 Introduction  The ZEUS PDF fit: an overview  Impact of future HERA data on the ZEUS fit - end of current HERA-II running scenario - additional studies (F L, sea quark asymmetry)  Conclusions Impact of HERA-II data on the ZEUS PDF fits 13 th International Workshop on Deep Inelastic Scattering (DIS05) Madison, Wisconsin, USA 27 th April – 1 st May 2005 Claire Gwenlan with the help of M. Klein (DESY), C. Targett-Adams (UCL), R. Thorne (Cambridge), A. Tricoli (Oxford)

2 Besides being interesting in their own right, it is essential to know the parton density functions (PDFs) of the proton as precisely as possible in order to maximise the physics potential at both current and future colliders e.g. high-x gluon is dominating uncertainty in several LHC processes Introduction _______________________________ Presented here are the results of studies that give a first look at the potential impact of future HERA measurements on the proton PDFs: 1.within current HERA-II running scenario - increased luminosity - cross sections optimised for sensitivity to PDFs 2.Other possibilities (F L, sea quark asymmetry) HERA data are now very precise and cover a wide range in (x,Q 2 )  determination of proton PDFs now possible within one experiment Most recently, ZEUS have performed a NLO QCD analyses on their full set of HERA-I e + and e - structure function data and high precision jet data  the ZEUS-JETS PDF With future measurements at HERA, hope to be able to do even better …

3 Observables used in QCD fits to determine PDFs:  Inclusive NC/CC DIS ep cross sections   direct sensitivity to quarks  only indirect sensitivity to gluon (scaling violation)  Jet cross sections:  directly sensitive to gluon through boson-gluon-fusion  xpxp Determination of proton PDFs at HERA _______________________________ Q 2 = -q 2 = -(k-k’) 2 Factorisation: observable = short range interaction  PDFs Gluon in proton Jet , Z 0, W Now, after the HERA-I (94-00) phase of data-taking, the full set of inclusive NC/CC e + /e - data, and high precision jet data are available for QCD analysis…

ZEUS-JETS NLO QCD fit _______________________________  parameter constraints:  momentum and quark number sum rules  low-x behaviour of u v and d v set equal  flavour structure of light quark sea set consistent with Gottfried sum and Drell-Yan  heavy quarks treated in variable flavour number scheme of Thorne and Roberts PDFParam. at Q 0 2 =7 GeV 2 u-val. (xu v )A uv x b uv (1-x) c uv (1+d uv x) d-val. (xd v )A dv x b dv (1-x) c dv (1+d dv x) total sea (xS)A S x b S (1-x) c S gluon (xg)A g x b g (1-x) c g (1+d g x) dbar-ubar (x  ) A  x b  (1-x) c  For more details on the ZEUS-JETS fit (hep-ph/ ), and the data included, see cont. to this workshop, “Proton PDFs using Structure Functions and Jet Data from ZEUS”, Juan Terron. all HERA-I ZEUS incl. NC/CC e + /e - (94-00) ZEUS inclusive jets in DIS (96-97) ZEUS dijets in photoproduction (96-97)  11 free parameters in total Data: Parameterisation: 4

5 Current HERA-II running scenario _______________________________  HERA-II is running efficiently... “ 700pb -1 integrated luminosity, equally divided between e + /e -, expected by the end of HERA-II running in mid-2007 ” We are here… Current running scenario HERA-I (92-00) HERA-II (02  ) e+e e-e- 27> 50 HERA delivered luminosity (pb -1 )

6 ValenceHigh Q 2 inclusive NC/CC e  cross sections SeaLow-x from inclusive NC DIS High-x ? Flavour ? (assumptions needed) GluonLow-x from HERA dF 2 /dlnQ 2 Mid-to-high-x from HERA jet data High-x from momentum sum rule Where does the information come from in a HERA-Only fit ? What impact will future HERA measurements have on the PDFs? _______________________________ HERA-I: statistics limited HERA-I: statistics limited at high-E T and high-Q 2

7 Example: high-Q 2 NC and CC data _______________________________ F 2 dominates NC cross section, HERA-I:  F 2 /F 2 ~30% e+ and e- needed for flavour separation, but high-Q 2 CC statistically limited at HERA-I, especially e - p data  HERA-II will provide greatly increased luminosity

8 ValenceHigh Q 2 inclusive NC/CC e  cross sections SeaLow-x from inclusive NC DIS High-x ? Flavour ? (assumptions needed) GluonLow-x from HERA dF 2 /dlnQ 2 Mid-to-high-x from HERA jet data High-x from momentum sum rule Where does the information come from in a HERA-Only fit ? What impact will future HERA measurements have on the PDFs? _______________________________ HERA-I: statistics limited HERA-I: statistics limited at high-E T and high-Q 2

