NA60 results on the spectral function in Indium-Indium collisions

1

NA60 results on the spectral function in Indium-Indium collisions

Sanja Damjanovic NA60 Collaboration

Asilomar, 14 June 2006

2

Outline

Isolation of excess dimuons above hadron decays

Mass spectra (published in PRL)

Shape analysis of mass spectra (new)

Comparison to theory

Acceptance-corrected pT spectra (new)

Comparison to theory

3

5-week long run in Oct.–Nov. 2003

Indium beam of 158 GeV/nucleon ~ 4 × 1012 ions delivered in total ~ 230 million dimuon triggers on tape

present analysis: ~1/2 of total data

Event sample: Indium-Indium

4

Main steps of the data analysis

reconstruction of the event vertex within the segmented target

matching of tracks from muon spectrometer and silicon vertex telescope

assessment of combinatorial background by event mixing

assessment of fake matches by overlay MC and/or event mixing

talk of Andre David

5

Subtraction of combinatorial background and fakes

For the first time, and peaks clearly visible in dilepton channel ; even μμ seen

Net data sample: 360 000 events

Mass resolution:23 MeV at the position

Fakes / CB < 10 %

Progress over CERES: statistics: factor >1000resolution: factor 2-3

6

Track multiplicity from VT tracks for triggered dimuons for

Centrality bin multiplicity ⟨dNch/dη⟩3.8

Peripheral 4–28 17

Semi-Peripheral

28–92 63

Semi-Central 92–160 133

Central > 160 193

Associated track multiplicity distribution

4 multiplicity windows:

opposite-sign pairs combinatorial background signal pairs

new: some part of the analysis also in 12 multiplicity windows

7

Understanding the Peripheral data

Fit hadron decay cocktail and DD to the data

5 free parameters to be fit:

DD, overall normalization

(0.12fixed)

do the fits for all pT and three bins in pT

Extrapolate fit parameters to full phase space

(using particle generator “Genesis”)

8

Comparison of hadron decay cocktail to data

all pT

Very good fit quality

log

9

Full-phase-space particle ratios from the cocktail fits

and nearly

independent of pT; 10% variation due to the

enhanced mostly at low pT (due to ππ annihilation, see later)

General conclusion: peripheral bin very well described in terms of known sources low M and low pT acceptance of NA60 under control

10

Isolation of an excess in the more central data

11

Goal Find excess above hadron decays without fits

Conservative approach Use particle yields so as to set a lower limit to a possible excess

12

● data

-- sum of cocktail sources

including the

Cocktail definition: see next slide

all pT

Comparison of data to “conservative” cocktail

Clear excess of data above cocktail, rising with centrality

fixed to 1.2

But: how to recognize the spectral shape of the excess?

13

Isolate possible excess by subtractingcocktail (without ) from the data

set upper limit, defined by “saturating” the measured yield in the mass region close to 0.2 GeV

leads to a lower limit for the excess at very low mass

and : fix yields such as to get, after subtraction, a smooth

underlying continuum

difference spectrum robust to mistakes even on the 10% level, since the consequences of such mistakes are highly localized.

14

Excess spectra from difference: data - cocktail

all pT

Clear excess above the cocktail , centered at the nominal pole and rising with centrality

Similar behaviour in the other pT bins

No cocktail and no DD subtracted

Phys.Rev.Lett. 96 (2006) 162302

15

Systematics

Systematic errors of continuum 0.4<M<0.6 and 0.8<M<1GeV 25%

Illustration of sensitivity to correct subtraction of combinatorial background and fake matches; to variation of the yield

Structure in region completely robust

16

Shape analysis of

excess mass spectra

17

Excess mass spectra in 12 centrality windows

18

Shape vs. centrality

3/2(L+U) “continuum”

R=C-1/2(L+U) “peak”

RR peak/continuum

nontrivial changes of all three variables at dNch/dy>100 ?

19

RMS of total excess

Consistency with shape analysis

Further rise starting at dNch/dy =100 significant!

