1: \section{Photoproduction and QCD}% (1+4+4+4+2+6 pages)

2: 

3: Given the double identity of the photon as a fundamental gauge boson

4: and a hadron-like object, studies of photoproduction reactions provide

5: a number of unique opportunities for exploring QCD.  Combining data

6: from photon-proton collisions with deep inelastic scattering (DIS),

7: photon-photon and proton-(anti)proton collisions can be a particularly

8: fruitful exercise. In this chapter we explore the major areas in QCD

9: where photoproduction data have had such an impact.

10: 

11: \subsection{Proton structure} % (4 pages)

12: \label{sec:proton_structure}

13: 

14: The parton densities in the proton constitute one of the best windows

15: onto the physics of the strong interaction. They encapsulate the

16: behaviour of quarks and gluons over a huge range of distance scales

17: and densities, and their scale dependence provides one of the most

18: precise measurements of the strong coupling, $\as$. In addition,

19: parton densities are a crucial input to all cross-section calculations

20: at proton colliders. Finally, measurements of a range of different

21: cross-sections at hadron colliders, all of which make use of the same

22: parton densities, provide one of the most stringent tests of QCD, and

23: in particular of factorisation.

24: 

25: As with photon structure (see Section~\ref{sec:photon_structure}), the

26: principle measurements used to extract information of the proton's

27: parton density function (PDF) are measurements of the structure

28: functions in DIS, in particular the structure function $\f2$. These

29: data are fitted by several

30: groups~\cite{Kretzer:2003it,Martin:2004dh,epj:c42:1,Adloff:2003uh,Alekhin:2002fv,epj:c5:461}

31: using the conventional DGLAP evolution equations. Put simply, this

32: approach makes no prediction as to the $x$ dependence of the PDFs

33: (apart from information indirectly derived from the conservation of

34: momentum) but for $\q2$ sufficiently above $\lqcd^2$ the $\q2$

35: dependence ({\it i.e}. the scaling violation) is accurately predicted. Fits

36: are currently carried out to NLO accuracy. However, the full

37: calculation of the DGLAP evolution to NNLO has recently been

38: performed~\cite{Vogt:2004mw,*Moch:2004pa,*Moch:2004xu,*Vermaseren:2005qc}. This

39: shows that the perturbative series converges well and that in most

40: cases the NLO approach is rather a good approximation.

41: 

42: Measurements of $\f2$ are only directly sensitive to the quark

43: densities. The gluon density is obtained from the fits since gluon

44: radiation from quarks, as well as gluon splitting to quarks, drives

45: the scaling violations. Nevertheless, measurements which are more

46: directly sensitive to the gluon density are important, partly to test

47: the QCD picture but also because they potentially have a

48: sensitivity to higher $x$ than do extractions based on scaling

49: violations. Since high-$x$ gluons are responsible for a big fraction

50: of the highest energy collisions in the world (currently at the

51: Tevatron and soon at the LHC), the high-$x$ gluon density is a very

52: important object to pin down.

53: 

54: Prompt photons and high transverse energy ($\ETJ$) jets are two

55: processes which have been used in fits to help constrain the

56: gluon. However, theoretical uncertainties (in the case of prompt

57: photons) and experimental uncertainties (in the case of jets) have

58: limited the impact of these data. In addition, high-$\ETJ$ jet 

59: cross-sections at the Tevatron are sensitive to new physics, and much of

60: this sensitivity is lost if the data is simply used to fit the gluon

61: density~\cite{Stump:2003yu}.

62: 

63: High $\ETJ$ jet photoproduction is also directly sensitive to the

64: gluon density in the proton. For direct photon events, this is via the

65: boson-gluon fusion diagram, Fig.~\ref{fig:feyn}a. Resolved events are

66: also sensitive (Fig.~\ref{fig:feyn}b) but the situation here is

67: complicated by the photon PDFs, which are less well known than those

68: of the proton. At high transverse energies (which also means high $x$)

69: the uncertainties due to hadronisation and underlying events are also

70: rather small.

71: 

72: Figure~\ref{fig:kinematicplane} shows the parton kinematics for

73: several experiments. The region over which jet measurements in

74: photoproduction at HERA constrain the gluon density is shown, as well

75: as the regions where Tevatron and HERA DIS measurements provide

76: constraints. The HERA DIS data, as well as fixed target data, extend

77: to higher $x$ and $\q2$ than shown, but give little constraint on the

78: gluon in that region. The region over which the LHC is sensitive to

79: the gluon density is also shown. However over much of this region,

80: actual measurement of the gluon at the LHC will be extremely

81: difficult since the main sensitivity is at low $\ETJ$, where

82: other uncertainties due to hadronisation and the underlying event are

83: important.