9 ValenceHigh Q 2 inclusive NC/CC e  cross sections SeaLow-x from inclusive NC DIS High-x ? Flavour ? (assumptions needed) GluonLow-x from HERA dF 2 /dlnQ 2 Mid-to-high-x from HERA jet data High-x from momentum sum rule Where does the information come from in a HERA-Only fit ? What impact will future HERA measurements have on the PDFs? _______________________________ HERA-I: measurements in only certain kinematic regions – potential to optimise cuts for sensitivity to gluon

10  Measure jet cross sections in kinematic regions “optimised” for sensitivity to gluon - ongoing ZEUS study: dijets in photoproduction (Q 2 < 1 GeV 2 )  data simulated using NLO QCD (Frixione-Ridolfi) and CTEQ5M1 proton PDF (500 pb -1 ) Optimised jet cross sections _______________________________ Christopher Targett-Adams (UCL) 10

11 Impact of HERA-II in current running scenario:- case study _______________________________ Data sampleL of HERA-I measurement (pb -1 ) assumed L of HERA-II measurement (pb -1 ) Central values taken from… Systematic uncertainties taken from… High-Q 2 NC e+63700existing data High-Q 2 NC e-16700existing data High-Q 2 CC e+61700existing data High-Q 2 CC e-16700existing data Inclusive DIS jets37500existing data Dijets in  p37500existing data Optimised dijets in  p-500NLO QCDNOT INCLUDED data exists for HERA-I running data does not yet exist for HERA-I running Impact of the projected HERA-II measurements has been studied in the context of the ZEUS-JETS fit

12 HERA-II projected fit _______________________________  The HERA-II projected PDF compared to the ZEUS-JETS fit  Impact from increased statistics on high-Q 2 NC/CC e+/e- data Impact from optimised cross sections (mostly) and increased statistics on jet data

13 u-valence uncertainties _______________________________ log-x scale (low-x region)linear-x scale (high-x region)  uncertainties on u-valence distribution significantly reduced over visible x range

14 d-valence uncertainties _______________________________ log-x scale (low-x region)linear-x scale (high-x region)  uncertainties on d-valence distribution significantly reduced over visible x range

15 log-x scale (low-x region)linear-x scale (high-x region)  uncertainties on sea-quark distribution significantly reduced at high-x  most significant improvement from increased statistics Sea-quark uncertainties _______________________________

16 gluon uncertainties _______________________________ log-x scale (low-x region)  uncertainties on mid-to-high-x gluon significantly reduced  most significant improvement comes from optimised cross sections linear-x scale (high-x region)

17 HERA Kinematic (x,Q 2 ) Range _______________________________  HERA data covers large region in (x,Q 2 )  also relevant x-region for LHC  HERA-II projected PDFs should have significant impact on e.g. high-p T jets, heavy particle searches etc.  sensitive to improvement to high-x partons  LHC cross sections under study

18 Impact of a HERA measurement of F L _______________________________ F L contributes at NLO (and HO) and is directly sensitive to the gluon density in the proton  measured at fixed target exps. (x > )  precision F L measurement at HERA requires low-E p running  vary y at fixed (x,Q 2 )

19  precision measurement of F L at HERA-II:  pin down gluon density at low-x  reduce uncertainties on gluon PDF Impact of a HERA measurement of F L _______________________________ F L contributes at NLO (and HO) and is directly sensitive to the gluon density in the proton

20  precision measurement of F L at HERA-II:  pin down gluon density at low-x  reduce uncertainties on gluon PDF  provide tests of higher order QCD  test the need for extensions to DGLAP at low-x Impact of a HERA measurement of F L _______________________________ F L contributes at NLO (and HO) and is directly sensitive to the gluon density in the proton Robert Thorne 20

21 Simulation of HERA-II F L : Max Klein (DESY)  F L simulated using GRV94 NLO PDF  statistical uncertainties correspond to:  systematic uncertainties from current H1 analysis of data (few %)  precision measurement of F L at HERA-II:  pin down gluon density at low-x  reduce uncertainties on gluon PDF  provide tests of higher order QCD  test the need for extensions to DGLAP at low-x Impact of a HERA measurement of F L _______________________________ F L contributes at NLO (and HO) and is directly sensitive to the gluon density in the proton E p (GeV) L (pb -1 )10532 For further details, see “On the future measurement of F L at low-x at HERA”, Max Klein, in proceedings, DIS04. 21

22 Impact on gluon distribution _______________________________  Fit including simulated F L data compared to the ZEUS-JETS PDF   Gluon uncertainties reduced at low-x and low-Q 2 (not relevant for LHC) 22

