1. From yesterday Why jets in heavy ion collisions? Jet Tomography!
Jet I: Intro & Motivations
Why jets in heavy ion collisions? Jet Tomography!
Jet quenching observed at RHIC & LHC via single high-pT hadron and di-hadrons
Access kinematics of the binary hard-scattering
Characterize the parton energy loss in the hot QCD medium
Study medium response to parton energy loss
Jet II: Full Jet Reconstruction
Jet-finding connects Theory and Experiment
Many Jet Finders on the market
 need to be collinear/infrared safe
Choice of R matters
Elena Bruna (Yale&INFN Torino)

2. Today and tomorrow Jet-finding connects Theory and Experiment
Many Jet Finders on the market
 need to be collinear/infrared safe
Choice of R matters
Goal: set the Jet Energy Scale
Different systematics to take into account (tracking,…)
Background fluctuations: the challenge
Jet II: Full Jet Reconstruction
Jet III: Results
Jet IV:
The Present: from RHIC to LHC
Elena Bruna (Yale&INFN Torino)

3. Jets in Heavy-Ion Collisions at RHIC and LHC
Central Au+Au √sNN=200 GeV
ETjet ~ 21 GeV
STAR EMC + tracking data
STAR preliminary
Central Pb+Pb√sNN=2.76 TeV
ALICE tracking data
Why measure jets in heavy ion collisions? [inclusive, di-jets, jet-hadron, g-jet,..]
Access kinematics of the binary hard-scattering
Characterize the parton energy loss in the hot QCD medium
modified fragmentation, energy flow within jets, quark vs gluon jet difference
flavor and mass dependence
Study medium response to parton energy loss – establish properties of the medium
Elena Bruna (Yale&INFN Torino)

4. Jets in p+p ATLAS Nov/Dec 2009 √s=2.36 TeV
Tevatron: √s = 1.8 TeV
pt per grid cell [GeV]
~ 21 GeV η ϕ
p+p JP trigger
RHIC √s = 200 GeV
STAR Preliminary
p+p
Jet
p+p:
the reference measurement
 need well calibrated probes
Elena Bruna (Yale&INFN Torino)

5. Jet-finding and systematics..
Hadronic and electron double counting
Electrons and hadrons can deposit
energy in the EMC and leave tracks in
the TPC.
Be aware of double counting!
Avoid double counting (p+p and A+A):
remove EMC towers that match to an electron
remove fraction f of tower energy for tower that match to hadrons
choice of f to be determined by experiment. f=100%, MIP,…
Elena Bruna (Yale&INFN Torino)

6. Jet Energy resolution – the Jet Energy Scale
PYTHIA:
p+p √s=200 GeV
X axis=Pythia jet pT (Particle Level)
Y axis=Pythia thru STAR detector simulation
(Detector Level)
Reconstructed Jet pT on average smaller than the Input (PYTHIA) jet pT
The reconstructed jet pT is smeared
Need to know (1) and (2) to correct the measured jet pT back to the “true” jet pT
Can use PYTHIA to determine the jet energy resolution
Elena Bruna (Yale&INFN Torino)

7. Jet-finding and systematics..
Tracking efficiency
[Remark: PYTHIA is OK for p+p. Data-driven correction scheme preferred for A+A]
Charged jet component has to be corrected for the pT dependent tracking efficiency:
Simulation:
5.5 TeV in ALICE
ε(pT)= tracking efficiency
pT,hi= single track transverse momentum
N = # of jet constituents
Other systematics:
tracking performance at high-pT, high-luminosity track distortion, unobserved neutral energy, … [see backup slides]
Elena Bruna (Yale&INFN Torino)

8. Jets in A+A
Goal
Reconstruct the full jet kinematics of hard scattering in unbiased way, even in presence of (underlying) heavy-ion collision. ϕ η
pt per grid cell [GeV]
STAR preliminary
~ 21 GeV
di-jet event
Elena Bruna (Yale&INFN Torino)

