Lecture 3: jets in heavy ion collisions

JET School, UC Davis, 19-21 June 2014

Marco van Leeuwen, Nikhef and Utrecht University

2

Jets and parton energy loss

Motivation: understand parton energy loss by tracking the lost energy

Qualitatively two scenarios: 1) In-cone radiation: RAA = 1, change of fragmentation 2) Out-of-cone radiation: RAA < 1

3

Jets at LHCALICE

η

ϕ

Transverse energy map of 1 event

Clear peaks: jets of fragments from high-energy quarks and gluonsAnd a lot of uncorrelated ‘soft’ background

4

Jet reconstruction algorithms

Two categories of jet algorithms: • Sequential recombination kT, anti-kT, Durham

– Define distance measure, e.g. dij = min(pTi,pTj)*Rij

– Cluster closest • Cone

– Draw Cone radius R around starting point – Iterate until stable η,ϕjet = <η,ϕ>particles

For a complete discussion, see: http://www.lpthe.jussieu.fr/~salam/teaching/PhD-courses.html

Sum particles inside jet Different prescriptions exist, most natural: E-scheme, sum 4-vectors

Jet is an object defined by jet algorithm If parameters are right, may approximate parton

5

Collinear and infrared safetyIllustration by G

. Salam

Jets should not be sensitive to soft effects (hadronisation and E-loss)

- Collinear safe - Infrared safe

6

Collinear safety

Note also: detector effects, such as splitting clusters in calorimeter (π0 decay)

Illustration by G. S

alam

7

Infrared safety

Infrared safety also implies robustness against soft background in heavy ion collisions

Illustration by G. S

alam

8

Jet algorithm examplesC

acciari, Salam

, Soyez, arX

iv:0802.1189simulated p+p event

Di-jet imbalance measurements

Di-jet asymmetry

10

Observation: some events have two jets with different energy!(However: one swallow does not make spring)

11

Jet energy asymmetryCentrality

12

12

EEEEAJ +

−=

ATLAS

, arXiv:1011.6182 (P

RL)

Jet-energy asymmetry Large asymmetry seen for central events

However: • Only measures reconstructed di-jets (don’t see lost jets) • Not corrected for fluctuations from detector+background • Both jets are interacting – Not a simple observable

Suggests large energy loss: many GeV ~ compatible with expectations from RHIC+theory

12

Energy dependence of asymmetryC

MS

, arXiv:1202.5022

(Relative) asymmetry decreases with energyHowever: difference pp vs PbPb remains – energy loss finite at large E

Energy dependence of AJ

13

150 200 250 300 350

⟩T,

1/p

T,2

p⟨

0.5

0.6

0.7

0.8

-1bµLdt = 150 ∫ = 2.76 TeV, NNsPbPb -1Ldt = 231 nb∫ = 2.76 TeV, spp

PYTHIA+HYDJET 50-100%

CMS

150 200 250 300 350

PbPb

- M

C

-0.2

-0.1

0

(GeV/c)T,1

p

150 200 250 300 350

⟩T,

1/p

T,2

p⟨0.5

0.6

0.7

0.8

20-50%

> 30 GeV/cT,2

p

π32 >

12φ∆

(GeV/c)T,1

p150 200 250 300 350

PbPb

- M

C

-0.2-0.15

-0.1-0.05

00.05 150 200 250 300 350

⟩T,

1/p

T,2

p⟨

0.5

0.6

0.7

0.8

0-20%

150 200 250 300 350

PbPb

- M

C

-0.2-0.15

-0.1-0.05

00.05

Peripheral Central

Asymmetry decreases for larger jet energy!Similar effect in pp (Pythia): difference stays ~constant

14

γ-jet imbalance

CMS, arXiv:1502.0206Centrality

γγT

jetT

J ppx =γ-jet asymmetry

Advantage: γ is a parton: know parton kinematics Disadvantage: low rate (+background π0→ γγ)

γ

Dominant contibution: qg → qγTranslates into: low pT,𝛾 cut > 60 GeV

Potentially important observable — more stats in next run(s)

Corrections

• AJ is basically uncorrected!• Background subtracted, detector effects corrected on average!• Hard to correct for detector effects+fluctuations!