84: 

85: \begin{figure}[!htp]

86: \begin{center}

87: ~\epsfig{file=Figures/bw_rev.epsi,height=18cm}

88: \caption[*]{\label{fig:kinematicplane} The parton kinematic plane. The

89: approximate regions in which the gluon in the proton is constrained by

90: various experiments are shown, as well as the whole region of

91: sensitivity at the LHC. The variable $M^2 = \q2$ is the dijet mass for

92: LHC, Tevatron and photoproduction, and the negative

93: four-momentum-transfer-squared for DIS. The dashed lines show the

94: rapidity of the dijet system at the LHC. The assumptions are that

95: precise jet measurements are possible down to $\Et = 60\,\Gev, \eta =

96: 2.0$ at Tevatron. The HERA jet region is taken from

97: \cite{epj:c23:615}.}

98: \end{center}

99: \end{figure}

100: 

101: The ZEUS collaboration has published~\cite{epj:c42:1} the first QCD

102: fits which incorporate jet photoproduction data~\cite{epj:c23:615}

103: (as well as DIS jet data~\cite{Chekanov:2002be}). The inclusion

104: significantly reduces the uncertainties on the gluon density at

105: intermediate and high $x$, as shown in Fig.~\ref{fig:fiterrors}.  This

106: is a graphic demonstration of the power of high energy photoproduction

107: data, but is not the full story. The data set used in these results

108: was taken in 1996-1997, and is statistically limited at high $\ETJ$,

109: the kinematic region corresponding to high $x$. This is only around

110: a third of the HERA I data set; HERA II should increase this again by a

111: factor of around 5. 

112: 

113: As well as simply using more data, the constraint on the gluon density

114: could also be improved by optimising the cross-section for sensitivity

115: to the high $x$ gluon by (for example) extending to the forward

116: region.  In such studies~\cite{cta,*Butterworth:2004is} it has also

117: been observed that resolved photon cross-sections exhibit a large

118: sensitivity to the gluon in the proton. For the existing

119: cross-sections, including the low-$\xgo$ cross-sections in the fit

120: reduces the statistical errors, but increases the systematic errors

121: due to the uncertainties associated with the photon structure, so that

122: there is no net benefit~\cite{epj:c42:1}. Currently no NLO fit for the

123: photon structure with full uncertainties exists, and so even this

124: statement is an estimate. Obviously, a combined NLO QCD fit to the

125: proton and the photon PDFs would bring great benefits. Finally, it is

126: also possible to determine $\as$ from such fits, as is discussed in

127: Section~\ref{sec:as}.

128: 

129: \begin{figure}[!thb]

130: \begin{center}

131: ~\epsfig{file=Figures/ZJZO_SIX.Correlated.eps,height=17cm}

132: \caption[*]{The fractional uncertainty in the gluon PDF before

133: (outer+inner shaded areas) and after (inner shaded area) the inclusion of DIS and

134: photoproduction jet data in the QCD fit~\cite{epj:c42:1}.

135: \label{fig:fiterrors}}

136: \end{center}

137: \end{figure}

138: 

139: It is to be hoped that these possibilities will be vigorously pursued

140: by the fitting groups and experimental collaborations during HERA II.

141: The whole topic of parton distributions and their uncertainties is a

142: very active area at present. Reviews and recent developments may be

143: found elsewhere~\cite{heralhc,Devenish:2004pb}.

144: 

145: \subsection{Measurement of the strong coupling constant} %(2 pages)

146: \label{sec:as}

147: 