(bad fit (2=3) for linear rise above dNch/dy=30)

20

Comparison of data

to model predictions

21

Two alternatives how to compare data to predictions

use predictions in the form

decay the virtual photons * into +- pairs, propagate these through the NA60 acceptance filter and compare results to uncorrected data at the output (presently done for mass spectra in selected pT regions)

correct data for acceptance in 3-dim. space M-pT-y and compare directly to predictions at the input (presently done for pT spectra in selected mass regions)

dydMdp

Nd

T2

*3

conclusions as to agreement or disagreement of data and predictions are independent of whether comparison is done at input or output

22

output:

white spectrum !

understanding the spectral shape at the output

By pure chance, for the M-pT characteristics of direct radiation, without pT selection,the NA60 acceptance roughly compensates for the phase-space factors and directly “measures” the <spectral function>

input:

thermal radiation based on white spectral function

all pT

functionspectralTMMfdMdN )/exp()(/

Acceptance filtering of theoretical prediction:

23

Predictions for In-In by Rapp et al. (2003) for ⟨dNch/d⟩ = 140, covering all scenarios

Theoretical yields normalized to data in mass interval < 0.9 GeV

Only broadening of (RW) observed, no mass shift (BR)

Comparison of data to RW, BR and Vacuum

Data and predictions as shown, after acceptance filtering, roughly mirror the respective spectral functions, averaged over space-time and momenta.

24

New theoretical developments since QM05

Brown and Rho, comments on BR scaling, nucl-th/0509001Brown and Rho, formal aspects of BR scaling, nucl-th/0509002

Rapp and van Hees, parameter variations for 2, hep-ph/0604269Rapp and van Hees, 4, 6… processes, hep-ph/0603084

Renk and Ruppert, finite T broadening, Phys. Rev. C71 (2005) Renk and Ruppert, finite T broadening and NA60, hep-ph/0603110 Renk, Ruppert, Müller, BR scaling and QCD Sum Rules, hep-ph/0509134 Renk and Ruppert, What the NA60 dilepton data can tell, hep-ph/0605330

Skokov and Toneev, BR scaling and NA60, Phys. Rev. C73 (2006) Dusling and Zahed, Chiral virial approach and NA60, nucl-th/0604071

25

Modification of BR bychange of the fireball parameters

Parameter variations for Brown/Rho scaling

even switching out all temperature effects does not lead to agreement between BR and the data

))/(1)(1( 2

0

0*cTTCmm

Van Hees and Rapp, hep-ph/0604269

26

Comparison of data to RH(2+4+QGP)

Vector-Axialvector Mixing: interaction with real ’s (Goldstone bosons). Use only 4 and higher parts of the correlator V in addition to 2

)0,(

),(

2

1)1( 00*

cAVV T

T

Use 4, 6 … and 3, 5… (+1) processes from ALEPH data, mix them, time-reverse them and get +- yields

van Hees and Rapp, hep-ph/0603084

27

Comparison of data to RH(2+4+QGP)

In this model, the yield above 0.9 GeV is sensitive to the degree of vector-axialvector mixing and therefore to chiral symmetry restoration!

whole spectrum reasonably well described, now even in absolute terms

direct connection to IMR results >1 GeV from NA60

Van Hees and Rapp, hep-ph/0603084

28

pT- dependences

Comparison of data to RW(2+4+QGP)

theoretical results plotted in absolute terms

29

Comparison of data to RR

Spectral function only based onhot pions (Dyson-Schwinger) , no baryon interactions included

Continuum contribution from partons, dominating the region >1GeV

broadening described, except for low-mass tail

Theoretical results obtained in absolute terms

talk of Joerg RuppertRenk and Ruppert, hep-ph/0603110

30

Comparison of data to RR

pT dependences

talk of Joerg Ruppert

theoretical results plotted in absolute terms

31

Acceptance-corrected excess pT spectra

Preliminary

32

reduce 3-dimensional acceptance correction in M-pT-y to 2-dimensional correction in M-pT, using measured y distribution as an input

use slices of m = 0.1 GeV and pT = 0.2 GeV

resum to three extended mass windows

0.4<M<0.6 GeV 0.6<M<0.9 GeV 1.0<M<1.4 GeV

Present strategy of acceptance correction

subtract charm from the data (based on NA60 IMR results)before acceptance correction