• Spectra and recoil measurements are corrected for!• Detector effects!• Background: !

• Overall background, measured outside jet cone, details differ between experiments!

• Background fluctuations!• Motivation: compare to theory without detector effects and without

background (may be ill-defined)

15

Always ask: for which effects is the measurement corrected?!Important for any measurement, but in particular for jets!

16

Measuring the jet spectrum

Charged and full jets

• Full jets: charged + neutral particles (except neutrinos)!• Hadronic + Electromagnetic Calorimetry

(ATLAS)!• + tracking (particle flow; CMS)!

• Tracking + EMCal (ALICE)!• Charged jets: only charged particles!

• Used by ALICE because of limited acceptance of EMCal

17

[GeV]coneTE

0 10 20 30 40 50 60 70 80 90 100

Cou

nts

/ 2 G

eV

0

200

400

600

800

1000

1200

aCharged + ChargedLeading charged particle

Reconstructed energy100 GeV jets (particle level)

Charge to neutral fluctuations!

Full jets strongly preferred for original goal: recover jet energy

Detector corrections

Standard practice:!• Charged jets are corrected to charged jets at

the particle level!• main effect: tracking efficiency!

• Full jets are corrected to full jets at the particle level!• Calorimetric jets: HCal response!• Tracking+EMCal: Unmeasured hadrons (neutrons,

K0L, tracking efficiency)

18

Definitions:!Particle level: as generated by event generator, e.g. Pythia!Detector level: as reconstructed (Pythia+detector simulation)!(Parton level: parton energy; ill-defined?)

part

T,ch jetp)/part

T,ch jetp- det

T,ch jetp(

-1 -0.8 -0.6 -0.4 -0.2 0 0.2

Pro

ba

bili

ty/B

in(0

.04

)

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

ALICE simulation

)c (GeV/part

T,ch jetp

30 - 40

50 - 60

70 - 80

= 0.3R TkAnti-

c > 0.15 GeV/track

Tp

ALI−PUB−64226

19

PbPb jet background

Cacciari et al

Background density vs multiplicity

η-ϕ space filled with jets Many ‘background jets’

Background contributes up to ~180 GeV per unit area

Statistical fluctuations remain after subtraction

Subtract background: App rawjetT

subjetT ρ−= ,,

Jet finding illustration

20

PbPb jet background

Toy Model

Main challenge: large fluctuations of uncorrelated background energy

Size of fluctuations depends on pT cut, cone radius

21

Background jetsRaw jet spectrum

Event-by-event background subtracted

Low pT: ‘combinatorial jets’ - Can be suppressed by requiring

leading track - However: no strict distinction at

low pT possible

Next step: Correct for background fluctuations and detector effects by unfolding/deconvolution

22

Removing the combinatorial jets

Correct spectrum and remove combinatorial jets by unfolding

Results agree with biased jets: reliably recovers all jets and removed bkg

Raw jet spectrum Fully corrected jet spectrum

ALICE arXiv:1311.0633

23

PbPb jet spectraCharged jets, R=0.3

Jet spectrum in Pb+Pb: charged particle jets Two cone radii, 4 centralities

RCP, charged jets, R=0.3

Jet reconstruction does not‘recover’ much of the radiated energy

ALICE arXiv:1311.0633

24

Pb+Pb jet RAA

Jet RAA measured byATLAS, ALICE, CMS

RAA < 1: not all produced jets are seen; out-of-cone radiation and/or ‘absorption’

For jet energies up to ~250 GeV; energy loss is a very large effect

ATLAS+CMS: hadron+EM jets

ALICE: charged track jets

Good agreementbetween experiments

Despite different methods:

Comparing hadrons and jets

25

)c (GeV/jetTp, track

Tp

10 210

CP

R

00.20.40.60.8

11.21.41.61.8

2 Ch. particles

(0-10%)/(50-80%)ALICE

(0-5%)/(50-90%)CMS

Jets

(0-10%)/(50-80%)=0.3RALICE Ch. Jets

(0-10%)/(60-80%)=0.3RATLAS Calo Jets

=2.76 TeVNNsPb-Pb

Suppression of hadron (leading fragment) and jet yield similar Is this ‘natural’? No (visible) effect of in-cone radiation?