148: According to QCD, jet cross-sections should exhibit a non-scaling

149: behaviour due to both the evolution of the structure functions of the

150: colliding beams and to the running of $\as$. Scaling violations

151: have been seen both in measurements of the inclusive deep inelastic

152: scattering cross-section, $F_2$ and for jet cross-sections in

153: $p\bar{p}$ collisions~\cite{prl:70:1376,prl:86:2523}. In the absence

154: of scaling violations, the ratio of jet invariant cross-section,

155: $(E_T^{\rm jet})^4 E^{\rm jet} d^3\sigma/dp_x^{\rm jet}dp_y^{\rm

156: jet}dp_z^{\rm jet}$, at one $\gamma p$ centre-of-mass energy,

157: $W_{\gamma p}$ to that at any other energy would be unity for all $x_T

158: \equiv 2 E_T^{\rm jet}/W_{\gamma p}$. Measurements of inclusive

159: jet photoproduction~\cite{Adloff:2003nr} are shown in

160: Fig.~\ref{fig:scaling_violations}a compared directly with data from

161: $p\bar{p}$

162: collisions~\cite{prl:86:2523,prl:82:2451,*pl:b172:461,*np:b309:405}. The

163: shape of the distributions for $\gamma p$ and $p\bar{p}$ data are

164: compatible for $x_T < 0.2$.  For larger values of $x_T$, the harder

165: distribution in $\gamma p$ collisions arises from the increasing

166: proportion of direct processes and the increasing proportion of the

167: quark density in the resolved photon relative to that in the

168: proton. These results are similar to the inclusive particle production

169: discussed in Section~\ref{sec:res_dir} which led to the observation

170: of the direct component. To test the scaling hypothesis, the ratio of

171: the scaled jet invariant cross-sections as a function of $x_T$ are

172: shown in Fig.~\ref{fig:scaling_violations}b for two chosen values of

173: $W_{\gamma p}$~\cite{pl:b560:7}. A clear deviation from unity, in

174: agreement with NLO QCD predictions, is seen. This is the first

175: observation of scaling violations in $\gamma p$ cross-sections.

176: 

177: \begin{figure}[htp]

178: \begin{center}

179: ~\epsfig{file=Figures/d02-225f9.eps,height=7cm}

180: \put(-25,132){\makebox(0,0)[tl]{\large (a)}}

181: \put(174,172){\makebox(0,0)[tl]{\large (b)}}

182: ~\epsfig{file=Figures/DESY-02-228_3.eps,height=7cm}

183: \end{center}

184: \caption{(a) Scaled jet cross-sections at one value of $W_{\gamma p}$ compared 

185: with results from $p\bar{p}$ collisions and {\sc Pythia}. (b) Ratio of scaled jet 

186: cross-section at two different $W_{\gamma p}$ values compared with NLO QCD 

187: predictions.}

188: \label{fig:scaling_violations}

189: \end{figure}

190: 

191: The inclusive jet cross-section over a wide range of $E_T^{\rm jet}$ also allows a 

192: determination of $\as$ and its energy scale dependence. The high precision 

193: of the data, in particular the well-understood energy scale of the calorimeter, 

194: yielded a value of $\as$ with a total experimental uncertainty of less than 2\%. 

195: The limitation of the analysis arises from the theory which has an uncertainty of 4\%. 

196: This measurement of $\as$ and the many in deep inelastic scattering have also 

197: clearly demonstrated the energy scale dependence or ``running'' of $\as$. The 

198: current collection~\cite{proc:zinnowitz:2004:bethke} of all data is shown in 

199: Fig.~\ref{fig:fitalphas}a. The consistency of the data with the running hypothesis 

200: is clearly seen and the impact of the HERA data is significant. 

201: 

202: The theoretical calculation used to fit the jet data also uses a parametrisation of 

203: the proton structure from global fits to deep inelastic scattering and other data. 

204: However, combining jet and inclusive $F_2$ data as described in the previous section, 

205: the full correlation of the gluon density and $\as$ can be accounted for and these 

206: both accurately extracted from HERA data alone~\cite{epj:c42:1}. The impact of the 

207: jet data can be seen in Fig.~\ref{fig:fitalphas}b, where the $\chi^2$ profile is shown 

208: for the fit to inclusive deep inelastic scattering data with and without the jet 

209: measurements. The extracted value is $\as(M_Z) = 0.1183 \pm 0.0028 {\rm (exp.)} 

210: \pm 0.0008 {\rm (model)}$.