33

Experimental results on the y distribution of the excess

use measured mass and pT spectrum as input to the acceptance correction in y (iteration procedure)

agreement betweenthe three pT bins

results close torapidity distribution of pions (from NA49) for the same √s, as expected (RR)

34

Excess pT spectra for three centrality bins

hardly any centrality dependence

significant mass dependence

(spectra arbitrarily normalized)

35

Centrality-integrated excess pT spectra

significant mass dependence (also vs. mT, see below )

possible origin:

different physics sources

radial flow

p-dependence of in-medium spectral function

(arbitrarily normalized at pT=1GeV)

36 at high pT, rho like region hardest, high-mass region softest !

Illustration of mass dependence of pT spectra

differential fits to pT

spectra, assuming locally 1-parameter mT scaling and using gliding windows of pT=0.8 GeV local slope Teff

very low Teff at low pT, enormous dynamic range, hardly compatible with flow alone (for only one component) systematic errors <15 MeV

37

Systematics of low-pT data: acceptance

pT spectrum of at low pT much flatter (higher Teff)

acceptance of inbetweenthat of the twomass windows

enhanced yield at low pT not due to incorrect acceptance

38

Systematics of low-pT data: combinatorial background

enhanced yield at low-pT seen at all centralities, including the peripheral bin

estimate of errors at low pT, due to subtraction of combinatorial background: peripheral 1%semiperipheral 10% semicentral 20%central 25%

enhanced yield at low pT not due to incorrect subtraction of combinatorial background

39

Comparison of data

to model predictions

40

Comparison to theory: mass window 0.6<M<0.9 GeV

1-parameter differential mT fits

at low pT: better description by RH at higher pT much better description by RR (freeze-out ρ incl.)

(arbitrarily normalized at pT=1 GeV)

in detail, data different from any theoretical prediction

41

Comparison to theory: mass window 1.0<M<1.4 GeV

1-parameter differential mT fits

(arbitrarily normalized at pT=1 GeV)

RH: dominantly hadronic processes (4); lower T; role of flow? RR: dominantly partonic processes (qq); high T; low flow

comparison to data inconclusive

42

Conclusions (I) : data

• pion annihilation seems to be a major contribution to the lepton pair excess in heavy-ion collisions at SPS energies

• no significant mass shift of the intermediate

• only broadening of the intermediate

• strong mass dependence of pT spectra

43

models predicting strong broadening roughly verified

Conclusions (II) : interpretation

all models predicting strong mass shifts of the intermediate including Brown/Rho scaling, are not confirmed by the data

theoretical investigation on an explicit connection between broadening and the chiral condensate clearly required

pT spectra not yet fully described by any theory, rich in detail, promising handle on emission sources and fireball dynamics

44

http://cern.ch/na60

Lisbon

CERN

Bern

Torino

Yerevan

CagliariLyon

Clermont

Riken

Stony Brook

Palaiseau

Heidelberg

BNL

~ 60 people13 institutes8 countries

R. Arnaldi, R. Averbeck, K. Banicz, K. Borer, J. Buytaert, J. Castor, B. Chaurand, W. Chen,B. Cheynis, C. Cicalò, A. Colla, P. Cortese, S. Damjanović, A. David, A. de Falco, N. de Marco,

A. Devaux, A. Drees, L. Ducroux, H. En’yo, A. Ferretti, M. Floris, P. Force, A. Grigorian, J.Y. Grossiord,N. Guettet, A. Guichard, H. Gulkanian, J. Heuser, M. Keil, L. Kluberg, Z. Li, C. Lourenço,

J. Lozano, F. Manso, P. Martins, A. Masoni, A. Neves, H. Ohnishi, C. Oppedisano, P. Parracho, P. Pillot,G. Puddu, E. Radermacher, P. Ramalhete, P. Rosinsky, E. Scomparin, J. Seixas, S. Serci, R. Shahoyan,P. Sonderegger, H.J. Specht, R. Tieulent, E. Tveiten, G. Usai, H. Vardanyan, R. Veenhof and H. Wöhri

