Photon Structure from HERA

1. Photon Structure from HERA Steve Maxfield University of Liverpool on behalf of the H1 and ZEUS Collaborations Stephen Maxfield DIS 2002, Krakow April/May 2002

2. Photon Structure from ep collisions Proton beam as source of partons probing photon structure • Dijet production in photoproduction and low Q2 DIS supplies: • Constraints on g pdf’s • Test of perturbative QCD (NLO) • Complements eg measurements (direct sensitivity to gluon) …because we have resolved photon processes: pQCD (NLO) QED Constrained elsewhere • Parton from proton probes photon with resolution determined by Et of jets Stephen Maxfield DIS 2002, Krakow April/May 2002

3. Photon Structure: Outline • Summarise status of photon structure measurements at HERA • Real photon: • Photoproduction of high Et dijets. Measurements from H1 and ZEUSsee talks by A. Lupi(ZEUS) and G.Frising(H1) • Virtual Photons: • Dijet cross-sections in Low Q2 DIS events. Measurements from H1 and ZEUS. see talks by K.Sedlak (H1) , M. Lightwood(ZEUS) • Not covered: - single-inclusive jet cross-section measurement see talk by J. Turnau-other important things I’ve forgotten For High Et jet production see talk byC.Foudas (ZEUS+H1) Stephen Maxfield DIS 2002, Krakow April/May 2002

4. Real Photon Structure: Photoproduction Measurements of dijet cross-sections in photoproduction (ZEUS) Eur. Phys. J. C (2002) • Comparison with NLO calculations • (Frixione and Ridolfi) • Hadronisation corrections estimated using HERWIG 6.1 / PYTHIA 6.1 • Parton densities: Photon GRV-HO and AFG-HO • Proton CTEQ5M1 • Integrated Luminosity 38.6 pb-1 (1996/97) • Q2 < 1 GeV2 • 134 GeV < Wgp < 277 GeV •  2 jets (kt algorithm) • -1 < h < 2.4; Etjet1 > 14 GeV Etjet2 > 11 GeV …similarly from H1  Stephen Maxfield DIS 2002, Krakow April/May 2002

5. Real Photon Structure Measurements of dijet cross-sections in photoproduction (H1) DESY 01-225 submitted to Eur. Phys. J. C • Comparison with NLO calculations • (Frixione and Ridolfi) • Hadronisation corrections estimated using HERWIG 5.9 / PYTHIA 5.7 • Parton densities: Photon GRV-HO and AFG-HO • Proton CTEQ5M • Integrated Luminosity 34.9 pb-1 (1995-1997) • Q2 < 1 GeV2 • 0.1 < y < 0.9 •  2 jets (kt algorithm) • -0.5 < h < 2.5; Etjet1 > 25 GeV Etjet2 > 15 GeV Note phase space and models slightly different from that of ZEUS Stephen Maxfield DIS 2002, Krakow April/May 2002

6. Real Photon Structure: NLO Matrix elements • Test the NLO matrix elements of the hard scattering process by looking at: • distribution of scattering angle in centre-of-mass frame, cosq*=tanh(Dh/2) • Minimum Mjj required to ensure distribution not biased by Et cuts Different propagators  different cosq*distributions Distinguish using: direct resolved Stephen Maxfield DIS 2002, Krakow April/May 2002

7. Testing the NLO Matrix element Mjj > 42 GeV • NLO prediction agrees with data for xgobs> 0.75 • underestimates resolved-like processes by (10-15)% • Reasonable description within errors† for resolved events with GRV-HO pdf • Shapes consistent with difference in dominant propagators †data 5-10% theory 5-10% Stephen Maxfield DIS 2002, Krakow April/May 2002

8. H1: GRV-HO • Mjj cut needed to reveal dynamics • NLO can describe data well for low xg • Slight overestimates for high xg • Expected steeper behaviour in resolved processes  dynamics understood What about the pdf’s? renormalisation/factorisation scale uncertainty Stephen Maxfield DIS 2002, Krakow April/May 2002

9. Real Photon Structure contd. • Sum over two highest Et jets. • In leading order, estimates xg • NLO calculations use pdf’s for real photons obtained from fits to eg scattering data • How well do these describe the data? • test of universality of pdf’s. • Can the ep data constrain the pdf’s? • Measure distribution of: Stephen Maxfield DIS 2002, Krakow April/May 2002

10. Real Photon Structure contd. • NLO predictions give a ‘reasonable’ description of the data. • Neither GRV-HO nor AFG-HO pdf’s can describe the data everywhere. • Cannot distinguish between pdf’s. • Existing pdf’s inadequate? Stephen Maxfield DIS 2002, Krakow April/May 2002

11. H1results • NLO predictions give a good description of the data. • GRV-HO or AFG-HO pdf’s can describe the data everywhere. • Cannot distinguish between pdf’s • Existing pdf’s adequate to describe the data Stephen Maxfield DIS 2002, Krakow April/May 2002

12. Etjet2 Sensitivity of NLO calculations to jet Et cuts • NLO calculations IR sensitive unless min Et cuts on jets are asymmetric.  NLO prediction depends on min(Etjet2) cut. How strongly? • Significant dependence seen.Choice of Etjet2 cutchanges prediction • Note that HERWIG (LO + LL parton showers) gives an excellent description of this behaviour!  Resummed or NNLO calculations needed Stephen Maxfield DIS 2002, Krakow April/May 2002

