Tagging Ingredients

T(SV)/p (jet)

Tp

0 0.2 0.4 0.6 0.8 1

A.U.

0

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0.16

0.18bcudsg

corrected mass [GeV]0 2 4 6 8 10

A.U.

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

bcudsg

tracks0 5 10

A.U.

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

bcudsg

Mcor =√

M 2 + p2 sin2 θ+p sin θ

secondary

vertex

θ

p


Tagging a Jet

A Pinch of Machine Learning

• 10 total observables (3 jet related) input to boosted decision trees• two BDTs: udsg vs. cb and c vs. b• fit 2-dimensional distribution of the two BDTs

)udsg|bcBDT(-1 -0.5 0 0.5 1

)c|bB

DT

(

-1

-0.5

0

0.5

1

LHCb simulation

-jetsudsg

)udsg|bcBDT(-1 -0.5 0 0.5 1

)c|bB

DT

(

-1

-0.5

0

0.5

1

LHCb simulation

-jetsc

)udsg|bcBDT(-1 -0.5 0 0.5 1

)c|bB

DT

(

-1

-0.5

0

0.5

1

LHCb simulation

-jetsb


Tagging a Jet

Tag and Probe

)c|bBDT(-1 -0.5 0 0.5 1

N(je

ts)

0

2000

4000

6000 LHCbdatabcudsg

0

200

400

600

800

1000

)udsg|bcBDT(-1 -0.5 0 0.5 1

)c|bBD

T(

-1

-0.5

0

0.5

1LHCb data

+jetD

)udsg|bcBDT(-1 -0.5 0 0.5 1

N(je

ts)

0

2000

4000

6000

8000 LHCbdatabcudsg

SV probe?

exclusive tag


Tagging a Jet

Some Cool Measurements

µ

µ

cg

c

Z

p

extrinsic c

intrinsic c

simulation

• W + jets: LHCb, Phys. Rev. D 92 (2015)• top production: LHCb, Phys. Rev. Lett. 115 (2015)• IC: Boettcher, Ilten and Williams, Phys. Rev. D 93 (2016)





J/ψ in a Jet

J/ψ in a Jet

LHCb, Phys. Rev. Lett. 118 (2017)



J/ψ in a Jet

The Polarization Puzzle

]c) [GeV/ψJ/(T

p0 5 10

)]c [

nb/(

GeV

/T

p/dσd

10

210

310

410

<4.5y, 2.0< ψJ/LHCb prompt <4.5yNRQCD, 2.0<

JHEP 10 (2015)

) [GeV/c]ψ(J/Tp

0 5 10 15

polarization

-1-0.8-0.6-0.4-0.20

0.20.40.60.81

NLO NRQCD(1)NLO CS

2.5 < y < 4.0

= 7 TeVsLHCb

EPJC 73 (2013)

J/ψ (1)

gg

g

c

J/ψ (8)

gg

g

c

colour singletlow pT

longitudinal pol.

colour octethigh pT

transverse pol.




J/ψ in a Jet

A Tale of Two Pictures

1 NRQCD hard process, octet states showered with QCD splittings2 shower with NRQCD splittings, match with hard process

cc

hard production shower production

arXiv:1603.06981 • build J/ψ → µµ candidates• build jets with J/ψ input• select jets with J/ψs• z ≡ pT(J/ψ)/pT(jet)



J/ψ in a Jet

LHCb Trigger

• real-time calibration and full event reconstruction in Run 2• full detector readout in Run 3

detector

hardware

software disk

Run�2Run�3

5�TB/s O(10)�GB/s

0.7�GB/s

50�GB/s1�TB/s


J/ψ in a Jet

Dimuon Trigger

310 410 510

210

310

410

510

610

710

m(µµ) [ MeV ]

Can

did

ates

/σ[m

(µµ)]

/2

LHCb√s = 13 TeV

µ+µ−

µ±µ±

prompt-like sample

pT(µ) > 1 GeV, p(µ) > 20 GeV

• fully reconstructed event written to disk, including particle flow• jets can be fully reconstructed, sans hadronic calorimetry


J/ψ in a Jet

Prompt and Displaced

J/ψ (1)

cJ/ψ (1)c

b→ J/ψ (1)