The NA60 experiment

45

BACKUPBACKUP

46

Peripheral DataPeripheral Data

47

48

enhanced yield at low-pT seen at all centralities, including the peripheral bin

Excess pT spectra for four centrality bins

peripheral bin at high-pT slightly softer

49

The region (small M, small pT)

is remarkably well described

Comparison of hadron decay cocktail to data

→ the (lower) acceptance of NA60

in this region is well under control

pT < 0.5 GeV

50

Comparison of hadron decay cocktail to data

Again good agreement

between cocktail and data

0.5 < pT < 1 GeV

pT > 1 GeV

51

Pt spectraPt spectra

52

very low Teff at low pT, enormous dynamic range, hardly compatible with flow alone (for only one component)

at high pT, rho like region hardest, high-mass region softest !

Illustration of mass dependence of pT spectra

integral fits to pT

spectra, assuming 1-parameter mT scaling and varying the upper cut in pT (bad χ2 for full pT range)

53

Comparison to theory: mass window 0.4<M<0.6 GeV

1-parameter differential mT fits

(arbitrarily normalized at pT=0.7 GeV)

at low pT: better description by RH/DZ at higher pT much better description by RR

in detail, data different from any theoretical prediction

54

Teff from differential fits to pT spectra

55

Shape analysis

56

RMS of continuum

If observed peak subtracted, the remaining continuum described by RMS of a flat distribution, independent of dNch/dy

Subtraction of cocktailcreates dip in the middle RMS>RMSflat

57

of peak (from Gaussian fit)

Sigma of peak consistent with sigma of cocktail/vacuum independent of dNch/dy

58

Further theoretical results

59

Comparison of data to RW, BR and Vacuum

pT dependence same conclusions

60

Modification of DM by

• “fusion” of the two scenarios

Dropping Mass (DM) vs Rapp/Wambach

Results of Rapp (8/2005):(not propagated through NA60 acceptance filter)

Neither fusion nor parameter change able to make DM scenario unobservable

ncTTCmm ))/(1)(1( 2

0

0*

• change of the fireball parameters

61

*2*2 gmgm

Skokov and Toneev on BR scaling

ncTTmm ))/(1)(15.01( 2

0

0*

full dynamical model including deconfinement transition

n=0.3

n=0

standard scaling of pole mass, no broadening

same modification to the vector dominance coupling

theoretical results normalized to data

even for n=0, no agreement with the data (like Rapp)

62

Chiral Virial Approach Dusling/Zahed

First attempt to describe the centrality dependence of the excess data.

Reasonable description, but increasing overestimate of central peak

63

DZ Integrated Dimuon Rates

225

150

B

T

a1a1

64

AcceptanceAcceptance

NA60 vs CERESNA60 vs CERES

65

CERES

reduction of acceptance at low M and low pT similar

Comparison of lepton pair acceptance

CERES

66

Acceptance comparison of CERES and NA60

acceptance variation of NA60 between <RW> and <BR> factor of 3

acceptance variation between 0.4 and 0.8 GeV different by a factor of 2 between CERES and NA60

Source:Rapp/Wambach

low pT dominated

67

CERES NA60

Phase space coverage in pT-mass plane

68

Comparison of NA60 to CERES

Suppression of low mass part of RW similar in CERES and in NA60

69

Phase space coverage in y-pT plane

Comparison to CERES

MC simulation with RW (low-mass and low-pt dominated):

both acceptances shifted relative to midrapidity, but difference only δy=0.3

70

Phase space coverage in y-pT plane

Examples from MC simulations

Optimal acceptance:

at high mass, high pT

<y> = 3.5

at low mass, low pT

<y> = 3.8

Shift of acceptance away from midrapidity not much different from CERES

71

Experimental set-up

72

2.5 T dipole magnet

hadron absorber

• Origin of muons can be accurately determined• Improved dimuon mass resolution

Matching in coordinate

and momentum space

targets

beam tracker

vertex trackermuon trigger and tracking

magnetic field

or

!