Comparison to JEWEL energy loss MC

26

JEWEL+PYTHIA

ALICE data

0

0.1

0.2

0.3

0.4

0.5

0.6R=0.2

RAA

30 40 50 60 70 80 90 100 110

0.6

0.8

1

1.2

1.4

p⊥ [GeV]

Ratio

ALICE data

JEWEL+PYTHIA

0

0.1

0.2

0.3

0.4

0.5

0.6R=0.3

RAA

40 50 60 70 80 90 100 110

0.6

0.8

1

1.2

1.4

p⊥ [GeV]

Ratio

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

10 100 1000

RA

A

pt [GeV]

charged hadrons

CMS data, 0-5% centrality

ALICE data, 0-5% centrality

JEWEL+PYTHIA

JEWEL shows the same feature: jet RAA ~ hadron RAA

Path length dependence: v2 of Jets

27

0.028

0.03

0.032

0.034 5 - 10 % 0.002± = 0.016 meas| jet

2v

-1 = 0.14 nbt dL∫ = 2.76 TeVNNsPb+Pb

10 - 20 % 0.002± = 0.032 meas| jet

2v

ATLAS

= 0.2R tkanti- < 80 GeV

Tp60 <

]-1

[GeV

mea

sφ

∆d Tpdje

tN2 d

jet

N1

0.028

0.03

0.032

0.034 20 - 30 % 0.002± = 0.042 meas| jet

2v30 - 40 %

0.002± = 0.041 meas| jet2v

φ∆0 0.5 1 1.50.028

0.03

0.032

0.034 40 - 50 % 0.003± = 0.034 meas| jet

2v

φ∆0 0.5 1 1.5

50 - 60 % 0.004± = 0.027 meas| jet

2v

ATLAS, arXiv:1306.6469

jet

2v

0

0.02

0.04

0.065 - 10 % = 0.2R tkanti- 10 - 20 % ATLAS

jet

2v0

0.02

0.04

0.0620 - 30 %

= 2.76 TeVNNsPb+Pb

-1 = 0.14 nbt dL∫ 30 - 40 %

[GeV] Tp

50 100 150 200

jet

2v

0

0.02

0.04

0.0640 - 50 %

[GeV] Tp

50 100 150 200

50 - 60 %

Significant azimuthal modulation of jet yield!jet v2 ~ 0.03 at high pT

Comparing to JEWEL energy loss MC

28

ATLAS dataJEWEL+PYTHIA

0.5

0.55

0.6

0.65

0.7

0.75

(5-10)%, 45GeV < p⊥ < 60GeV

1/Njetd2Njet/dp⊥d

∆φ∣ ∣ ∣

corr

[GeV

−1]

0 0.2 0.4 0.6 0.8 1 1.2 1.4

0.6

0.8

1

1.2

1.4

∆φ

MC/data

ATLAS dataJEWEL+PYTHIA

0.5

0.55

0.6

0.65

0.7

0.75

(40-50)%, 45GeV < p⊥ < 60GeV

1/Njetd2Njet/dp⊥d

∆φ∣ ∣ ∣

corr

[GeV

−1]

0 0.2 0.4 0.6 0.8 1 1.2 1.4

0.6

0.8

1

1.2

1.4

∆φ

MC/data

5-10% centrality 40-50% centrality

Good agreement between JEWEL and jet v2 results

K. Zapp, arXiv:1312.5563

Geometry: Glauber overlap with Bjorken expansion

29

Generic expectations from energy loss

• Longitudinal modification: – out-of-cone ⇒ energy lost, suppression of yield, di-jet energy

imbalance – in-cone ⇒ softening of fragmentation

• Transverse modification – out-of-cone ⇒ increase acoplanarity kT

– in-cone ⇒ broadening of jet-profile

λ

kT~µEjet

fragmentation after energy loss?

