gluon polarisation overview

A.Magnon IWHSS ‘08 – TorinoMarch 31, 2008

1

Gluon Polarisation Overview

quark contribution to nucleon spin. Why G ? G from scaling violations G from hadron production - Open charm - COMPASS

- High pT hadrons pairs & single - COMPASS/HERMES

G from pp collisions - RHIC A. Magnon (CEA-Saclay/IRFU & COMPASS)


2

Early measurements of (1)

SLAC Polarized electrons

large, “as expected” 1976-1983

g1p

xBj

SLAC

Compatible with =0.6

∫g1p

xBj

xBjg1p

Ellis-Jaffe = 0.6

EMC

EMC @ CERN polarized

Access lower x, = 0.12 ± 0.17

→ ” Spin crisis ’’ 1988


3

Early measurements of (2)

HERMES, SLAC high precision, SMC @ CERN lower x

g1 for proton & neutron (deuteron) …

Bjorken Sum Rule relates proton & neutron g1=∫g1dx003.0181.0)(

6

1 2111 QC

g

g NS

V

Anp

024.0012.011 174.0

np

Theory, Q2=5 GeV2

SMC

= 0.2 - 0.3 confirmed to be small

Bjorken OK + s determination + first flavor separation …

1998


4

COMPASS @ CERN, 160 GeV

HERMES @ DESY, e- 27 GeV

= 0.30 ± 0.01 (stat) ± 0.02 (evol)

=0.33 ± 0.011 (stat) ± 0.025 (theo) ± 0.028 (evol)

COMPASS fit to g1 p, n, d world data, MS scheme, Q2 = 3 (GeV/c)2

PLB 647 (2007) 8

HERMES from g1d data, MS scheme, Q2=5 (GeV/c)2,

neglecting x < 0.02 contrib., PRD75 (2007) 012007

s + s= - 0.08 ± 0.01 (stat) ± 0.02 (syst)COMPASS data alone

s + s= - 0.085 ± 0.013 (th) ± 0.008 (exp) ± 0.009(evol)

Recent measurements of


5

Why measure G ?

½ = ½ΔΣ + ΔG + <Lq> + <Lg>

Measurement of G important :

1 – How are gluons polarized ? 2- Low value of a0 could be due to axial anomaly if G is large. (A. Efremov O.Teryaev, G. Altarelli – G. Ross)

3 – How large is parton orbital angular momentum

a0 =


6

G from scaling violations G from hadron production - Open charm - High pT hadrons (pairs, single) G from pp collision

How to measure G ?


7

Q2 = 3 GeV2

COMPASS NLO QCD fit

Comparison of fits - disagreement of data with previous QCD fits (LSS05, GRSV, BB)


8

Q2 = 3 GeV2

New COMPASS g1d

G > 0 or G < 0, |G| ~ 0.3 a0 = 0.33 ± 0.03 ± 0.05 s = -0.08 ± 0.01 ± 0.02

COMPASS NLO QCD fit

Δ G

ΔG = - 0.31

ΔG = 0.34

G = 0.34

G = - 0.31

2006


9

G from scaling violations G from hadron production (PGF) - Open charm - High pT hadrons (pairs, single) G from pp collision

How to measure G ?


10

measure

calculate and

using Monte Carlo

Photon-gluon fusion (PGF)

Gluon polarisation is measurable in PGF

GGaRA

pgfpgfLL

LLA

pgfR

pgfa

N


11

G/G from open charm

c

c

c -> D* -> s D0 -> Ks

cleanest process wrt physical bkgr

combinatorial bkgr, limited statistics

so far LO analysis, NLO in progress

Open charm, single D meson

N


12

G/G from open charm

πKD0 ππKπDDs

0*

COMPASS Data: 2002,2003,2004 & 2006

160 GeV beam & 6LiD target

nD0 = 37398

nD* = 8675


13

G/G from open charm

)(/gLLLLx

GGa

BS

SDA

Analysis uses both aLL and S/(S+B) weighting

aLL obtained from Neural Network trained on MC (AROMA):

input variables : Q2, xbj, y, pT, zD

S/(S+B) given by a parameterization: input variables : target cell,

fPμaLL, pK, θK, zD, cosθ*, pT, RICH Likelihoods

Weighting brings significant improvement in statistics due

to large variations of aLL and S/(S+B) in phase-space


14

G/G from open charm

5 bins in = S/(S+B)