• determine J/ψ signal yield with mass fits• separate prompt (direct) from displaced (b → J/ψ) yields with

pseudo-lifetime fits, τ̃ ≡ (xz − xz(PV))m/pz

) [GeV]−µ+µ(m3 3.05 3.1 3.15 3.2

Can

dida

tes

/ 5 M

eV

0

5000

10000

LHCb = 13 TeVs

(jet) < 30 GeVT

p20 <

) < 0.5ψ/J(z0.4 <

Data

Total Fit

ψ/J

Background

[ps]t~-10 -5 0 5 10

Can

dida

tes

/ 0.1

ps

1

10

210

310

410LHCb = 13 TeVs

(jet) < 30 GeVT

p20 <

) < 0.5ψ/J(z0.4 < Data

Total Fitψ/JPrompt

ψ/J→b

Background

Wrong PV


J/ψ in a Jet

Unfolding

• correct for z resolution and pT(j) resolution, ≈ 20− 25%• perform 2D unfolding in z and pT(j) (iterative D’Agostini)

0

1

1) [true]ψ/J(z

) [r

eco]

ψ/J(z(jet) [reco]

Tp

15-20 GeV

20-30 GeV

>30 G

eV

(jet) [true]T

p

15-20 GeV 20-30 GeV >30 GeV

LHCbsimulation


J/ψ in a Jet

Displaced Results

)ψ/J(z0 0.2 0.4 0.6 0.8 1

σ/σd

0

0.1

0.2

0.3

Data (syst)

Pythia 8

LHCb = 13 TeVs

ψ/J→b


J/ψ in a Jet

Displaced Results

)ψ/J(z0 0.2 0.4 0.6 0.8 1

σ/σd

0

0.1

0.2

0.3

0.4Pythia 8

-splitgno

no MPI

LHCb simulation = 13 TeVs

ψ J/→b


J/ψ in a Jet

Prompt Results

)ψ/J(z0 0.2 0.4 0.6 0.8 1

σ/σd

0

0.1

0.2

0.3

DPS

SPS

Data (syst)

LO NRQCD

LHCb = 13 TeVsPrompt

σeff = 31 mb (Pythia default)


J/ψ in a Jet

Prompt Results

)ψ/J(z0 0.2 0.4 0.6 0.8 1

σ/σd

0

0.2

0.4

0.6

0.8

1

+ MPI(1)1LO 3S

(1)1LO 3S

(1)1NLO* 3S

+ MPI(8)1LO 3S

(8)1LO 3S

(8)1NLO* 3S

LHCb simulation = 13 TeVsPrompt


Jets in a Jet?

Jets in a Jet?

Ilten, Rodd, Thaler and Williams,Phys. Rev. D 96 (2017)




Jets in a Jet?

SoftDrop and Jet Sub-structure

• what happens with boosted topology when Qhard � Qobs,e.g. W ,Z ,H → qq̄?

• anti-kt produces a single jet → need jet sub-structure• use jet sub-structure technique like SoftDrop

j0

j1→j0j2

j1 j2

final state particlesmin(pT1, pT2)

pT1 + pT2> zcut

(∆R12R

)β

1 create fat anti-kt jets2 build Cambridge/Aachen tree

for each fat jet3 split j0 into sub-jets j1 and j2

4 if j1 and j2 fulfil SoftDropcondition, terminate

5 otherwise, assign j0 to largerpT sub-jet and return to 3


Jets in a Jet?

Averaged Massless Splittings

+

+

=

1− zz + z

1− z + 12

zg ≡pT1

pT1 + pT2

0123456789

1σ

dσdz g

CMS 2010 Open DataTheory (MLL)Pythia 8.219Herwig 7.0.3Sherpa 2.2.1

p PFCT > 1.0 G eV

AK5; | η | < 2.4p jet

T > 150 G eV

mMDT / SDβ=0 : z cut = 0.1

0.0 0.1 0.2 0.3 0.4 0.5 0.6zg

0.5

1.0

1.5

2.0

Rat

ioto

Pyt

hia

arXiv:1704.05842

• SoftDrop provides direct access to the hardest 1→ 2 splitting



Jets in a Jet?

Jet Anatomy

1 find all tags in event and treat as ghosts2 build anti-kt jets with R = 1, including tags3 apply SoftDrop with zcut > 0.1 and β = 04 consider sub-jet tagged if ptag

T /(pT1 + pT2) > 0.05

Δy

Δφ

0 1-1

0

-1

1

Δy

Δφ

0 1-1

0

-1

1

Δy

Δφ

0 1-1

0

-1

1

(0, 0)Q (0, 1)Q (1, 1)Q


Jets in a Jet?