Out-of-cone effects are large, so expect combination of all of the above

30

Looking outside the jet coneC

MS

, arXiv:1102.1957

In Cone R<0.8 Out of Cone R>0.8

PYTHIA+HYDJET

CMS measured

∑ −=tracks

jetTppmissT

)cos(//,

ϕϕ

Momentum imbalance restored by hadrons at large angle R>0.8 and

small pT < 2 GeV/c

Jet broadening: R dependence of RAA

31

= 0.2R = 0.3R = 0.4R = 0.5R0

0.5

1 < 44 GeV

Tp38 <

0

0.5

1 < 58 GeV

Tp50 <

0

0.5

1 < 103 GeV

Tp89 <

CP

R

0

0.5

1

1.5

2

< 182 GeVT

p158 <

= 2.76 TeVNNsPb+Pb -1bµ = 7 L dt ∫

ATLAS0-10 % Centrality

ATLAS, arXiv:1208.1867

However, R = 0.5 still has RAA < 1 – Hard to see/measure the radiated energy

[GeV] T

p

0.2

CP

R/R CP

R

0.8

1

1.2

1.4

1.6

1.8

2

40 50 60 70 100 200

= 0.3R = 0.4R = 0.5R

0 - 10 %ATLAS

= 2.76 TeVNNsPb+Pb -1bµ = 7 L dt∫

Larger jet cone: ‘catch’ more radiation ! Jet broadening

Ratio of spectra with different R

Changes in fragmentation

32

!PAS C

MS-H

IN-12-013

0-10%CMS preliminary

Longitudinal fragment distributions

CM

S, arX

iv:1310.0878

Transversefragment distributions

Enhancement at large R, low pT

No modification at small R, large pT: physics or auto-correlation?

Again: background fluctuations

33

Toy model spectrum

At fixed pT: pick up above-average background contributions!

10-2

10-1

1

10

102

103

0 0.2 0.4 0.6 0.8 1

1/N

jet d

Nh/

dz

z

PythiaPythia+Hydjet

z-subtracted

10-2

10-1

1

10

102

0 1 2 3 4 5 6 7 8

1/N

jet d

Nh/

dj

j

PbPb 3s=2.76 TeV0-10% (Hydjet)anti-kt, R=0.4, |y|<3

0

0.5

1

1.5

2

2.5

0 1 2 3 4 5 6

[(1+N

)/2]4 M

N

N

psubt,full>100 GeV

N-subtracted

10-2

10-1

1

10

102

103

0 0.2 0.4 0.6 0.8 1

1/N

jet d

Nh/

dz

z

10-2

10-1

1

10

102

0 1 2 3 4 5 6 7 8

1/N

jet d

Nh/

dj

j

0

0.5

1

1.5

2

2.5

0 1 2 3 4 5 6

[(1+N

)/2]4 M

N

N

psubt,full>200 GeV

Cacciari et al, arXiv:1209.6086

Background fluctuations migrate yield to higher pT!

Fragment distributions (simulation)

ξ ≳ 4 ⇔ pT ≲ 2 GeV

Current measurements mostly pT > 2 GeV

Jet Quenching Summary I• So, jet RAA is not close to 1!• Large out-of-cone radiation, low pT, large angles!• NB: even the fragmentation measurements do not capture the

‘initial energy’

34

What is the (dominant) mechanism?!Several lines of investigation!

!- No angular ordering the the medium; large angle radiation allowed (Mehtar-

Tani, Salgado, Tywoniuk)!- Interplay of scales: medium density/mean free path vs opening angle of radiation!

- Multiple interactions ‘thermalise’ the radiation (Renk, Wiedemann, Caselderrey-Solana)!

- Large angle ‘democratic’ gluon splitting allowed in the medium (Blaizot, Iancu, Mehtar-Tani)!

- Kinematics, (trigger-)biases also play a role!- Thorsten Renk: effect of Angular Ordering is small in Pythia

Comparing to energy loss models

35

R=0.2

R=0.3 (×10−1)

R=0.4 (×10−2)

R=0.5 (×10−3)

ATLAS data

JEWEL+PYTHIA

40 60 80 100 120 140 160 180 200

10−3

10−2

10−1

1

p⊥ [GeV]

RCP K. Zapp et al, arXiv:1212.1599

JEWEL gets the right suppression for R=0.2,!but not the increase with R!

