Real and virtual photon structure Leif Jönsson University of Lund representing the H1 and ZEUS collaborations

1. Real and virtual photon structureLeif JönssonUniversity of Lundrepresentingthe H1 and ZEUS collaborations Outline of the talk Physics processes; di-jet events Virtual photon structure Real photon structure Di-jet events with charm production Conclusions

2. Physics Processes • Electron-proton scattering proceeds via the exchange of a virtual photon • Photoproduction Q2~0 GeV2 • Deep inelastic scattering Q2>>0 GeV2 • Pointlike photons (direct) Q2>kT2 • Resolved photons Q2<kT2 • Direct processes with kT non-ordered parton emissions (CCFM) DGLAP evol. CCFM evol. sepeX = dy fg/e(y,Q2) sgp Y: electron momentum fraction taken by the photon fgT/e dominating; fgL/e contributes as y gets small Direct: sgp = i dxp fi/p(xp, mp) sig Resolved: sgp = ij dxg fj/p(xg, mg) dxp fi/p(xp, mp) sij xg: the fractional photon momentum entering the hard scattering xp: the fractional proton momentum taken by the interacting parton Leif Jönsson

3. xgobs = jetsETe-/2yEe A cut at xg around 0.7-0.8 gives good separation between direct and resolved processes Leif Jönsson

4. Virtual photon structureTriple differential cross sections • Direct processes only describe data in the region Q2>(ET)2 • In the region Q2<(ET)2 the resolved processes become important Leif Jönsson

5. Virtual photon structureIncluding longitudinal photon polarisation • Recently QCD parametrisation of longitudinally • polarised photons has been implemented in Herwig • Herwig comes much closer to data Leif Jönsson

6. Virtual photon structureComparisons with Cascade (CCFM) • Cascade provides kT non-ordered parton showers • Cascade with less degrees of freedom (no photon structure) describes data reasonably well Leif Jönsson

7. Structure of real photons ds/dxgOBS compared to NLO calculations • NLO calculations give reasonable description of data • Only slight dependence on photon PDF’s Leif Jönsson

8. Structure of real photonsds/dcos* compared to NLO calculations Mjj>42 GeV H1 xg< 0.75 xg> 0.75 • cos*=|tanh(h1-h2)/2 • NLO calculations give reasonable agreement with data Leif Jönsson

9. Cross section ratio of resolved and direct processes as a function of Q2 • The cross section ratio decreases with increasing Q2 as the contribution from resolved processes gets less important • SaS1D falls below the data Leif Jönsson

10. Q2 dependence in charm productionR=sgobs<0.75)/sgobs>0.75) vs Q2 • Data are not able to distinguish between Q2 suppression or not • Cascade (CCFM) gives good description of data • Aroma (DGLAP) falls below • Extrapolation to the full D* phase space confirms no Q2 suppression • Two different scales come into play Leif Jönsson

11. ds/dxgOBS vs xgOBS in charm productionPredictions by Cascade (CCFM) • In a significant fraction of the events the gluon is the hardest parton • Cascade on hadron level gives reasonable agreement with data Leif Jönsson

12. Conclusions Virtual photon structure • Direct photon processes only describes data in the kinematic region Q2>ET2 • The inclusion of resolved photon processes provides better agreement in the region Q2<ET2 • Considering also longitudinally polarized photons improves the agreement even more • CASCADE with kT non-ordered parton emissions (CCFM) gives similar agreement over the full kinematic range; do we need resolved photons? Real photons • NLO calculations reproduce data reasonably well • The dependence on photon PDF’s seems small • The dominant error comes from NLO scale uncertainties Charm production • No Q2 suppression observed in contrast to the case where no charm requirement is made • Suppression due to charm and Q2 not independent Leif Jönsson