211: 

212: \begin{figure}[htp]

213: \begin{center}

214: ~\epsfig{file=Figures/bethke.eps,height=7cm}

215: \put(-27,192){\makebox(0,0)[tl]{\large (a)}}

216: ~\epsfig{file=Figures/Chi2Profile.eps,height=7.25cm}

217: \put(-27,192){\makebox(0,0)[tl]{\large (b)}}

218: \caption[*]{(a) The running of $\as$ from all 

219: experiments~\cite{proc:zinnowitz:2004:bethke}. The HERA data in both 

220: photoproduction and deep inelastic scattering are shown as the open triangles. (b) 

221: The $\chi^2$ profile for $\as$ before and after the inclusion of DIS and photoproduction 

222: jet cross-sections into the QCD fit~\cite{epj:c42:1}.}

223: \label{fig:fitalphas}

224: \end{center}

225: \end{figure}

226: 

227: %\begin{figure}[!thb]

228: %\begin{center}

229: %\special{psfile=Figures/cta_xsec.ps angle=270 hscale=50 vscale=50} 

230: %\vspace{10.0cm}

231: %\caption[*]{ The potential of HERA II jet photoproduction

232: %data~\cite{cta}. The upper plot compares the uncertainties on the

233: %ZEUS-O~\cite{epj:c42:1} fit to the relevant published jet photoproduction

234: %data~\cite{epj:c23:615}. The lower plot compares the uncertainties from the

235: %same fits, but now for a cross-section optimised to constrain the

236: %gluon. The data in the lower plot are faked, but give an indication of

237: %the experimental uncertainties expected for $500~pb^{-1}$ of HERA II

238: %data.\label{fig:optjets}}

239: %\end{center}

240: %\end{figure}

241: 

242: %Need some (NLO) pp predictions...?

243: 

244: %Ollie should have these, and will publish them himself in/before the

245: %HERALHC proceedings, so we can reference them. I hope.

246: 

247: %%% \subsection{Heavy Flavour Production} % (4 pages)

248: %%% \label{sec:hf}

249: 

250: \subsection{Open heavy-quark production}

251: \label{sec:hf}

252: 

253: %Measurements of the production of heavy quarks provide a wealth of

254: %information on high energy particle collisions. 

255: Photoproduction of heavy quarks is again governed by QCD within the

256: Standard Model. 

257: %The importance of a precise understanding of QCD is

258: %apparent when considering current and future accelerators. Many of

259: %these accelerators will use protons and photons as the colliding

260: %particles, both of which have a hadronic structure and hence are

261: %described by QCD. 

262: Figure~\ref{fig:feyn_ee_gg}b shows the production of heavy quarks in a

263: resolved $\gamma p$ collision. In fact, a similar diagram would apply

264: for resolved $\gamma \gamma$ collisions, or hadron-hadron

265: process. Thus knowledge accumulated at HERA, LEP and the Tevatron will

266: directly benefit future programmes such as the LHC and a future linear

267: collider where heavy quarks will be produced by the same

268: mechanism. Since heavy quarks feature in many signatures for new

269: physics, understanding the QCD production mechanisms becomes even more

270: important.

271: 

272: Theoretically, heavy quarks provide ideal tools for probing QCD due to

273: their relatively large mass, $m_Q \gg \Lambda_{\rm QCD}$, which entails

274: a fast convergence of the perturbative expansion of the cross-section. 

275: The production of heavy quarks is also {\em directly}

276: sensitive to the gluon density in the colliding hadron (see

277: Fig.~\ref{fig:feyn_ee_gg}b). The gluon density is usually determined

278: in the DGLAP-evolution fits to measurements of structure functions in

279: inclusive deep-inelastic scattering, and more recently by including

280: jets (Section~\ref{sec:proton_structure}).  In the case of jets, the

281: non-zero photon virtuality (in DIS) or the $\xgo$ cut (in

282: photoproduction) reduces the uncertainties associated with photon

283: structure. Requiring charm in the final state is another method of

284: achieving the same effect, with a different set of associated benefits

285: and problems.

286: 

287: Several measurements of charm photoproduction have been made at HERA,

288: some of which were discussed in Section~\ref{sec:dual} in the context

289: of the information they provide on the structure of the

290: photon. Measurements of charm production in deep inelastic scattering

291: provide information on the charm contribution to the proton structure

292: function. This subject, although related, is outside the remit of this

293: review and the interested reader is referred to the most recent

294: relevant publications~\cite{pr:d69:012004,*epj:c40:349}.  There are

295: also several results on charm production from the HERA experiments

296: which are preliminary and not yet published. Only the published

297: results on charm and beauty production will be discussed here.