13. Real Photon Structure Summary • NLO calculations give a reasonable description of the data • Taking into account systematic errors and differences in the phase space, the H1 and ZEUS results are not inconsistent. • Both experiments agree that data shows sensitivity to g pdfs’s (but do not allow discrimination between the pdf’s obtained from eg.) • …and that the data should be included in future global fits. • NNLO, Resummed calculations highly desirable to improve the constraints. Stephen Maxfield DIS 2002, Krakow April/May 2002

14. Virtual Photon Structure + + • ‘Target’ photon with finite Q2 : Low Q2 DIS events • Introduces a second hard scale. Now have Et2, Q2 • For Q2 large enough, non-perturbative contributions to the g* structure (VDM) strongly suppressed (~1/Q4) w.r.t to calculable perturbative terms. • g*pdf’snot strictly required. • But useful as includes resummation of near collinear terms • Provides framework for a smooth transition from photoproduction to DIS Stephen Maxfield DIS 2002, Krakow April/May 2002

15. Virtual Photon Structure • Triple differential cross-sections measured by H1 and ZEUS • Cross sections compared with LO QCD • + pdf’s for resolved photon including suppression of resolved component with increasing Q2 (SaS) Stephen Maxfield DIS 2002, Krakow April/May 2002

16. ZEUS • HERWIG5.9 • proton pdf MRSA • photon pdf SaS1D • no s.u.e • Normalised to highest xg bin • Direct component sufficient where Q2 > Et2 • Resolved virtual photon component needed whereEt2 > Q2 • See expected evolution of resolved component Stephen Maxfield DIS 2002, Krakow April/May 2002

17. ZEUS • In SaS 1D model: • Resolved component everywhere dominated by perturbative component Stephen Maxfield DIS 2002, Krakow April/May 2002

18. Q2 suppression modified in charm enriched events… • 103.7 pb-1 (1996-2000) • Etjet1 > 7.5 GeV Etjet2 > 6.5 GeV • |hjet| < 2.4 • 0.2 < y < 0.65 • Tag charm with D*+→ K-p+p+s +c.c. • …and require D* to have • pT > 3 GeV |h| < 1.5 + Look at ratio of resolved to direct events: Q2 evolution modified • Data consistent with flat in Q2 • described by LO MC with SaS1D pdf’s for g*. Stephen Maxfield DIS 2002, Krakow April/May 2002

19.  not all due to restricted D* phase space Interplay of three hard scales: Et2, Q2, mc2 Stephen Maxfield DIS 2002, Krakow April/May 2002

20. H1 • HERWIG 5.9 • RAPGAP 2.8 • proton pdf GRV LO • photon pdf SaS1D • 10% s.u.e • Both LO predictions underestimate the cross section • normalisation uncertain What else? Stephen Maxfield DIS 2002, Krakow April/May 2002

21. Longitudinally polarised photons g*L • Lepton beam acts as source of transverse and longitudinal photons: • Fluxes comparable for small y: • Gauge invariance ensures g*L contributions to cross-sections vanish as Q2→0 • Explore sensitivity to g*L by measuring y-dependence of cross-section Stephen Maxfield DIS 2002, Krakow April/May 2002

22. PDF’s of g*L Notation: M2 = hard scale (e.g. Et2) P2 = photon virtuality • QED (vanishes for P2→0) • QCD improved PDF’s: • J. Chýla Phys. Lett. B488,289, hep-ph/0006232 • g*L contribution significant • exhibits hadron-like scaling behaviour. • cf g*T ~ ln M2/P2 •  g*Lsignificance increases as P2 → M2 Stephen Maxfield DIS 2002, Krakow April/May 2002

23. g*L • Version of HERWIG5.9 modified to include g*L pdf • Inclusion ofg*Limproves description Stephen Maxfield DIS 2002, Krakow April/May 2002

24. y-dependence of cross-section • Shape of y-distribution not described by gT* pdf’s alone Stephen Maxfield DIS 2002, Krakow April/May 2002

25. g*L • Description of y dependence also improved Stephen Maxfield DIS 2002, Krakow April/May 2002

26. Direct Photon DGLAP Other possibilities:Unordered emissions: CASCADE Resolved Photon ~unordered emission CASCADE: Event generator based on CCFM H.Jung and G.P.Salam, Eur. Phys. J. C19 (2001), 351-360 Stephen Maxfield DIS 2002, Krakow April/May 2002

27. CASCADE •  • CASCADE also gives better description of the data in general But… Q2 dependence not well-reproduced Stephen Maxfield DIS 2002, Krakow April/May 2002

28. CASCADE xgdependence similar Stephen Maxfield DIS 2002, Krakow April/May 2002

29. Next Steps: • Would really like to see NLO predictions compared with the data. • H1 and ZEUS both working on this. • Expect that the ‘share’ of the cross-section carried by g pdf’s will be reduced. • Extraction of gL? Stephen Maxfield DIS 2002, Krakow April/May 2002

30. Conclusions • Dijet cross-section measurements in photoproduction of sufficient precision to constrain pdf’s of real photons. • Constraints would be tighter if NNLO/resummed calculations were available. • Virtual photon structure shows sensitivity to longitudinal polarisation of photon. • Potential to measure pdf of gL* • Comparison of data with NLO predictions including virtual photon structure underway Stephen Maxfield DIS 2002, Krakow April/May 2002