D0 -untagged

D* -tagged


15

G/G from open charma

LL

gen

era

ted

aLL reconstructed

2006 aLL parameterization


16

“ New ”

2002 – 2006 data D0 + D* G/G = -0.49 ± 0.27 (stat) ± 0.11 (syst) @ <xg> ~ 0.11, <2> ~ 13 (GeV/c)2

G/G from open charm


17

G from scaling violations G from hadron production (PGF) - Open charm - High pT hadrons pairs, Q2 > 1 GeV/c2

G from pp collision

How to measure G ?


18

ΔG/G from high pT hadron pairs

Photon Gluon Fusion ~ 30%

Leading Order QCD Compton

Q2 > 1 (GeV/c)2

g q q +

Resolved

Q2 < 1 (GeV/c)2

q,g

q,g

BKGRPGFLLPGFLL

AGGaRA


19

Analysis uses parameterization of RPGF, RQCDC, RLO, aLL

PGF,

aLLQCDC, aLL

incl, xg, xC, … etc based on Neural Network

trained on MC (LEPTO for Q2 > 1).

No cut on NN which assigns to each evt. a probability to

originate from LO, PGF or COMPTON.

Dependence on PDFs studied

Parton shower (NLO process) added

Detailed studies of systematics



20


Two parameters:

O1 & O2 to express

fractions R (PGF, LO

or QCDC) for each

high pT event


21

Leading hadron

Sub-leading hadron



22


Probabilities (fractions) of LO, QCDC, PGF : Monte Carlo vs Neural Network


23

“ New ”

2002 – 2004 data: High pT, Q2 > 1 GeV/c2 G/G = 0.08 ± 0.10 (stat) ± 0.05(syst) @ <xg> = 0.082, (range: 0.055 – 0.123) 2 ~ 3 (GeV/c)2



24

(Released 2 Oct. 2006, SPIN2006)

2002 – 2004 data: High pT, Q2 < 1 GeV/c2

G/G = 0.016 ± 0.058 (stat) ± 0.055 (syst)

@ <xg> = 0.085, 2 = 3 (GeV/c)2



25

New high pT

G/G, direct measurements

New open charm

GRSV, G

std, 0.6

min, 0.2

QCD Fits

|G| ~ 0.3

max, 2.5


26

Accurate G/G from COMPASS data (2002 – 2004) from high pT hadron pairs, Q2 < 1 GeV/c2 and Q2 > 1 GeV/c2 (new) G/G small (~ 0) @ <xg> = 0.08

Significant improvement for G/G from open charm (2002 – 2004 + 2006) and aLL + S/B weighting. Also G/G (~ 0) @ <xg> = 0.11

Similar conclusion from new HERMES analysis

G/G, direct measurements


27

G from scaling violations G from hadron production - Open charm - high pT hadrons (pairs, single) G from pp collision

How to measure G ?


28

RHIC: polarized pp collider

Year P L(pb-1) P4L(*)

2004 40% 3 0.08

2005 50% 13 0.8

2006 60% 46 6

(*) G.Bunce Dubna Spin07


29

q

q

G

G

G

G

G

G

pp collisions @ PHENIX & STAR

Reactions pp -> X, jet X, X, cc X, probe gluon Measure always product of 2 observables MC required to determine fraction of process

q

q

q

q


30

pp collisions @ PHENIX & STAR

Considerable progress in pQCD NLO calculations

Jäger,Schäfer, Stratmann,Vogelsang; de Florian

Jäger,Schäfer, Stratmann,Vogelsang; Signer et al.