298: 

299: Much of the recent interest in heavy-quark production arose from

300: discrepancies between data from the Tevatron

301: experiments~\cite{prl:71:500,*prl:71:2396,*pr:d53:1051,*pr:d55:2546,*pl:b487:264},

302: CDF and D0, and NLO QCD predictions. The data, measured in many decay

303: channels, was a factor of $2-3$ above NLO QCD for all measured

304: regions. This was in contrast to earlier UA1

305: results~\cite{pl:b256:121} at the $Sp\bar{p}S$ collider and led to a

306: mini-crisis in QCD. Many explanations were put forward as to reasons

307: for this discrepancy: the input parameters to the QCD calculations

308: such as the structure function of the proton and the beauty-quark

309: mass; inaccurate values for the extrapolation of the data outside the

310: measured region; and extensions beyond the Standard Model such as the

311: presence of a light supersymmetric

312: particle~\cite{prl:86:4231}. Results from HERA have contributed to the

313: understanding of heavy-quark production in QCD as discussed below.

314: Measurements from other colliders are also briefly discussed.

315: 

316: A measurement of inclusive $D^*$ mesons currently provides the largest

317: data sample for charm production at HERA. Experimentally, the tagging

318: of $D^*$ mesons via the decay chain $D^* \to D^0 \pi_s \to K \pi

319: \pi_s$ is very clean and although the branching ratio is small

320: ($\approx 2\%$) high-precision physics with about 60~000 events is

321: possible~\cite{epj:c38:29}. Published results on $D^*$ photoproduction

322: cross-sections use a smaller sample but already are of sufficient

323: precision for an informative comparison with QCD calculations. Data

324: can be compared to different types of calculations: the ``massive''

325: and ``massless'' scheme. In the fixed-order, or ``massive'', scheme,

326: $u$, $d$ and $s$ are the only active quarks in the proton and

327: photon; charm and beauty are only produced in the hard scatter. This

328: scheme is expected to work well in regions where the transverse

329: momentum of the outgoing $c$ quark is of the order of the quark

330: mass. At higher transverse momenta, the ``massless'' scheme should be

331: applicable in which charm and beauty are regarded as active flavours

332: (massless partons) in the structure functions of the proton and photon

333: and are fragmented into massive quarks only after the hard process. A

334: calculation which matches the two schemes, called FONLL~\cite{Cacciari:1998it},

335: is also available for inclusive $D^*$ meson production at HERA but has

336: not yet been compared to published data.

337: 

338: Data as a function of the pseudorapidity of the $D^*$ meson are shown

339: in Fig.~\ref{fig:c_hera} compared to two different types of NLO QCD

340: calculations for charm production. Although roughly compatible with

341: the data, none of the calculations fully reproduce the trends of the

342: data, with discrepancies concentrated at high values of $\eta^{D^*}$

343: and medium $p_T^{D^*}$. This could hint at, for example, a low value

344: of the charm-quark mass or inadequacies in the fragmentation

345: model. 

346: 

347: \begin{figure}[htp]

348: \begin{center}

349: ~\epsfig{file=Figures/DESY-98-085_3.eps,height=10.3cm}

350: ~\epsfig{file=Figures/desy98085cedar.eps,height=9.2cm}

351: \end{center}

352: \vspace{-0.5cm}

353: \caption{Cross-sections $d\sigma/d\eta^{D^*}$ for different regions of

354: $p_T^{D^*}$ in charm photoproduction at HERA. The data are compared to

355: calculations of NLO QCD in the massive and massless schemes. The data 

356: are also compared to predictions from {\sc Pythia} and {\sc Herwig} Monte 

357: Carlo programs. Note the {\sc Herwig} prediction is not shown for the 

358: two lowest $p_T^{D^*}$ regions as there is an implicit minimum parton 

359: $p_T$ cut in the calculation which would make the comparison invalid.}

360: \label{fig:c_hera} 

361: \end{figure}

362: 

