1

A first study of the A first study of the feasibility of HBT feasibility of HBT

analysis analysis in the ALICE in the ALICE

Inner Tracking Inner Tracking System stand-aloneSystem stand-alone

Alberto Pulvirenti (INFN and University of Catania)

21st Winter Workshop on Nuclear DynamicsBreckenridge, 8 February 2005

22

OutlineOutline

HBT utility and forecasts for LHCLHC, ALICE and the Inner Tracking SystemTracking in ALICEHBT simulation in the off-line ALICE frameworkTracking in the ITS stand-aloneStudy of the detectability of HBT signals in the ITS stand-aloneConclusions and outlook

33

Intensity interferometry and QGPIntensity interferometry and QGP

HBT source characterization in heavy ion collisions> Information about phase-space density S(x,k)> Sketch of the source state at freeze-out > Study of source final state for comparison with

theoretical models> Evidence of collective effects which can be

explained with QGP formation and NOT without it◊ R “out” / R “side” ratio greater than 1

– Long the emission duration at freeze-out due to a permanence of the fireball in the mixed phase while QGP hadronizes

» Rischke-Gyulassy Nucl.Phys. A608 (1996), 479

44From recent experiments (SPS – RHIC):From recent experiments (SPS – RHIC):KKTT dependence and collective dynamics of dependence and collective dynamics of fireballfireball

t

fflong m

TtR

f

ft

geomside

T

m

RR

2

1

Using the value for the freeze-out temperature = 120 MeV:> Average expansion duration (tf) = 6-8 fm / c> Transverse expansion velocity ~ 0.5c> Transverse source radius: 7-9 fm

(R2out – R2

side) w.r. to KT: emission duration:> ~ 2-3 fm/c anomalously small

“HBT puzzle”

TK

NA49 from QM2004 proceedings

RHIC

Rou

t,si

de

Rlo

ng

55Expectation for ALICE (from ALICE Expectation for ALICE (from ALICE PPR)PPR)

Source size (≈ RoRsRl) should grow with dN/dy> the increment of the three radii can be different.

Initial expectation (based also on the recent results from RHIC): 8-12 fm> but the estimation cannot be extremely definite.

Observed increment in the average phase space density: [Bertsch, Phys. Rev. Lett. 72, 2349 – 77, 789]> influence on the HBT radii expectation > Possibility of smaller radii.

TK

S

J. Cramer, INT/RHIC workshop, Dec 15, 2002

lsoTT RRRdydmm

Nd

EKxS

i

32

2

1,

k

Bertsch, Phys. Rev. Lett. 72, 2349 – 77, 789

66

Large Hadron ColliderLarge Hadron Collider

http://www.cern.ch

~9 km

LHC

SPS

CERN

77

worst (confusion) seen up to now:

STAR

Some numbers related to ALICE and Some numbers related to ALICE and LHCLHC

5.5 A TeV Pb-PbExpected multiplicity (dNch/dy)y=0 :> Major uncertainties not completely

resolved> Still no safe way to extrapolate> Simple scaling form RHIC

~2500> Safe guess ~1500 – 6000> Worst case ~8000

◊ Baseline in the project

Luminosity for Pb-Pb:> Lmax = 11027 cm-2s-1

…and the worst case for

ALICEALICE

88

The ALICE detectorThe ALICE detector

ITSSmall pt tracking, parameter refinement at vertexVertexing

ITSSmall pt tracking, parameter refinement at vertexVertexing

TPCTracking, dE/dxTPCTracking, dE/dx

TRDElectron identificationTRDElectron identification

TOFPID at high pt

TOFPID at high pt

HMPIDPID (RICH) at very high pt

HMPIDPID (RICH) at very high pt

PHOS,0 PHOS,0

MUON-Arm -pairs MUON-Arm -pairs

PMD multiplicityPMD multiplicity

99

Inner Tracking System (ITS)Inner Tracking System (ITS)

6 Layer, 3 technologies (occupancy ~2% at max multiplicity)

> Silicon Pixels (0.2 m2, 9.8 Mchannels. Single chip size = 50x425 μm)> Silicon Drift (1.3 m2, 133 kchannels) > Double-sided Strip (4.9 m2, 2.6 Mchannels)

Rout=43.6 cm

Lout=97.6 cm

SPD

SDD

SSD

1010

ITS performancesITS performances

Layer 1, 2Pixel

Layer 3, 4Drift

Layer 5, 6Strip

(r) (m) 12 38 20

z (m) 100 28 830Two track resolution r (m) 100 200 300Two track resolution z (m) 850 600 2400

Cell size (m2) 50X425 150X300 95X40000

X/Xo (1%) 2.0 2.2 1.76

1111

Track propagation to the closest distance from the interaction point

> Best resolution for all track parameters

Tracking of low transverse momentum particlesPrimary vertex estimation before tracking (and used as a constraint for primary particles tracking)Secondary vertices detection

> D, B> strangeness

Event characterization in a “high-rate acquisition phase”

> TPC powered off ITS unique tracking device

◊ Neural Tracking in the ITS stand-alone

> Can HBT be done with these data?