9. Background in A+A pT (Jet Measured) ~ pT (Jet) + ρA ± F η ϕ
pt per grid cell [GeV]
STAR preliminary
~ 21 GeV
di-jet event
rA [Gev]
Reference multiplicity (~centrality)
Au+Au 0-20%
Rc=0.4, no pt cut, out-of-cone area
STAR Preliminary
STAR preliminary
Out-of-cone area
pT (Jet Measured) ~ pT (Jet) + ρA ± F
STAR Preliminary
Three main components:
Background energy in R=0.4 ~ 45 GeV at RHIC, ~90 GeV at LHC
Substantial region-to-region background fluctuations described by F
“Fake” jets: random association of uncorrelated soft particles (i.e. not due to hard scattering)
Elena Bruna (Yale&INFN Torino)

10. Background in A+A pT (Jet Measured) ~ pT (Jet) + ρA ± F
Event-wise background estimate:
reconstruct event with kT (jets+bkg)
all jets in acceptance {pT,i}
A = jet area in η-ϕ
ρA [Gev]
Reference multiplicity (~centrality)
Au+Au 0-20%
Rc=0.4
STAR preliminary
Why the median?
The background estimate has to be independent of the ‘true signal jets’.
The ‘true jets’ do not pull the median compared to the mean.
STAR Preliminary
Elena Bruna (Yale&INFN Torino)

11. Background in A+A pT (Jet Measured) ~ pT (Jet) + ρA ± Fϕ η
pt per grid cell [GeV]
STAR preliminary
~ 21 GeV
di-jet event
In a given Area, the background is subject to fluctuations around the median ρ
How to quantify fluctuations and fake jets?
STAR Preliminary
Elena Bruna (Yale&INFN Torino)

12. Background in A+A pT (Jet Measured) ~ pT (Jet) + ρA ± Fϕ η
pt per grid cell [GeV]
STAR preliminary
di-jet event
In a given Area, the background is subject to fluctuations around the median ρ
How to quantify fluctuations and fake jets?
STAR Preliminary
Embed a probe particle in the event: pTemb
Reconstruct hybrid event with anti-kT
Match reconstructed jet with embedded probe in (h,f):
pTcluster, Acluster
Quantify response via:
Elena Bruna (Yale&INFN Torino)

13. Assessing background fluctuations
f (pT,clus Meas – ρA – pTemb)
Example: pT,clus for only background clusters (no true jet)
pTemb=0
How to characterize the full shape of the bkg fluctuations?
pT,clus Meas – ρA
No fluctuations
Gaussian fluctuations
“Thermal” fluctuations
R=0.2
σ=3.5 GeV
Akt, R=0.2
Simulation
T=290 MeV
dN/dη=650
Elena Bruna (Yale&INFN Torino)

14. Background fluctuations: δpT
Single particle embedding
in real Au+Au
pT=30 GeV
h=-0.2
Gaussian fit to left-hand side (LHS):
LHS: good representation
RHS: non-Gaussian tail (real jets are there!)
centroid non-zero(~ ±1 GeV)
 contribution to jet energy scale uncertainty
Elena Bruna (Yale&INFN Torino)

15. Background fluctuations: δpT
Simple model: uncorrelated particle emission
M(A) = particle multiplicity in area A 
<pT> = mean pT in a given area A 
Poisson
Gamma
M. Tannenbaum
Phys. Lett. B498 (2001) 29
Background fluctuation distribution in a given area A in (η,ϕ):
No hard scattering
No correlations
Two parameters
Simple uncorrelated-emission model accounts for the bulk of background fluctuations (!)
Elena Bruna (Yale&INFN Torino)

16. Background fluctuations: δpT
Systematics
So far, embedded single particles
But jet ≠ single particles
investigate dependence on fragmentation patterns:
PYTHIA, QPYTHIA
δpT insensitive to different fragmentation
Crucial for quenched jets, whose fragmentation is unknown!
arXiv:
What do we do once we know the dpT shape?
Elena Bruna (Yale&INFN Torino)