Gordon,Vogelsang; Contogouriset al.;Gordon, Coriano

Bojak, Stratmann;


31

Unpol. Cross Section in pp

pp0 X : hep-ex-0704.3599

pp X: PRL 98, 012002

Good agreement between NLO pQCD calculations and data confirmation that theory can be used to extract spin dependent pdf’s from RHIC data

PHENIX data


32

Log10(xgluon)NLO pQCD: 0 pT=29 GeV/c xgluon=0.020.3 GRSV model: G(xgluon=0.020.3) ~ 0.6G(xgluon =01 ) Each pT bin corresponds to a wide range in xgluon, heavily overlapping with other pT bins. These data is not much sensitive to variation of G(xgluon) within our x range. Any quantitative analysis should assume some G(xgluon) shape

From pT to xgluon (PHENIX, 0X)

√s=200 GeV

G.Bunce

Dubna Spin07


33

Calc. by W.Vogelsang and M.Stratmann

GRSV “standard”, G(Q2=1GeV2)=0.4, is excluded

by data on >3 sigma level: 2(std)2min>9

Only exp. stat. uncertainties are included (the effect of syst. uncertainties is expected to be small in the final results) Theoretical uncertainties are not included

From ALL to G (PHENIX, 0 with GRSV)

“3 sigma”

G.Bunce

Dubna Spin07


34

From ALL to G (STAR, jetX with GRSV)

2005 STAR preliminary

Systematic error band

Measured Jet PT (GeV)

GRSV DIS

Large gluon polarisation scenario is not consistent with data

J.C.Dunlop

Dubna Spin07


35

Different sensitivity of + and - to the sign of G …. e.g. G > 0

πLL

πLL

πLL AAA

ο

STARSTAR

No constraint on G yet …

STAR, inclusive ± production (mid - η)

J.C.Dunlop

Dubna Spin07

Dramatic increase in precision in Run 2006


36

G from RHIC

High statistics available to constrain G in xg range (0.02 – 0.3)

G not large (consistent with zero)

“Standard” scenario, G (Q2=1GeV2) = 0.4, is excluded

by data on > 3 sigma level: 2(std)2min > 9

(PHENIX ?)

Theoretical uncertainties might be significant


37

RHIC, prospects

Improve exp. (stat.) uncertainties, move to higher pT

- more precise G in probed x range - probe (lower) and higher x and constrain G vs x Different channels - different systematics - different x, - gq -> q (pp -> jet), sensitive to G sign, parton kinematics well constrained, theoretically clean Different √s, 62 GeV, 200 GeV, 500 GeV

Substantial FOM = P4L needed


38

Conclusion, possible scenarios

G Lq Lg

½ = 1/2 × 0.3 + 0.35 + 0 + 0

½ = 1/2 × 0.3 + 0.0 + 0.35

½ = 1/2 × 0.3 - 0.35 + 0.70

COMPASS/RHIC JLab/HERMES/COMPASS

From COMPASS & RHIC:

G =|∫G(xG)| ≤ 0.4 ?

≈ a0 = 0.3

a0 =


39

Additional slides


40

Measurements of G/G

xg binning

High pT

in 2006


41

PHENIX, different s

s=62 GeV 0 cross section described by NLO pQCD within theoretical uncertainties

Sensitivity of Run6 s=62 GeV data collected in one week is comparable to Run5 s=200 GeV data collected in two months, for the same xT=2pT/s

s=500 GeV will give access to lower x; starts in 2009


42

Method- parameterize polarised parton distributions at Q0

2

e.g. qi ~ xi (1-x)i(1+ix)

- DGLAP evolution to measured Q2 - calculate g1 and fit all existing g1 data together

DGLAP evolution equations rule ∂/∂ lnQ2

dependence of parton distribution functions

and G coupled in the evolution

→ Extract G(x)

G from scaling violations


43

AAC – NLO, hep-ph/0603213 including g1 new data from HERMES, COMPASS and JLAB + PHENIX ALL 0

G = 0.31 ± 0.32 at Q2=1 GeV2

xGxuv

xdv

Global QCD analysis: AAC - NLO

xq

q = - 0.050 ± 0.32


44

New COMPASS A1d data

PLB647 (2007) 8

gluon polarisation overview

Documents

open charm analysis

open charmcompass data

open charmccc d

data d0 d

open charm5 bins

d world data

progressopen charm

g1d data