ITS purposesITS purposes

10 100pt (GeV/c)50

p/p

(%

)

10

30

50

ALICE PPR CERN/LHCC 2003-049

1212

Tracking in ALICETracking in ALICE

“Standard TPC + ITS” tracking Seed determined in the external pad-row of TPC Seed propagation through TPC via a Kalman Filter algorithm Track propagation in the ITS still with the Kalman Filter

[rif.: A. Badalà et al.: NIM A 485 (2002) 15] Back-propagation to outermost TPC TRD TOF Final back-propagation again to the interaction point

TIME PROJECTION TIME PROJECTION CHAMBER CHAMBER

Up to 180 pts / track [starting point]

INNER TRACKING SYSTEM INNER TRACKING SYSTEM Up to 6 pt / track

[resolution improvement]

1313

Why an ITS stand-alone tracking?Why an ITS stand-alone tracking?

“high-rate acquisition mode”:> how: getting data from the “fast” modules only (ITS is

one)> target: physics analysis requesting huge statistics> requirements: good efficiency in the high transverse

momentum range (pt >1 GeV/c).

Algorithm: Denby-Peterson neural network

1414

How the neural network worksHow the neural network works

1

exp1

T

awa j jij

i

Activation variation becomes smaller after a number of cycles

i i

i

aa

N1Final stabilization

0

0

1,0

ij

ij

i

w

w

a correlation increases activation competition decreases activation

Fully connected ANN

1515

Map of real-valued activations between 0 and 1

Comparison with a threshold S

Binary activation map (“on” / “off”)

otherwise 0

if 1 Saa real

bin

How the neural network worksHow the neural network works

1616

ITS rec-points Neurons = segments between points in consecutive layers

Competition negative constant weight Bw I

Chain positive variable weight depending on the alignment

(NB: we look for HIGH pt particles) n

E Aw sin1

BAD alignment:

wE 0 GOOD alignment:

wE A

ImplementationImplementation

No relations among non connected segments

1717Stand-alone tracking efficiency & Stand-alone tracking efficiency & resolutionresolution

Parameter Resolution

Δpt / pt (%) 7.45±0.08

φ (mrad) 1.90±0.03

λ (mrad) 1.83±0.02

Dt (μm) 77.6±0.9

Dz (μm) 164.8±1.5

Track reconstruction: Kalman Filter>Seed: from a Riemann Sphere global fit>Propagation: vertex layer 6 vertex

Efficiency calculation:> “Found” track: a track where at least 5 of 6

points share the same GEANT label> “Fake” track: otherwise> “Findable” track: a track with generates at least

5 points in the ITS> EFFICIENCY = “found tracks” / “findable tracks”

T 4.0

40000

B

d

dN

y

good

“fake”

1818

HBT analysis packageHBT analysis packageDeveloped by the ALICE group from the Warsaw University of Technology (J. Pluta. P. Skowronski & C.)

1919

HBT simulation in ALICEHBT simulation in ALICE

HBT processor (after-burner)> shifts the vector momentum of generated particles,

to generate the correlations> Correlations are “fixed” for each event.

Weight method [R. Lednicky, Heavy Ion Physics 3 (1993), 93]> A weight is defined for each particle pair according to:

◊ HBT correlation◊ FSI interactions◊ Space-time coordinates of the pair◊ 4-momenta of single particles

> The correlation function is determined by normalizing the “weighted” pair distribution (w.r. to q) to the “un-weighted” one.

> Advantages:◊ It is possible to modify the source parameters or the particle pair

to study without having to re-generate the event – Much CPU time

◊ It is possible to make a systematic study of HBT signals in different condition with the same statistical sample.

2020

Ingredients of the used simulation Ingredients of the used simulation

HBT effects simulation in AliRoot:> Weight method

Generator: parameterized HIJING> 160 events at multiplicity (dNch/dy)y=0 = 4000

Magnetic field: 0.4 T“Perfect” PID:> Realistic PID for high-momentum particles in the ITS requires

matching with outer ALICE modules (TRD, TOF)◊ Under study

Gaussian 3-dimensional source simulation (only one R parameter to fix)

> R = 6, 8, 10, 12 fm> λ = 0.5, 0.75, 1> Coulomb effect included

1-D correlation function studied (Qinv)

2121Reconstruction effects and evidence of Reconstruction effects and evidence of HBT signalHBT signal

Double-Track efficiency Cluster finding Tracking

HBT enhancement

2222Comparison:Comparison:HBT generator turned HBT generator turned onon vs. vs. HBT generator HBT generator turned turned offoff

5.0R = 6 fmR = 8 fmR = 10 fmR = 12 fm

2323Compatibility between CF and flat (=1) Compatibility between CF and flat (=1) distributionsdistributions(outside the fall-down due to double track (outside the fall-down due to double track efficiency)efficiency)

λ = 0.75

R χ2 / Ndof Comp.

R = 6 fm 285.6 / 23

<0.01%

R = 8 fm 90.9 / 23 <0.01%

R = 10 fm

40.6 / 23 5%

R = 12 fm

13.5 / 23 95%

λ = 1.00

R χ2 / Ndof Comp.

