Gluon polarisation from high transverse momentum hadron pairs COMPASS

Transcription

1 Gluon polarisation from high transverse momentum hadron pairs COMPASS Luís Silva LIP Lisbon lsilva@lip.pt On behalf of the COMPASS Collaboration 22 Jul 2008

2 Contents Motivation Direct measurement G/G COMPASS experiment High pt analysis, Q2 > 1 (GeV/c)2 High pt analysis, Q2 < 1 (GeV/c)2 G/G results Conclusion and Outlook

3 The Nucleon Spin SN = ½ = ½ ΔΣ The naïve Quark Parton Model (QPM) considers only the contribution from quarks =1 Applying Relativistic Corrections 0.75 Using the Ellis Jaffe Sum rule and Hyperon decays 0.6 Phys.Rev.D9,(1974)1444 Erratum ibid.d10,(1974)1669. (J.Ellis and R.Jaffe)

4 The Nucleon Spin SN = ½ = ½ ΔΣ In 1988 EMC measured = 0.12 ± 0.17 (Phys.Lett.B206,364) Today world data results, including COMPASS, gives: = 0.30 ± 0.01(stat.) ± 2 3 GeV/c)2 (using QCD NLO fits) Phys.Lett.B647, (2007)8 Spin crisis! Where is the remaining part? How is the nucleon spin composed?

5 The Nucleon Spin SN = ½ = ½ ΔΣ + ΔG + L Adding the following contributions: G from gluons L = Lg + Lq from orbital angular momenta of quarks and gluons How much is the contribution from gluons and from L? Spin Puzzle

6 Direct measurement of ΔG/G * g q q A PGF = * N N PGF PGF N PGF N PGF G/G Photon-gluon fusion process (PGF) Experiments with polarised beam and target could be sensible to gluon helicity

7 Direct measurement of ΔG/G To tag this process there are two procedures concerning event selection : Open-charm meson (C.Franco talk) * g q q Provides the purest sample of PGF events, almost free from background contamination. Not much MC dependent. * Low statistics. NLO corrections can be important. High transverse momentum hadrons (Q2<1 and Q2>1 (GeVc)2) Much more statistics. Physical background: strongly model dependent, requires a very good agreement between Data and MC. Photon-gluon fusion process (PGF)

8 The COMPASS Spectrometer Trackers Magnets RICH NIM A577 (2007) 455 Electromagnetic Calorimeters Hadronic Calorimeters 50 Absorbers m Two staged spectrometer: Target SM2 LAS and SAS Polarised beam and target SM1 + μ V e 0G 16 Acceptance: 70 mrad (2004) About 350 detector planes Track reconstruction p > 0.5 GeV PID (RICH), K and p above 2, 9 and 18 GeV

9 High pt Analysis, Q2 >1 (GeV/c)2

10 High pt Analysis, Q2 >1 GeV2 How is ΔG/G measured? ΔG/G from PGF process: Estimated from MC ΔG A = a G The asymmetry for PGF process PGF PGF PGF LL LL Fraction of PGF process Tot a PGF LL Δ PGF = PGF What we want to extract

11 High pt Analysis, Q2 >1 GeV2 We access APGF by measuring of the helicity asymmetry of two high pt hadrons at large Q2, this measurement includes also contribution other physical processes: PGF COM LO A2h = A A A LL PGF C ΔG LO 2h PGF LO C LO A LL x = x g a LL A1 x C a LL Tot A1 x Bj D Tot Tot G A1LO e 2i Δqi QCDCompton i e i2 q i PGF LO DIS i The same decomposition can be done for inclusive asymmetry : A incl LL x incl Bj ΔG incl incl, PGF PGF = x g a LL G Tot LO 1 A x incl incl C a incl, C LL C Tot incl LO 1 A x incl Bj LO D Tot incl

12 G/G for High pt, Q2 >1 GeV2 The final formula for the gluon polarization: 2h LL R LO A1 x C A1 x C ' ΔG av A x Bj A1 x Bj x g = D β 1 β2 G β β R LO β β incl β =a β 1= PGF LL R PGF a PGF, incl LL R incl PGF R LO incl LO R incl R LO a PGF,incl LL RC R incl PGF Rincl LO 1 C C, incl a R a RC incl LL C LL incl R LO R LO C LL a D i R i = Tot incl C C, incl R C R C a LL β 2 =a LL 2 D Rincl LO A2hLL is the measured 2 h asymmetry. all and R are estimated using MC. A1 are taken using a parametrisation on inclusive data. (EPJ C52 (2007)255)

13 Event selection Interaction vertex which contains an incoming and a scattered muon and at least 2 outgoing hadrons For Deep Inelastic Scattering variables: Q2>1 (GeV/c)2 and 0.1 < y < 0.9 Each hadron is required to have: pt > 0.7 GeV/c For the pair of hadrons is required an invariant mass m > 1.5 GeV/c2 and z1 + z2 < 0.95 Years Statistics all years k Q 2= q2 q=k k ' = E E ' y= E Q2 x= 2M q k'