17. Unfolding the underlying event
Jet Energy resolution distorts measured jet cross section
Background distorts measured jet cross section
Unfolding technique used to extract the ‘true’ jet spectrum  jet energy scale
unfolding
Pythia
unfolded
smeared
Elena Bruna (Yale&INFN Torino)

18. Unfolding/deconvoluting/unsmearing
Given true measure mj (i.e. true jet pT) and response function Rij (inefficiencies, irresolutions, …) the experiment will measure:
Ideally (i.e. with infinite statistics) we can determine mj from ni by inverting Rij
Don’t have infinite stats. so
need to solve for m iteratively.
Example of response matrix used for unfolding the underlying event.
[dpT=Gaussian, s=6.5 GeV]
(m)
(n)
- RooUnfold:
- 5 methods: D. D’Agostini, NIM.A362:487 (2005), …
Elena Bruna (Yale&INFN Torino)

19. Jet III: Results
Elena Bruna (Yale&INFN Torino)

20. Jets in p+p: calibrated probes?
Tevatron: √s = 1.8 TeV
√s=2.36 TeV
ATLAS Nov/Dec 2009
pt per grid cell [GeV]
~ 21 GeV η ϕ
p+p JP trigger
RHIC √s = 200 GeV
STAR Preliminary
p+p
Jet
p+p:
the reference measurement
 need well calibrated probes
Elena Bruna (Yale&INFN Torino)

21. Jets are calibrated probes
Tevatron
RHIC
Calibrated probes
Jet cross section in p+p (STAR), p+p, DIS, well described by pQCD
Jets in p+p are a good reference for A+A
Elena Bruna (Yale&INFN Torino)

22. Fragmentation Functions in p+p
Data not corrected to particle level.
“PYTHIA” = PYTHIA +GEANT
Preliminary
Preliminary
R=0.4
20 <Jet pTreco< 30 GeV/c
30 <Jet pTreco< 40 GeV/c
Preliminary
Preliminary
Reasonable agreement between data and PYTHIA
Jets in p+p are a good reference for A+A
Elena Bruna (Yale&INFN Torino)

23. The underlying event in p+p
|Δφ| − Angle relative to leading jet
“Toward” |Δφ| < 60o
“Away” |Δφ| > 120o
“Transverse” 60o < |Δφ| < 120o
TransMax - Trans. region with highest ΣpT or ΣNtrack
TransMin Trans. region with
least ΣpT or ΣNtrack
Underlying event = what is contained in the Transverse region, i.e. everything BUT the hard scattering
Elena Bruna (Yale&INFN Torino)

24. The underlying event in p+p
Agreement between PYTHIA and data
Underlying event is decoupled from the hard scattering
Elena Bruna (Yale&INFN Torino)

25. The underlying event in p+p
Agreement between PYTHIA and data
Underlying event is decoupled from the hard scattering
Elena Bruna (Yale&INFN Torino)

26. Jets in d+Au: why?
Control experiment: Measure possible initial state/Cold Nuclear Matter (CNM) effects
Probe the “cold medium” via d+Au collisions (compare to p+p)
d+Au
Jet
Elena Bruna (Yale&INFN Torino)

27. Jets in d+Au
σkT,raw (p+p) = 2.8 ± 0.1 GeV/c
σkT,raw (d+Au) = 3.0 ± 0.1 GeV/c
-kt broadening in pp
No add broadening in dAu
3Gev=pt sinDeltaPhi (sqrt(20^2 + 3^2)=20.2  1% effect)
No strong Cold Nuclear Matter effect on jet kT broadening seen
Systematics under investigation
Elena Bruna (Yale&INFN Torino)