(Treatment of recoil partons?)

JEWEL: RCP vs RJet observables: need explicit modelling of multi-particle final states

Mehtar-Tani, Tywoniuk, arXiv:1401.8293

Fragment distributions sensitive to coherence effects

(NB: no geometry model yet)

36

Jet trigger

Hadron trigger

Hadron trigger vs jet trigger

Hadron trigger: strong “surface bias” maximizes recoil path length

Full jet trigger: no geom. bias partially cancelled by bkg fluctuations

T.Renk, P

RC

85 064908

If hadron and jet RAA are similar, why not use hadron observables?Are jets an unnecessary complication?

Biases are different! Can be exploited to constrain models

Hadron-recoil jet measurements

37

38

G. de Barros et al., arXiv:1208.1518

pT,jet< 20 GeV/c: No change with trigger pT

Combinatorial background

Hadron-triggered recoil jet distributions

pT,jet> 20 GeV/c: Evolves with trigger pT

Recoil jet spectrum

39

Remove background by subtracting spectrum with lower pT

trig:

Δrecoil =[(20-50)-(15-20)]

Reference spectrum (15-20) scaled by ~0.96 to account for

conservation of jet density

Background subtraction: Δrecoil

Unfolding correction for background fluctuations and detector response

Δrecoil measures the change of the recoil spectrum with pTtrig

40

Recoil jet yield ΔIAAPYTHIA ≈0.75, approx. constant with jet pT

R=0.4 !Constituents: pT

const > 0.15 GeV/c !no additional cuts (fragmentation bias) on recoil jets

pp reference: PYTHIA (Perugia 2010)

Ratio of Recoil Jet Yield ΔIAAPYTHIA

41

R=0.4

Recoil Jet ΔIAAPYTHIA: R dependence

Similar ΔIAAPYTHIA for R=0.2 and R=0.4

R=0.2

No visible broadening within R=0.4(within exp uncertainties)

42

Hadrons vs jets II: recoil

PR

L108 092301

Hadrons Jets

Hadron IAA = 0.5-0.6In approx. agreement with models; elastic E-loss would give larger IAA

Jet IAA = 0.7-0.8Jet IAA > hadron IAA Not unreasonable

NB/caveat: very different momentum scales !

(GeV/c)ch

T,jetp

0 10 20 30 40 50 60 70

Pyt

hia

reco

il∆/

Pb-P

bre

coil

∆ =

P

ythia

AA

I∆

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6 > 0.15 GeV/cconst

T, R=0.4, pTanti-k

(20-50)-(15-20)PYTHIAAA I∆ALICE

(20-50) 0-10%PYTHIAAAYAJEM I

(20-50)-(15-20)PYTHIAAA I∆JEWEL

= 2.76 TeVNN

sPb-Pb 0-20%

ALI−DER−44225

43

Model comparison IAA

JEWEL: Zapp et al., EPJ C69, 617

JEWEL correctly describes inclusive jet RAA

JEWEL ΔIAA~0.4, below measuredYAJEM agrees with measurement

Difference in energy loss or geometry?

44

Summary• Jets: a ‘new’ tool for parton energy loss measurements

– Large out-of-cone radiation (R = 0.2-0.4) • Energy asymmetry • RAA < 1, similar to hadrons • IAA < 1 • Radial shapes

– Remaining jet has small modifications: • Longitudinal and transverse structure similar at small r, large z • Deviations at large r, low z

– Most of the radiation is at low pT • Scale set by medium temperature? • Democratic branchings?

Does this constrain the energy loss mechanism(s)?!Ongoing work…

Interplays of many effects: impossible to read simple conclusions off the plots!Need (detailed) calculations to draw conclusionse.g. JEWEL and YaJEM energy loss MCs agree

with many of the observed effects!