Some Numbers

σ(Pythia) [µb] σ(Herwig++) [µb](0, 0)c 9.96× 102 5.28× 102

(0, 1)c 7.56× 101 2.64× 101

(1, 1)c 6.87 × 100 2.87 × 100

(0, 2)c 1.00× 101 5.64× 100

otherc 8.86× 10−1 2.47 × 10−1

(0, 0)b 1.07 × 103 5.52× 102

(0, 1)b 1.34× 101 9.58× 100

(1, 1)b 8.40× 10−1 5.03× 10−1

(0, 2)b 9.50× 10−1 5.94× 10−1

otherb 1.13× 10−2 7.75× 10−3

• missed tags migrate category up → minimal contamination• efficiency of tagging well understood from data


Jets in a Jet?

Heavy Flavour Splittings

0.0 0.2 0.4 0.6 0.8 1.0z

0

1

2

3

4

5

(1/σ

)(d

σ/d

z g)

c-hadron tagPythiaR = 1.0

LHCb: pT > 20 GeV, z tag > 0.1, η ∈ [3, 4]

(0, 0)c

(0, 1)c

(1, 1)c

g

(0, 0)c

(0, 1)c

(1, 1)c


Jets in a Jet?

Quarkonia Splitting

0.0 0.2 0.4 0.6 0.8 1.0z

0

1

2

3

4

5

(1/σ

)(d

σ/d

z g)


LHCb: pT > 20 GeV, z tag > 0.1, η ∈ [3, 4]

(0, 1)c

(0, 2)c

(0, 1)J/ψ

g

(0, 1)c

(0, 2)c

c

(0, 1)J/ψ


Conclusions

Conclusions


Conclusions

Outlook

• LHCb has unique jet tagging capabilities• valence and strange quark PDFs• top asymmetry• intrinsic charm

• J/ψ production is not isolated!• reconsider theory with parton shower framework• implications for J/ψ production in medium• more measurements are needed

• SoftDrop accesses fundamental 1→ 2 QCD splittings• paired with jet tagging, probe heavy flavour splitting• new technique to understand J/ψ production

Thank you!Ilten Quarkonia in Jets 33 / 33

Appendix

Appendix


Appendix

Efficiencies

N (probe SV)N (total)

• N (probe SV) from BDT fit• N (total) from hardest χ2

IP fit

)IP2χmuon log(

-5 0 5 10 15

cand

idat

es

0

500

1000

1500

2000

LHCbdatabc

udsg

+jetD

light-parton mistag probability0.001 0.002 0.003 0.004 0.005

(b,c

)-je

t tag

eff

icie

ncy

0

0.2

0.4

0.6

0.8

1

-jetb-jetc

LHCb simulation

(jet) < 4.2η2.2 <

(jet) < 100 GeVT

20 < p

(jet) [GeV]T

p20 40 60 80 100

SV-t

ag e

ffic

ienc

y

0

0.2

0.4

0.6

0.8

1LHCb

-jetb

-jetc


Appendix

Probing the Proton

10−6 10−5 10−4 10−3 10−2 10−1 100

x

100

101

102

103

104

105

106

107

108

Q2

[GeV

2 ]

Tevatron

HERA

fixed target

GPD 13 TeV

LHCb 13 TeV

g

qi

qj

W

g

s

c

W

qj

qi

b

b

W


Appendix

W with a Jet

muon

tag

1 trigger on a high pT muon2 build jets in the event3 require jet containing muon jµ and tagged jet j4 bin data as a function of isolation, pT(µ)/pT(jµ)5 determine flavour in each isolation bin with BDT fit6 fit isolation to determine signal