28. Jets in d+Au No significant deviation from Nbin scaling in d+Au
-kt broadening in pp
No add broadening in dAu
3Gev=pt sinDeltaPhi (sqrt(20^2 + 3^2)=20.2  1% effect)
No significant deviation from Nbin scaling in d+Au
Initial state/Cold nuclear matter effects in the kinematic range as measured in d+Au seem to be small
Systematics under investigation
Elena Bruna (Yale&INFN Torino)

29. ✓ ✓ So far, so good p+p: the reference measurement
Jet
p+p:
the reference measurement
 calibrated probes ! ✓
d+Au
Jet
d+Au:
the control measurement
 No strong Cold Nuclear Matter effect ✓
Elena Bruna (Yale&INFN Torino)

30. Jets in Au+Au: what to expect?
- for unbiased jet reconstruction -
Jet energy fully recovered even in case of quenching
Jet is a hard process, scales as Nbin
Inclusive spectra:
Di-jet analyses:
Ratio of recoil spectra Au+Au/p+p = 1
Modified fragmentation in case of dense medium
Elena Bruna (Yale&INFN Torino)

31. Inclusive measurements
Inclusive Jet spectrum measured in central Au+Au collisions at RHIC
Extended the kinematical reach to study jet quenching phenomena to jet energies > 40 GeV
Elena Bruna (Yale&INFN Torino)

32. Inclusive measurements
Inclusive Jet spectrum measured in central Cu+Cu collisions at RHIC
Extended the kinematical reach to study jet quenching phenomena to jet energies > 40 GeV
Elena Bruna (Yale&INFN Torino)

33. Jet RAA
Inclusive RAA
RAAjet<1 We see a substantial fraction of jets - in contrast to x5 suppression for light hadron RAA (RAAjet > RAA )
kT and Anti-kT known to have different sensitivities to background
Elena Bruna (Yale&INFN Torino)

34. Jet energy profile: first look
Jet inclusive measurements: 0.2 vs 0.4
R=0.2
R=0.4
Solid lines:
Pythia – particle level
p+p:
jets more collimated with increasing pT
PYTHIA (fragmentation + hadronization) describes the data
From 0.2 to 0.4
Elena Bruna (Yale&INFN Torino)

35. Jet energy profile: first look
Jet inclusive measurements: 0.2 vs 0.4
G. Soyez – priv. comm. 2010
Solid lines:
Pythia – particle level
From 0.2 to 0.4
NLO ≈ PYTHIA parton level
PYTHIA hadron level ≈
HERWIG hadron level
Be careful when comparing to theory: Hadronization broadens the jet
Elena Bruna (Yale&INFN Torino)

36. Jet energy profile: first look
Jet inclusive measurements: 0.2 vs 0.4
R=0.2
R=0.4
p+p:
jets more collimated with increasing pT
PYTHIA (fragmentation + hadronization) describes the data
Au+Au:
ratio lower than p+p
“Deficit” of jet energy for jets reconstructed with R=0.2
From 0.2 to 0.4
Elena Bruna (Yale&INFN Torino)

37. Jet energy profile: first look
Jet inclusive measurements: 0.2 vs 0.4
R=0.2
R=0.4
Red: p+p
Blue: Au+Au
From 0.2 to 0.4
Ratio of energy within 0.2 relative to energy in 0.4 smaller in AuAu due to broadening
Suggests strong broadening of the energy profile
Elena Bruna (Yale&INFN Torino)

38. Jets in p+p and Cu+Cu in PHENIX
PHENIX uses a Gaussian filter approach
Cone-like, but no fixed angular cut-offs
Implements fake jet rejection
Elena Bruna (Yale&INFN Torino)

39. Jets in A+A: possible biases
CAVEAT:
jet-finder based on unmodified jet-shapes
⇒ veto against modified/quenched jets
“Anti-quenching” biases!
pT cut to minimize background
⇒ bias towards less-interacting jets
Can we exploit the biases?
Elena Bruna (Yale&INFN Torino)