R = 6 fm 451.5 / 23

<0.01%

R = 8 fm 151.9 / 23

<0.01%

R = 10 fm

81 / 23 <0.01%

R = 12 fm

34.1 / 23 5%λ = 0.50

R χ2 / Ndof Comp.

R = 6 fm 142.2 / 23

<0.01%

R = 8 fm 42.8 / 23 1%

R = 10 fm

20.2 / 23 5%

R = 12 fm

6.6 / 23 99%

2424Correction for double track Correction for double track efficiencyefficiencyNormalization of CF with HBT effect with respect to the CF without the HBT effect, in order to remove double track efficiency

0.5fm 8 R

2525

Reconstruction of Reconstruction of RR

Comparison of reconstructed R vs. simulated R

Straight line = “optimum” line (reconstructed = generated)

2626

Reconstruction of Reconstruction of λλ

Comparison of reconstructed λ versus simulated λ

Straight line = “optimum” line (reconstructed = generated)

2727

Conclusions and outlookConclusions and outlook

HBT enhancement signal:> Generally visible, for the examined radii & lambda> Reconstructed parameters are underestimated due to

Coulomb FSI◊ Mapping of rec vs. sim◊ More suitable fit being developed within the framework package

> Fundamental requirement: high statistics◊ Statistics problems affect the study at large radii

> Anyway, an encouraging result for further study

TODO:> Increment of statistics

◊ Improve the results for larger radii ◊ Check feasibility for 3-D HBT

> Improvement of low-momentum particles tracking> Study of influence of “realistic” PID

2828

2929

Correlation FunctionsCorrelation Functions

Clear merging effect in Qside and Qlong

Very good resolution and PID> We will anyway correct it but first

we have get rid of the merging effect

3030

Fits to Correl. Functions: 3DFits to Correl. Functions: 3D

3D fit, (range 0-50MeV):

> Qout=7.92 ± 0.03 fm

> Qside=7.84 ± 0.02 fm

> Qlong=8.16 ± 0.02 fm

> =0.87 ± 0.01

> 2/NDF=1.48

2 depends strongly on the maximum range> If wide range is used 2 is good, of course

> Fitted values does not depend on it

Fitted values depend on minimum of the range> Already mentioned merging effect

3131

Coverage in [η - pCoverage in [η - ptt] in ALICE] in ALICE

3232

PID capability of ALICEPID capability of ALICE

0 1 2 3 4 5 (GeV/c)

1 10 100 GeV/c

TRD e / PHOS /

TPC + ITS (dE/dx)

/K

/K

/K

K/p

K/p

K/p

e /

e /

HMPID (RICH)

TOF

hadrons (, K, p): 60 MeV/c < p < 5 GeV/cdE/dx in silicon (ITS) and gas (TPC) + (TOF) + Cherenkov (HMPID)

leptons (e, ), photons, 0 : TRD: p > 1 GeV/c

muons: (Muon Spectrometer)

p > 5 GeV/c

0 in PHOS: 1 < p < 80 GeV/c

3333

ALICE: overwiewALICE: overwiew

La Collaborazione ALICE include più di 1200 persone da oltre 90 istituzioni in 29

paesi

ALICE On-line System multi-level trigger per filtrare il fondo e ridurre la quantità di dati registrati

level 0 – custom hardware8 kHz (160 GB/sec)

level 1 – embedded processors

level 2 – PC’s

200 Hz (4 GB/sec)

30 Hz (2.5 GB/sec)

30 Hz

(1.25 GB/sec)

data storage &

off-line analysis

Peso: 10.000tDiametro esterno: 16,00mLunghezza totale: 25mCampo magnetico: 0.2-0-5Tesla

3434

Event generator: HIJINGEvent generator: HIJING

Main standard of the collaboration

Parameterized version for generation of a “signal-free” event

Variable multiplicity, HIJING-like shaped

Useful for study of tracking efficiency and event reconstruction performances.

Distrib. In ptDistrib. In η

3535

Framework di simulazione e analisiFramework di simulazione e analisi

AliROOT ROOT-based> Object Oriented

◊ Modularità◊ Riutilizzabilità del codice

> Implementazione completa di tutte le fasi di simulazione e analisi> Simulazione dettagliata di tutte le parti (sensibili e non) del rivelatore

ALICE

3636

Multiplicity expectationsMultiplicity expectations

The major uncertainties in the energy dependence are still there (only some improvement with the RHIC data!). Still no safe way to extrapolate> shadowing/saturation (might decrease Nch)> jet quenching (might increase it)> A-scaling (soft vs. hard)

Simple scaling form RHIC (log-log plot) ~2500> safe guess dNch/dη ~ 1500 – 6000

ALICE conservative design: all hardware able to handle ~ 8000

3737

Source-averaged phase space densitySource-averaged phase space density

lsoTTi

i

RRRdydmm

Nd

ExxSd

xSxdKxS

i

32

3

23

2

1

,

,,

k

kk

This is compatible with a not so huge increment of HBT radii going to higher energy.

Important issue for forecasts on HBT radii which will be measured in ALICE

Bertsch, Phys. Rev. Lett. 72, 2349 – 77, 789

S

TK