Comparison of PDFs: CTEQ6.5 vs MSTW08 vs ZEUS2005 vs H12003

1. PDF4LHC: Comparison studiesFeb 2008 A M Cooper-Sarkar, Oxford • First change over time CTEQ6.1 to CTEQ6.5 • MRST01 to MSTW08 • Then CTEQ6.5 to MSTW08 to ZEUS2005 to H12003 • All with asymmetric errors rather than • All with MRTS04 as a common comparison line

2. CTEQ6.5 vs CTEQ6.1 Valence very similar light sea quarks larger (massive heavy quark treatment) gluon slightly different shape/larger errors As Q2 increases these trends persist but are less dramatic

3. Valence, Total Sea and gluon CTEQ6.1 CTEQ6.5 Valence very similar, sea quarks larger (heavy quark treatment), gluon?

4. Valence, Total Sea and gluon CTEQ6.5 CTEQ6.1 Valence very similar, quarks larger (heavy quark treatment), gluon?

5. Valence, Total Sea and gluon CTEQ6.5 CTEQ6.1 AS Q2 increases these trends persist but are less dramatic

6. Valence, Total Sea and gluon CTEQ6.5 CTEQ6.1 AS Q2 increases these trends persist but are less dramatic

7. Light quarks u, d, ubar dbar CTEQ6.1 CTEQ6.5 Light sea quarks larger (heavy quark treatment)

8. u, d, ubar dbar CTEQ6.5 CTEQ6.1 AS Q2 increases these trends persist but are less dramatic

9. also look at u/d and dbar-ubar CTEQ6.1 CTEQ6.5 These have slightly different shapes Don’t change much with Q2

10. strange, charm, beauty CTEQ6.5 CTEQ6.1 For alternative strange PDFS see CTEQ6S or6.6 For alternative charm see CTEQ6C Strange a bit larger as for other light quarks Massive c and b treatment now

11. MSTW08 vs MRST01 MRST04 was not much different from MRST01 but MSTW08 has Significantly asymmetric errors a different d-valence shape, smaller low-x sea very much larger low-x gluon errors Which feed to larger low-x sea errors as Q2 increases More detailed treatment of strange sector

12. Valence, Total Sea and gluon MRST01 MSTW08 different d-valence shape, smaller low-x sea and very much larger low-x gluon errors

13. Valence, Total Sea and gluon MRST01 MSTW08 Let’s take a look at the gluon at slightly larger Q2…..it’s not as –ve as it used to be

14. Valence, Total Sea and gluon MSTW08 MRST01 And you can see the large low-x gluon errors feeding into large low-x sea errors

15. Valence, Total Sea and gluon MRST01 MSTW08

16. Valence, Total Sea and gluon MRST01 MSTW08 At large Q2 the difference in low-x sea quark and gluon shapes is well within errors, but the larger uncertainty estimates remain

17. Light quarks u, d, ubar dbar MRST01 MSTW08 Light sea quarks slightly smaller, but within errors Large low-x uncertainties

18. Light quarks u, d, ubar dbar MRST01 MSTW08 Same trends at higher Q2

19. also look at u/d and dbar-ubar MRST01 MSTW08 Similar shapes, slight differences at highest-x

20. strange, charm, beauty MSTW08 MRST01 Charm and beauty not very different though slightly less beauty and slightly more charm Strange treatment allows for a shape difference from other light quarks- evident at high-x, and for s – sbar non zero

21. ZEUS05 vs H103 • Both use restricted data sets • H1PDF2000 actually published 2003 H1+BCDMS • ZEUS only including JET data published 2005 • Most obvious difference is in error estimates: ZEUS Offset/ H1 Hessian with Δχ2=1 H1 use of restricted valence parametrization also contributes to this. • Low-x gluon shape is also rather different • Shape differences wash out as Q2 increases • Both have high-x gluons which are soft compared to MRST04

22. Valence, Total Sea and gluon H103 ZEUS05 Most obvious difference is in error estimates: ZEUS Offset/ H1 Hessian with Δχ2=1 H1 use of restricted valence parametrization also contributes to this. gluon shape is also rather different

23. Valence, Total Sea and gluon H103 ZEUS05 gluon shape is rather different H1 Sea quarks are somewhat smaller at x 10-2 to 10-1

24. Valence, Total Sea and gluon ZEUS05 H103 Even by Q2=10 the difference in gluon shape is not so dramatic H1 Sea quarks are somewhat smaller at x 10-2 to 10-1