40. Di-jet measurements Trigger jets are biased towards the surface.
EMC trigger
Trigger jet
Recoil jet
Trigger jets are biased towards the surface.
Recoil jets are exposed to a maximum path-length in the medium.
Large energy loss expected.
σ=6.5 GeV/c
Anti-kT, R=0.4
Trigger Jet: pT,cut=2 GeV/c, pT(trig)>20 GeV/c
Coincidence rate:
how often I measure a recoil jet once the trigger jet is found
Elena Bruna (Yale&INFN Torino)

41. BACKUP
Elena Bruna (Yale&INFN Torino)

42. Jet Energy resolution with di-jets
Particle-Detector jet Res:
pTJet(Part.Lev) – pTJet(Det.Lev)
~10-25 %
di-jet Res:
pTJet 1– pTJet 2
(PY Det. Lev.) ~ (dijet data) : good!
But:
(dijet PY Det. Lev.) > (Part-Det)
di-jet imbalance includes both energy resolution and kT (initial state) effect!
[kT=pTjet sinDfdijet]
kT: good agreement between data and simulation
Use PYTHIA to determine the jet energy resolution
Elena Bruna (Yale&INFN Torino)

43. Jet-finding and systematics..
Tracking performance
Tracking is limited by misalignment, luminosity, resolution…
Rare processes as high-pT jets are likely to come from high luminosity runs
Example of high-luminosity distortion? Space-charge effect  accumulation of space charge in the TPC that causes an anomalous transport of drifting electrons in the TPC, affecting the tracking performance by shifting the momentum up or down (depending on the charge)
Tracking resolution at high-pT is
expected to deteriorate  need to apply
an upper pT cut on tracks
PYTHIA simulation: p+p 200 GeV
effect of upper pT cut on jet energy scale
Elena Bruna (Yale&INFN Torino)

44. Jet-finding and systematics..
Unobserved neutral energy
Experiments like STAR and ALICE do not detect neutral, long-lived particles (neutrons, K0L)
PYTHIA simulation:
p+p at 200 GeV
mean missed E ~ 9%
median missed E <0.3 %
50% of jets loose no energy
model dependent
Elena Bruna (Yale&INFN Torino)

45. Fragmentation Functions
large uncertainties due to background (further systematic evaluation needed)
STAR preliminary
pT Jet(trig)>20 GeV
pTcut=2 GeV
AuAu (Jet+Bkg)
Charged particle FF: R(FF)=0.7
Jet energy determined in R=0.4
AuAu (Bkg)
high z
low z
xrec=ln( pT,Jet rec / pT,hadr)
AuAu: FF(Jet)=FF(Jet+Bkg)-FF(bkg) Bkg estimated from charged particle spectra out of jet cones Bkg dominates at low pT
Elena Bruna (Yale&INFN Torino)

46. Fragmentation Functions
“recoil” jet
“trigger” jet
EMC trigger
Smaller energy in the jet cone. We need to better determine the jet energy
If ptjet(AuAu)>ptjet(pp) due to bkg fluctu, FF(AuAu) artificially softer before unfolding.
After unfolding  FF(AuAu) is harder.
If the energy is not recovered in R=0.4  E(AuAu) smaller than E(pp). A 20 GeV pp jet is actually a higher energy AuAu jet. The FF is ~ indep of jet energy. Using the “small” energy, FF(Au) should be by construction harder than FF(pp). The fact that it is “unmodified” could mean that actually FF(Au) is modified but due to the choice of the “small” E, the softening does not show up in the ratio.
No apparent modification of FF of recoil jets with pTrec>25 GeV would imply non-interacting jets, but: Jet broadeningEnergy shift harder FF Need to better determine the jet energy
Elena Bruna (Yale&INFN Torino)

47. Jet Yields in ALICE
Elena Bruna (Yale&INFN Torino)

48. DCal for Di-Jet analysis @ ALICE
Elena Bruna (Yale&INFN Torino)