14 Monte Carlo Simulation Two MC samples were used in the analysis: high pt and inclusive samples. Full chain of MC has been used: Generator (LEPTO) + Apparatus Simulation (GEANT) + Reconstruction Program. PDF: MRST2004LO. High pt sample: MC parton shower on has been used in the analysis. MC parton shower off used to evaluate systematics. Two generator tunings were used: Default and so called COMPASS tunings: 2 b m 1 T D z 1 z a exp 2 z PARJ(21) PARJ(23) PARJ(24) a b Default Compass

15 Data Monte Carlo comparison, Q2>1(GeV/c)2 high pt sample (x,y and Q2) The agreement between data and MC is good

16 Data Monte Carlo comparison, Q2>1(GeV/c)2 Data/MC high pt sample: hadron variables: pt1, pt2

17 Weighting method The idea is to enhance the PGF events sample and to reduce the physical background. A weight is applied on event by event basis: W = fdpb Therefore for every event we have to know: incl incl R PGF, R C, R LO,R incl,r, R PGF C LO, PGF PGF, incl C C, incl a LL,aLL,aLL,a LL, xc, xg, f, D, Pb f,d,pb are directly obtained from data; the rest has to be estimated/parameterised.

18 Weighting method Using a Neural Network to assign to each event a probability of originating from each of the three processes (LO, PGF or Compton). MC is used to train the Neural Network (NN). MC training NN A parametrization is constructed. A weight is built from the parametrization. Data parametrization Data is weighted in an event by event basis. Event by event calculation G/G

19 Fractions Ri We parametrise fractions R (probabilities). Two variables O1 and O2 are used (R sum up to 1). 1 R PGF =1 O1 O2 3 1 R QCDC =O1 O2 3 i Ri = Tot 2 R LO= O2 3

20 Results ΔG =0.08±0.10±0.05 G x =0.08 G 2 =3 (GeV/c) 2

21 Systematics errors Study What has been checked? False asymmetries: Neural Network stability: Several training MC samples Systematic errors due to MC: Parton shower radiation on/off and tuning Pb, Pt, f A1 parametrisation Different parametrisations were used ( G/G)false ( G/G)NN ( G/G)MC ( G/G)f,Pb,Pt ( G/G)A Total 0.045

22 High pt Analysis, Q2 <1 (GeV/c)2

23 High pt 2 2 Analysis, Q < 1 (GeV/c) For this analysis, the same selection as Q2 > 1 (GeV/c)2 analysis was applied plus a slightly tighter set of cuts : xf>0.1 and z >0.1 xbj< 0.05 pt2 > 2.5 GeV2 10% of statistics ~90 % of our statistics in this sample

24 Processes Contribution for high pt, Q2<1 (GeV/c)2 PGF and Background events: MC Generator PYTHIA Same background as Q2 > 1 (GeV/c)2 case PGF QCD-C LO DIS Additional background from resolved photon events qq qq direct resolv. γ qg qg gg gg low pt Additional processes sensitive to gluons in the nucleon neglect LO DIS and low pt

25 Data Monte Carlo comparison, Q2<1 (GeV/c)2 dn/dpt,1st hadron dn/dq2.data MC dn/dpt, 2nd hadron.data.data MC MC pt The agreement between MC and data is good

26 Results, Q2<1 (GeV/c)2 Determination of ΔG/G : ALL/D = ± 0.019stat. ± 0.003syst x g =0.085 G = ± 0.089stat. ± 0.014exp. syst. ± MC syst. ± 0.018photon < > = (GeV/c)2 G G = ± 0.089stat. ± x g = G/G = ± (stat) ± 0.014(exp syst) ± 0.052(MC syst) ± 0.013( ) G G/G = ± 0.089(stat) ± 0.057(syst) 0.055(syst) Phys. Lett. B 633 (2006) 25-32

27 G/G Results Q2>1 GeV2 : G/G = 0.08 ± 0.10stat ± 0.05sys xg = Q2<1 GeV2 : G/G = 0.02 ± 0.06stat ± xg = independent analyses with quite different backgrounds lead to compatible results

28 G/G Results Q2>1 GeV2 : G/G = 0.08 ± 0.10stat ± 0.05sys xg = Q2<1 GeV2 : G/G = 0.02 ± 0.06stat ± xg = independent analyses with quite different backgrounds lead to compatible results

29 Conclusions and Outlook Recent results on G/G from COMPASS high pt analysis have been presented These measurements are consistent with xg and 2007 data to be analyzed Increase statistics for 2006 and 2007 data due to the new COMPASS magnet

30 Spares

31 Uncertainty for Q2 <1 GeV2 Contribution from resolved photons Problem: polarised PDFs of the photon is not measured! use unpolarised PDFs to constrain polarised q γ x, Q 2 Δq γ x, Q 2 q γ x, Q 2 This leads to 2 extreme (max & min) scenarios additional uncertainty band. Glück, Reya, Sieg, Eur. Phys. J. C20 (2001) 271