25. Valence, Total Sea and gluon H103 ZEUS05 Differences wash out at high Q2

26. Light quarks u, d, ubar dbar H103 ZEUS05 Tendency for ZEUS sea quarks to be larger at mid-x

27. H103 ZEUS05 Differences wash out as Q2 increases

28. also look at u/d and dbar-ubar ZEUS05 H103 Different high-x shape and errors for d/u Different uncertainty estimate for dbar-ubar: ZEUS imports information from E866

29. strange, charm, beauty H103 ZEUS05 Implementation of H1 beauty in LHAPDFv5 cannot be right, was there any b-quark? Strange distributions not very different Charm not so different at Q2=100, despite massive (ZEUS) vs massless (H1) heavy quark treatment

30. CTEQ6.5 vs MSTW08 • Previously uncertainties were generally larger for CTEQ, • - but now look at low-x errors • And differently asymmetric errors • Valence: d-valence shape rather different • Sea quarks smaller, gluon more negative for MSTW

31. Valence, Total Sea and gluon CTEQ6.5 MSTW08 Sea errorslarger for CTEQ, differently asymmetric errors Valence: d-valence shape rather different Quarks smaller, gluon more negative for MSTW- but look at low-x gluon errors.

32. Valence, Total Sea and gluon CTEQ6.5 MSTW08 Sea errorslarger for CTEQ, differently asymmetric errors Valence: d-valence shape rather different Quarks smaller, gluon more negative for MSTW- but look at low-x gluon errors.

33. Valence, Total Sea and gluon CTEQ6.5 MSTW08 AS Q2 increases these trends persist but are less dramatic Note that large low-x uncertainty on gluon has fed into large low-x uncertainty of sea quarks for MSTW..much larger than for CTEQ

34. Light quarks u, d, ubar dbar CTEQ6.5 MSTW08 low-x quarks and gluon somewhat smaller for MSTW and low-x uncertainty on gluon has fed into large low-x uncertainty of sea quarks

35. Light quarks u, d, ubar dbar CTEQ6.5 MSTW08 Even at high Q2 low-x quarks and gluon somewhat smaller for MSTW and low-x uncertainty on gluon has fed into large low-x uncertainty of Sea quarks

36. also look at u/d and dbar-ubar CTEQ6.5 MSTW08 Rather different shapes for dbar-ubar, difference in high-x error estimate for d/u

37. strange, charm, beauty CTEQ6.5 Most obvious difference is in low-x error estimate Strange shape is not so different – a bit smaller at low-x for MSTW Charm/beauty also are preferentially smaller at low-x for MSTW

38. ZEUS05 vs CTEQ65 • Comparison of a fit to restricted data sets (ZEUS) to a global fit (CTEQ) • Comparison of a fit using Offset method (ZEUS)to a fit using Hessian method Δχ2=100 (CTEQ) • At Q2=1, slightly different u,d-valence shapes, • less low-x sea for ZEUS, • different gluon shapes but.. • PDFs pretty much compatible at Q2 > 10, and even uncertainty estimates are similar

39. Valence, Total Sea and gluon ZEUS05 CTEQ6.5 Slightly different u,d-valence shapes, less low-x sea for ZEUS, different gluon shapes but..

40. Valence, Total Sea and gluon Difference in sea quarks already less pronounced, some difference in gluon persists

41. Valence, Total Sea and gluon CTEQ6.5 Pretty much compatible at high Q2, even uncertainty estimates similar (and hence smaller than MSTW08 at low-x) ZEUS Offset/ CTEQ Hessian Δχ2=100

42. Light quarks u, d, ubar dbar ZEUS05 CTEQ6.5 Light quarks pretty much compatible even at Q2=10

43. Light quarks u, d, ubar dbar ZEUS05 CTEQ6.5

44. also look at u/d and dbar-ubar ZEUS05 CTEQ6.5 d/u compatible, dbar-ubar more realistic error estimate from CTEQ

45. strange, charm, beauty ZEUS05 CTEQ6.5 Heavy quarks pretty much compatible

46. No conclusions • Hard to summarize- purpose of meeting • Latest CTEQ, MSTW are significantly different from previous • All difference wash out as Q2 increases BUT • MSTW low-x uncertainties remain much larger than those of other groups - beware • PDFs from HERA alone are suprisingly compatible with global PDFS over large parts of phase space, but beware high-x gluon