Appendix

Flavour Determination


Appendix

Signal Determination

0.5 0.6 0.7 0.8 0.9 1

Can

dida

tes/

0.1

0

500

Data

W

Z

Jets

= 8 TeVs, +µLHCb-jetc+µ

) µj(T

p)/µ(T

p0.5 0.6 0.7 0.8 0.9 1

cand

idat

es

0

500

= 8 TeVs, −µ

0.5 0.6 0.7 0.8 0.9 1

Can

dida

tes/

0.1

0

500

1000 = 8 TeVs, +µLHCb

-jetb+µ

) µj(T

p)/µ(T

p0.5 0.6 0.7 0.8 0.9 1

cand

idat

es

0

500

1000 = 8 TeVs, −µ Data

W

Z

Jets


Appendix

Topping the W

Tevatron LHC

y=0 y=0

topanti-top

tg

g

b

W

t

tq

q

b

W

t

tqj

qi

b

W

b

gluon fusion quark fusion single top


Appendix

A Tricky Background

) [GeV]c+µ(T

p

)c+

W(N

0

50

100

150

200

Data

SM

LHCb

20 45 70 95 ∞

• constrain W b with W j• scale with W b/W j theory• validate with W c

) [GeV]b+µ(T

p

)W

+b

(N

0

100

200

Data+topWb

Wb

LHCb

20 45 70 95 ∞) [GeV]b+µ(

Tp

Cha

rge

Asy

mm

etry

-0.4

-0.2

0

0.2

0.4

Data

+topWb

Wb

LHCb

20 45 70 95 ∞


Appendix

Putting Everything Together

−0.09± 0.08± 0.04 A(Wc)0.51± 0.20± 0.09 A(Wb)5.80± 0.44± 0.75 σ(Wc)/σ(Wj)× 102

0.66± 0.13± 0.13 σ(Wb)/σ(Wj)× 102

6.61± 0.19± 0.33 σ(W−j)/σ(Zj)10.49± 0.28± 0.53 σ(W +j)/σ(Zj)

−0.01± 0.05± 0.04 A(Wc)0.27± 0.13± 0.09 A(Wb)5.62± 0.28± 0.73 σ(Wc)/σ(Wj)× 102

0.78± 0.08± 0.16 σ(Wb)/σ(Wj)× 102

6.02± 0.13± 0.30 σ(W−j)/σ(Zj)9.44± 0.19± 0.47 σ(W +j)/σ(Zj)

7 TeV

8 TeV

(exp - thr)/max(δthr)-5 0 5 10 239± 53± 41 [fb]7 TeV

289± 43± 49 [fb]8 TeV

(exp - thr)/max(δthr)-1 0 1 2


Appendix

Intrinsic Charm

µ

µ

cg

c

Z

p

extrinsic c

intrinsic c

simulation


Appendix

Bonus Intrinsic Charm

10−6 10−5 10−4 10−3 10−2 10−1 100

x

100

101

102

103

104

105

106

107

108

Q2

[GeV

2 ]

Tevatron

HERA

fixed target

GPD 13 TeV

LHCb 13 TeV

LHCb 110 GeV

type timepNe 30m

PbNe 30mpNe 12hpHe 7hpAr 20hpAr 11h

PbAr 100hpHe 20hpHe 87h


Appendix

Heavy Flavour Production

• understanding heavy flavour production critical for many signals• two approaches typically taken

1 hadron-level: good angular properties, poor energy proxy2 tagged jet-level: poor angular properties, good energy proxy

0 0.2 0.4 0.6 0.8 10

0.5

1

1.5

2

2.5

3

π σd

σd|Δ

φ∗|

|Δ φ∗| / π

PowhegPythiauncorrelatedLHCb

arXiv:1708.05994


http://arxiv.org/abs/1708.05994

Appendix

FlavorCone

• good angular properties, good energy proxy• collinear and infrared safe by jet-axis definition

y

φ

3 42-3

3

-2

-1

2

1

0

5y

φ

3 42-3

3

-2

-1

2

1

0

5

1 given n tags define njet-axes

2 particles outside of R withan jet-axis is not clustered

3 remaining particles areclustered with nearest axis

4 jet momenta is sum ofconstituents


Appendix

Comparison

0.0 0.5 1.0 1.5 2.0 2.5 3.0

Δφ

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4(1

/σ)(

dσ/

dΔ

φ)c-hadron tagPythiaR = 0.5

LHCb: pleadT > 20 GeV, zsub > 0.1, η ∈ [2.5, 4.5]

FlavorConeFlavorConeJetAnti-kt

Q-hadron× 0.9


Appendix

Variable Discrimination

0.0 0.5 1.0 1.5 2.0 2.5 3.0

Δφ

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

(1/σ

)(d

σ/dΔ

φ)



gluon splittingnon-splittingtotal

− 3 − 2 − 1 0 1 2 3

Δ y

0.0

0.2

0.4

0.6

0.8

1.0

(1/σ

)(d

σ/dΔ

y)




0 1 2 3 4 5

ΔR

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

(1/σ

)(d

σ/dΔ

R)




0 20 40 60 80 100

m j j [GeV]

0.00

0.01

0.02

0.03

0.04

0.05

(1/σ

)(d

σ/d

mjj

[GeV

])





Documents

Quarkonia in Jets - University of Birmingham · 2018. 5. 9. · Quarkonia in Jets Philip Ilten May 9, 2018 Birmingham Seminar Ilten Quarkonia in Jets 1 / 33