1. Diffractive partons from perturbative QCD
Alan Martin, Misha Ryskin and Graeme Watt
Conventionally DDIS analyses use two levels of factorisation
- collinear factorization and Regge factorization
We replace Regge factorization by pQCD
Collinear factorization, which holds asymptotically, needs
modification in the HERA regime:
-- inhomogeneous term in DGLAP evolution
-- direct charm contribution
-- twist-4 FLD component
We present universal diffractive partons
XVIIth Rencontre de Blois, May 2005
Alan Martin (IPPP,Durham)

3. factorization proved for DDIS (Collins), but
important modifications in the HERA regime

4. H1 Large rapidity gap selection: MY<1.6 GeV and |t|<1 GeV2
H1 LPS proton selection: MY= mp
extrapolated to |t|<1 GeV2
Good agreement between two
methods and two experiments
Data well described by
H1 QCD fit to LRG data
M.Kapishin, ICHEP04, Beijing

7. x=0.18
b=0.67

8. lots of gluons at high b,
theoretically puzzling from
a perturbative viewpoint

9. H1 NLO fits to
H1 LRG data
ZEUS MX data

10. H1 NLO fits to
H1 LRG data
ZEUS MX data

11. Hint of problem with Regge factorisation assumption
assumes Pomeron ~ hadron of size R
Regge factn occurs in non-pert region m<m0, where m~1/R
but aP(0)~1.2 from DDIS > aP(0)~1.08 from soft data
 small-size component from pQCD domain with larger aP(0)
MRW study impact of applying pQCD to DDIS
find DDIS factorization OK asymptotically, but
important modifications in HERA (subasymptotic) regime

12. rapidity gap

13. given by pQCD

14. m2 fP  flux does not behave as 1/m2
convergence decreases
as xP decreases

15. inclusion of the inhomogeneous term makes gP smaller
inhomog. term
xP=0.003
Q2=15 GeV2

16. In practice, more convenient to solve standard DGLAP for each m starting
from its own scale m (provided m>Q0). Then to integrate over m. m2

17. dm2 m2 input forms: first try: assume xg ~ x -l following Wusthoff,
BEKW loosely based
parametrizations on
such forms

18. Fit to ZEUS and H1 diffractive DIS (prelim.) data
xg = x -l

19. xg = x -l gD SD ZEUS l=0.22 H1 l=0.13 combn l=0.17 H1 2002 fit
use as=0.1085
cf (PDG)
H1 have steeper
Q2 dep. of S,
hence larger g.
Also no twist-4 FL

20. g: valence-like S: Pomeron-like whereas expect lg ~ lS ~ 0.1
but one of the
HERA surprises….
global partons at
g: valence-like
S: Pomeron-like
whereas expect
lg ~ lS ~ 0.1
(xg ~ x –lg
xS ~ x –lS )

21. Now Pomeron flux factors depend on Sp as well as gp
need to introduce
Pomeron made of
col. singlet qq pair
Pomeron made of
two gluons
as well as
Now Pomeron flux factors depend on Sp as well as gp

22. dm2 m2
input forms:
use MRST/CTEQ partons

23. gD SD Fit with qq as well as gg Pomeron Uses MRST gp and Sp
to calculate
Pomeron
flux factors
v.good fit

24. diffractive charm production
constrains gluon
direct contribution
from gPom
add separately.
No evolution and
no DDIS factorization

25. the description of ZEUS diffractive charm production
from gPom
direct

26. The corresponding
MRW fit to the
ZEUS LPS data 
and H1 (prelim.) data

28. MRW 2005
ZEUS LPS, charm, MX
MRW 2005
ZEUS LPS, charm, H1
H1 fit 2002
MRW 2004
ZEUS LPS,MX, H1
MRW 20042005
parametrize bgP in
DIS scheme then
transform to MSbar

29. H1
MRW
CDF dijet data g*
rapidity gap
survival prob.

30. Conclusions
Analysis of DDIS, which goes beyond collinear + Regge factn
Regge factorization replaced by pQCD
--need quark-antiquark Pomeron, in addition to two-gluon Pomeron
--the input forms of the Pomeron PDFs given by QCD diagrams
Collinear DDIS factorization modified in HERA regime:
--inhomogeneous DGLAP evolution  smaller gPom
Good description of H1, ZEUS DDIS data
We obtain universal diffractive partons
--diffractive gluon considerably smaller than that of H1 fit
--for hadron-hadron diffraction, must include rap. gap survival probability
Moreover---the diffractive fit allows an estimate
of the absorptive corrections in global DIS fit

31. I P I P I P Contribution of diffractive F2 to inclusive F2
Apply the AGK cutting rules to contrib. I P I P I P
AGK in QCD: Bartels & Ryskin
Im Tel ~ stot
DF2abs ~ - F2D
negative (~Glauber shadowing)