Measuring dimuons in NA60: concept

73

BackgroundBackground

74

Analysis Topics

LMR IMR HMR

1) Low Masses:vector mesons + continuum rad. (LMR)

2) Intermediate Masses: charm + continuum rad. (IMR)

3) High Masses:J/, ’ and DY (HMR)

75

Selection of primary vertex

Beam Trackersensors

windows

The interaction vertex is identified with better than 20 m accuracy in the transverse plane and 200 m along the beam axis.

(note the log scale)

Present analysis (very conservative):

Select events with only one vertex in the target region,

i.e. eliminate all events with secondary interactions

76

A certain fraction of muons is matched to closest non-muon tracks (fakes). Only events with 2 < 3 are selected.

Fake matches are subtracted by a mixed-events technique (CB) and an overlay MC method (only for signal pairs, see below)

Muon track matching

Matching between the muons in the Muon Spectrometer (MS) and the tracks in the Vertex Telescope (VT) is done using the weighted distance (2) in slopes and inverse momenta. For each candidate a global fit through the MS and VT is performed, to improve kinematics.

77

Determination of Combinatorial Background

Basic method:

Event mixing

takes account of

charge asymmetry

correlations between the two muons, induced by magnetic field sextant subdivision trigger conditions

78

Combinatorial Background from ,K→ decays

Agreement of data and mixed CB over several orders of magnitude

Accuracy of agreement ~1%

79

Fake Matches Fake matches of the combinatorial background are automatically subtracted as part of the mixed-events technique for the combinatorial background

Fake matches of the signal pairs (<10% of CB) are obtained in two different ways:

Overlay MC : Superimpose MC signal dimuons onto real events. Reconstruct and flag fake matches. Choose MC input such as to reproduce the data.

Event mixing : More complicated, but less sensitive to systematics

80

Fake-match background

example from overlay MC: the fake-match contribution localized in mass (and pT) space: = 23 MeV, fake = 110 MeV; fake prob. 22%

complete fake-match mass spectrum agreement between overlay MC and event mixing, in absolute level and in shape, to within <5%

81

Signal and background in 4 multiplicity windows

S/B

2 1/3

1/8 1/11

Decrease of S/B with centrality, as expected

82

Mixing events

The event mixing requires 12 “pools” of single muons Two muon charges, times Six spectrometer trigger sextants

And we can only mix single muons from The same target The same centrality class The same field polarities, running conditions, etc(~6000 event samples)

For the mixing we only use single muons from the like-sign dimuon triggers

The mixed event technique gives both shape and absolute scale

83

Combinatorial background: normalization Assume that the probability to accept the

pair ij is Pij = Pi Pj,

Pi and Pj are the single muon probabilities for sextants i and j (as if NA60 would have collected single muon triggers)

The observed number of pairs ij is Nij

Then the single muon underlying probabilities are

And the absolute probability to have a given muon is

Pi

N

146

3 52

jiij PPNN

ij

jiij

jiiii PPNPPNN )1(

PPP

ii

i 1,

Muon spectrometertrigger system sextants

84

Sensitivity of Sensitivity of

difference proceduredifference procedure

85

Sensitivity of the difference procedure

Change yields of , and by +10%:

enormous sensitivity, on the level of 1-2%, to mistakes in the particle yields.

The difference spectrum is robust to mistakes even on the 10% level, since the consequences of such mistakes are highly localized.

86

Excess spectra from difference data-cocktail

No cocktail and no DD subtracted

pT < 0.5 GeV

Clear excess above the cocktail , centered at the nominal pole and rising with centrality

Similar behaviour in the other pT bins

87

medium modification medium modification of of

88

89

Masking of by pion annihilation

Central Bin

The chances are better at high pT but no effect is visible there!

90

Medium modification of

Flattening of the pt distributions develops very fast with centrality.

Compatible with radial flow?

Or indication of broadening in the medium??

91

results from KEK-E325results from KEK-E325

92

Results from KEK-E325

ee pairs from 12 GeV pA collisions

superb mass resolution of 10 MeV

Tail below the interpreted as , shifted down in mass

93

Alternative interpretation: / interference

Rob Veenhof, Winter Workshop on Nucl.Dynamics, Breckenridge 2005