23  F L predictions very sensitive to underlying theory  choice of PDF, order of QCD calculation etc.  How sensitive is the NLO QCD fit to inclusion of “extreme” sets of simulated F L data Sensitivity of the NLO QCD fit _______________________________ PDFQCD theory Max. F L MRSG95NNLO* Mid. F L GRV94NLO Min. F L MRST2003NLO  ZEUS fit relatively stable to inclusion of extreme F L data-sets  fit cannot describe “just any data”  an F L measurement of this precision should have power to discriminate between theoretical models Simulated F L data extremes provided by Robert Thorne (Cambridge) 23

24 Summary _______________________________  HERA-I has provided precision measurements over large region in (x,Q 2 )  PDFs of the proton already determined to high precision using HERA data only  most recently: ZEUS-JETS fit to HERA-I inclusive NC/CC and jet data  In current running scenario:  HERA-II will provide significantly increased statistical precision - high-Q 2 NC/CC data  significant improvement to valence distributions - high-Q 2 and high-E T jet data  improvement to high-x gluon  potential for measuring cross sections optimised for sensitivity to gluon - significant improvement to high-x gluon  In low-E p running scenario a precise measurement of F L at low-x possible  inclusion in fit indicates improvement in gluon uncertainties at low-x and low-Q 2  stability of fit to extreme F L sets indicates ability of HERA F L to discriminate between theory

25 Bonus Extra: - sea quark asymmetry, a study by Max Klein, Burkard Reisert _______________________________ ū=đ was a natural assumption for long time, until E866, HERMES found a difference at x ~ 0.1  all global fits followed  Indications for strange-anti-strange asymmetry What causes rise of F 2 at low x? Have measured 4ū+đ but ū and đ are unknown at low x  would be accessible via deuteron (eD) running at HERA-II Are sea and anti-quarks equal ? Are up and down quarks equal at low x?

26  No information on sea flavour composition - only one measurement at low-x:  Assume quark and anti-quark distributions are equal at low-x and u=d The H1 NLO QCD fit _______________________________ PDFParam. at Q 0 2 =4 GeV 2 xU=x(u+c)A U x b U (1-x) c U (1+d U x+e U x 3 ) xUbarA Ubar x b Ubar (1-x) c Ubar xD=x(d+s)A D x b D (1-x) c D (1+d D x) xDbarA Dbar x b Dbar (1-x) c Dbar xgA g x b g (1-x) c g (1+d g x) HERA-I H1 incl. NC/CC e + /e - (94-00): H1-Only + BCDMS (p,D data): H1+BCDMS experimental errors model uncertainties

27 Releasing the dbar-ubar constraint _______________________________ H1-Only fit (default assumptions) experimental errors model uncertainties H1-Only PDF experimental errors model uncertainties H1-Only PDF 27

28 Releasing the dbar-ubar constraint _______________________________ H1+BCDMS fit experimental errors model uncertainties H1-Only PDF experimental errors model uncertainties H1-Only PDF 28

29 simulated accuracy (20pb -1 e-D, 40pb -1 e-p) Q 2 =5 GeV 2 The light sea quark asymmetry is expected and has been assumed to vanish at low x. However, F 2 rises strongly towards low x which deserves to be studied. Releasing the dbar-ubar constraint _______________________________ H1+BCDMS fit

30 simulated accuracy (20pb -1 e-D, 40pb -1 e-p) Q 2 =5 GeV 2 The light sea quark asymmetry is expected and has been assumed to vanish at low x. However, F 2 rises strongly towards low x which deserves to be studied. Releasing the dbar-ubar constraint _______________________________ H1+BCDMS fit

31 Mini-summary (sea-quark asymmtery) _______________________________  So far HERA has not resolved the light sea quarks at low x  QCD fits employ (“reasonable”) assumption that ū=đ and u=ū and d=đ at low-x - this reasonable assumption was proven to be wrong at larger x ~ 0.1  without these requirements the fits become unstable  deuteron data from fixed target experiments (e.g. BCDMS) do help but cannot solve the problem since data lie at higher-x  Would need eD running at HERA-II to resolve this issue

32 Extras … _______________________________

33  Uncertainties with full HERA-II inclusive data-set comparable to global fits HERA-II Projected PDF Comparison with global fit (u-val.) _______________________________ ZEUS-S global PDF 33

34 Comparison with global fit (d-val.) _______________________________  Uncertainties comparable to or better than current global fit HERA-II Projected PDFZEUS-S global PDF 34

35 Impact on sea/gluon uncertainties _______________________________  already at HERA-I, measurement of optimised jet cross sections would have a significant impact on the high-x gluon