Extra slides

A consistent view of jet quenching

46

Consistent with 2010 result !Recall (2010 vs 2011): •Track pT > 4 GeV vs pT > 1 GeV •Leading vs inclusive jet •0-30% vs 0-10% and 10-30%

2010 data: arXiv:1205.5872

arXiv:1205.5872

Change from

“ξ” to “pT”

PbP

b –

pp (1

/GeV

)

Broadening/excess at large r, low pT

!(~2% of jet energy)

Narrowing/depletion at intermediate r, pT

No change at small r, high pT

Radius r

G. Roland@QM2012

A consistent view of jet quenching

47

Charged particles from pT =50-100 GeV:

z = pT(track)/pT(jet) = 0.4-0.6 ξ < 1

Looking at the same parton pT range

Consistent message from charged hadron RAA, inclusive jet RAA and fragmentation functions!

PbPb fragmentation function = pp for ξ <1

G. Roland@QM2012

48

Jet fragment distributionsPbPb measurement Ratio to pp

Low pT enhancement: soft radiation

Intermediate z: depletion: E-lossNB: z is wrt observed Ejet ≠ initial Eparton

ATLAS

M.R

ybar@Q

M12

M. Rybar@QM2012

49

Jet fragment distributions

Low pT enhancement: soft radiation

Intermediate z, pT: depletion: E-loss

CMS, Frank Ma@QM12

r 0 0.1 0.2 0.3

(r)

ρ

-110

1

10

70-100%

= 2.76 TeV NNsCMS, jets: R = 0.3Tanti-k

> 100 GeV/cjetT

p| < 2jetη0.3 < |

>1 GeV/ctrackT

p

r 0 0.1 0.2 0.3

pp(r)ρ/

PbPb

(r)ρ

0.5

1

1.5 r 0 0.1 0.2 0.3

(r)

ρ

-110

1

10

50-70%

PbPbpp reference

-1L dt = 5.3 pb∫pp,

r 0 0.1 0.2 0.3

pp(r)ρ/

PbPb

(r)ρ

0.5

1

1.5 r 0 0.1 0.2 0.3

(r)

ρ

-110

1

10

30-50%

-1bµL dt = 150 ∫PbPb,

r 0 0.1 0.2 0.3

pp(r)ρ/

PbPb

(r)ρ

0.5

1

1.5 r 0 0.1 0.2 0.3

(r)

ρ

-110

1

10

10-30%

r 0 0.1 0.2 0.3

pp(r)ρ/

PbPb

(r)ρ

0.5

1

1.5 r 0 0.1 0.2 0.3

(r)

ρ

-110

1

10

0-10%

r 0 0.1 0.2 0.3

pp(r)ρ/

PbPb

(r)ρ

0.5

1

1.5

50

Jet broadening: transverse fragment distributions

PbPb PbPb

CM

S PA

S H

IN-12-013

Jet broadening: radiation at large angles

CM

S, arX

iv:1310.0878

Comparing to JEWEL energy loss MC

51

R=0.2

R=0.3 (×10−1)

R=0.4 (×10−2)

R=0.5 (×10−3)

ATLAS data

JEWEL+PYTHIA

40 60 80 100 120 140 160 180 200

10−3

10−2

10−1

1

p⊥ [GeV]

RCP

0.6

0.8

1

1.2

1.4

MC/data

R=0.2

0.6

0.8

1

1.2

1.4

MC/data

R=0.3

0.6

0.8

1

1.2

1.4

MC/data

R=0.4

40 60 80 100 120 140 160 180 200

0.6

0.8

1

1.2

1.4

p⊥ [GeV]

MC/data

R=0.5

K. Zapp et al, arXiv:1212.1599

JEWEL gets the right suppression for R=0.2,!but not the increase with R!

May be treatment of recoil patrons

Full jet comparison

52

)c (GeV/T,jet

p50 100 150 200 250 300

AA

R

0

0.5

1

= 2.76 TeVNNsALICE Preliminary Pb-Pb

CMS: Read from HIN-12-004-PAS

= 0.3RCMS: Syst. Unc.

= 0.2R Tkanti-

ALICE 0-10%

ALICE 10-30%

CMS 0-5%

CMS 10-30%

ALI-PREL-84783

Good agreement between experiments; hint of pT dependence