1. 1 Diffractive dijets at HERA Alice Valkárová Charles University, Prague Representing H1 and ZEUS experiments

2. 2 HERA collider experiments 27.5 GeV electrons/positrons on 920 GeV protons → √s=318 GeV two experiments: H1 and ZEUS HERA I: 16 pb -1 e-p, 120 pb –1 e+p HERA II: ∼ 550 pb -1, ∼ 40% polarisation of e+,e- e DIS: Probe structure of proton → F 2 Diffractive DIS: Probe structure of color singlet exchange → F 2 D

3. 3 W ´ HERA: ~ 10% of low-x DIS events are diffractive Diffraction and diffraction kinematics momentum fraction of color single exchange fraction of exchange momentum, coupling to γ* t 4-momentum transfer squared MyMy

4. 4 1) Proton Spectrometers: ZEUS: LPS (1993-2000) H1: FPS (1995-2008),VFPS (2004-08) t measurement access to high x IP range free of p-dissociation background at low x IP small acceptance  low statistics ☠ 2) Large Rapidity Gap, H1, ZEUS: Require no activity beyond η max t not measured, some p-diss background ☠ 3) M x method, ZEUS: Diffractive vs non-diffractive: exponential fall off vs constant distribution in ln M x 2 Some p-diss contribution ☠ diff. non-diff Diffractive Event Selection e p

5. 5 QCD factorization Get PDF from inclusive diffraction predict cross sections for exclusive diffraction Hard scattering QCD matrix element,perturbatively calculated, process dependent Universal diffractive parton densities, identical for all processes → DPDFs – obey DGLAP, universal for diff. ep DIS (inclusive,dijet,charm) → universal hard scattering cross section (same as in inclusive DIS) proven for DIS (J.Collins (1998)) not proven for photoproduction!

6. 6 Proton vertex (Regge) factorization Additional assumption – there is no proof! pomeron flux factorpomeron PDF Exctracted from inclusive diffraction! Fits 2006 A and B

7. 7 H1 2006 fits -results Two fits, A and B Fit B: gluon parameterised as a constant at starting scale quarks are very stable gluons carries ~75% of pomeron momentum gluons similar at low z no sensitivity to gluon at high z z (z IP )- longitudinal momentum fraction of gluon rel. to colorless exchange

8. 8 Dijets in diffractive DIS, H1 z IP is the most sensitive variable to test gluonic part of DPDFs – difference between fits A and B at high z IP. Data agree with NLO prediction, NLO with fit B is more close to data 4< Q 2 < 80 GeV 2 0.1 < y < 0.7 x IIP < 0.03 P* tjet1 > 5.5 GeV P* tjet2 > 4 GeV -3. < η* jets < 0. Data 99/00

9. 9 Dijets in diffractive DIS, H1 For z IP < 0.4 NLO predictions using fits 2006 A and B agree with data very well Combined QCD fit for inclusive and dijet DIS data…..

10. 10 H1 QCD jet fit Low sensitivity of fits to inclusive cross section to gluon PDF especially at large z IP → use jets to combined fits! Largest difference Largest difference JHEP 0710:042,2007

11. 11 Data 99/00 Eur.Phys.J.C52: 83 (2007) Dijets in diffractive DIS, ZEUS E* tjet1 > 5 GeV The best agreement of data and NLO for H1 2006 fit B and MRW 2006 H1, ZEUS: for DIS dijets factorization holds….. R=data/NLO(ZEUS LPS)

12. 12 pp  γ*p Exporting DPDFs from HERA to Tevatron….. x IP integrated effective DPDFs from CDF single diff. dijets (run I) Factorization broken by β-dependent factor ~ 10 ! If for HERA incl. data rapidity gap survival factor =1, at Tevatron ~0.1 Hadron-hadron collisions

13. 13 Photoproduction, γ ⋆ p, Q 2 →0 direct photoproduction (Q 2 ≃ 0): photon directly involved in hard scattering x γ =1 Expect: gap survival probablity= 1 resolved photoproduction (Q 2 ≃ 0): photon fluctuates into hadronic system, which takes part in hadronic scattering,dominant at Q 2 ≃ 0 x γ <1 Expect: gap survival probability < 1 Theory (Kaidalov at al) ~0.34 x γ - fraction of photon’s momentum in hard subprocess

14. 14 H1: E tjet1 > 5 GeV suppression of factor ~0.5 ZEUS: E tjet1 > 7.5 GeV weak (if any) suppression (0.6-0.9) Neither collaboration sees difference between the resolved and direct regions, in contrast to theory! Possible explanation of differences between H1 and ZEUS (DIS 2007) Different phase space of both analyses ……..? One year ago….

15. 15 E t dependence of suppression? H1ZEUS H1 and ZEUS observe the data have harder E t slope than NLO From the DIS 2008 talk of W.Slomiński, ZEUS results

16. 16 Double ratios of γ ⋆ p & DIS ( data/NLO) Very useful – full or partial cancellation of many uncertainties (energy scales for data, DPDFs used…etc ). E tjet1 >5 GeV There is no clear W dependence but what about E tjet1 dependence?? H1 collab. Eur.Phys.J,C51 (2007),549

17. 17 H1 - ratio photoproduction/DIS Double ratio of Data/NLO for photoproduction and DIS → uncertainties of diffractive PDFs cancel! Figure derived from published results Double ratio is within errors E t dependent!

18. 18 New H1 analysis – data 99/00 Tagged photoproduction, luminosity 3x larger than for 97 diffractive events found by Large Rapidity Gap method (LRG)

19. 19 Two cut scenarios To crosscheck previous H1 results To approach closest to ZEUS cuts

20. 20 Theoretical predictions 2 programs for NLO calculations, 3 sets of DPDFs: Frixione/Ridolfi → H1 2006 Fit A H1 2006 Fit B H1 2007 Fit Jets → Kramer/Klasen → H1 2006 Fit B The aim was to estimate the effect on DPDF used and to crosscheck NLO programs inclusive measurement DIS dijets

21. 21 Integrated survival probabilities (ISP) No difference in survival probabilities for resolved and direct regions of x γ, like in previous H1 and ZEUS analyses Within errors no difference in ISP using different DPDFs Lower E t cut scenario

22. 22 Lower E t cut scenario Another hint of E t harder slope for data than NLO Hadronization corrections δ hadr =MC(hadr)/MC(parton) ZEUS – W.Slominski

23. 23 Higher E t cut scenario Now much more „direct-like“ events than in low E t analysis, peak at higher x γ Integrated survival probabilities (ISP) Larger ISP than for lower E t cut scenario → more close to ZEUS results!!!

24. 24 Higher E t cut scenario E t dependence not excluded but cannot be independently verified ZEUS – W.Slomiński

25. 25 What DPDF is the „best“? The most sensitive variable → z IP Fits 2006 A,B valid only to z IP = 0.8 Fit 2007 jets to z IP = 0.9 Fits 2006 A and 2007 Jets represent extremes. Fit 2006 B is in the middle….

26. 26 Summary new H1 dijet photoproduction data – older H1 results confirmed – within errors is gap survival probability (GSP) ~ 0.5. in higher E t cut scenario (similar to ZEUS) GSP is ~ 0.6, more close to ZEUS results. hint that GSP is dependent on E t of the leading jet, for low E t jets seems to be suppression more significant. the evidence that GSP is not different for direct and resolved events remains (originally not expected ) to understand better diffractive dijets in photoproduction is challenging!!!