363: In making the predictions, the fragmentation of the charm quark into a

364: $D^*$ meson is performed using the Peterson function~\cite{pr:d27:105}

365: fitted to $e^+e^-$ data. The applicability of this is unclear at HERA

366: particularly at high $\eta^{D^*}$, {\it i.e.} in the direction of the

367: proton, where the proton remnant may affect the production of the

368: $D^*$ meson. Clearly more phenomenological work on fragmentation is

369: needed before stronger conclusions can be drawn. Combining NLO

370: calculations with sophisticated models of hadronisation available in

371: Monte Carlo models, such as the MC@NLO

372: programme~\cite{jhep:0206:029,*jhep:0308:007} is one promising path to

373: understanding charm production in more detail. Another is to check the

374: assumptions which go into the calculations in more detail. For

375: example, the universality of hadronisation may be tested by measuring

376: the production fractions for different charmed hadrons, as has been

377: done very recently in DIS~\cite{Aktas:2004ka} and

378: photoproduction~\cite{zeusfrag}. In general the results are in

379: agreement and, in the case of the photoproduction data, of competitive 

380: precision with $e^+e^-$ data. However, in photoproduction the fraction

381: of charm quarks fragmenting to $\Lambda_c^+$ is around $3\sigma$

382: higher and the $D^{*+}$ fraction lower by a similar amount than the

383: corresponding fractions in $e^+e^-$. About half of the difference in

384: the $D^{*+}$ fraction is due to the $\Lambda_c$ difference. Since

385: there is no asymmetry between $\Lambda_c^+$ and $\Lambda_c^-$, this

386: effect, if real, is unlikely to be due simply to baryon number

387: flow. However, one may speculate that it might be a hint of some

388: non-trivial effects in fragmentation due to the incoming proton.

389: 

390: The first measurements~\cite{pl:b467:156,*epj:c18:625} of beauty

391: production at HERA were higher than the NLO prediction by a similar

392: factor to those from the Tevatron. However, given the large

393: uncertainties on the ZEUS result it was consistent with the

394: theory. The precision of the data subsequently improved, and by

395: measuring cross-sections in kinematic regions where the detector

396: acceptance is good (rather than relying on models to extrapolate into

397: unmeasured regions), the estimates of the systematic errors also

398: became more reliable. The cross-section as a function of the

399: pseudorapidity of the decayed muon for beauty events is shown in

400: Fig.~\ref{fig:b_hera}. Data from both H1~\cite{epj:c41:453} and

401: ZEUS~\cite{pr:d70:012008} collaborations are shown compared with NLO

402: QCD predictions performed in the massive scheme. The data are

403: well-described by the theoretical calculation. Similar measurements

404: have been made in deep inelastic

405: scattering~\cite{pl:b599:173,epj:c41:453}; here the data are above

406: the NLO QCD predictions but consistent within the quoted uncertainties.

407: 

408: \begin{figure}[htp]

409: \begin{center}

410: ~\epsfig{file=Figures/desy03212cedar.eps,height=10.95cm}

411: \end{center}

412: \caption{Cross-section $d\sigma/d\eta^\mu$ for beauty photoproduction

413: at HERA. The data are compared to predictions of NLO QCD in the

414: massive scheme and from {\sc Pythia} and {\sc Herwig} Monte

415: Carlo programs.}

416: \label{fig:b_hera}

417: \end{figure}

418: 

419: Recent theoretical improvements and their comparison with measured,

420: rather than extrapolated, cross-sections have also led to an improved

421: description of beauty production in $p\bar{p}$ collisions. Two new

422: calculations have become available.  The FONLL calculation discussed

423: previously, matches the massive and massless schemes and should

424: therefore be applicable for all energies. This prediction also has an

425: improved fitting technique to $e^+e^-$ data for the fragmentation

426: function.  The second, MC@NLO, matches a NLO calculation to

427: parton showers and hadronisation in the {\sc Herwig} MC

428: programme. These calculations are shown in Fig.~\ref{fig:b_tev_lep}

429: compared with recent CDF data~\cite{Acosta:2004yw}; both describe the

430: data well. The rate of beauty production has also been measured in

431: $\gamma \gamma$ collisions by the L3 collaboration~\cite{Acciarri:2000kd} at

432: LEP. The data are measured within the acceptance of the detector and

433: then extrapolated by factors of $\sim 20-30$ to the full phase space

434: for comparison with NLO QCD predictions. This measurement is 

435: experimentally challenging due to the poor separation power between 

436: beauty decays to leptons and background. The NLO QCD prediction, shown

437: in Fig.~\ref{fig:b_tev_lep}, is a factor of three below the data

438: although the difference is below three standard deviations. Were this

439: result to persist with measurements from other LEP collaborations it

440: would certainly be very puzzling as a large part of the cross-section

441: comes from electromagnetic coupling of two photons to two beauty

442: quarks.

443: 

444: \begin{figure}[htp]

445: \begin{center}

446: ~\epsfig{file=Figures/cdfpsi.eps,height=6.5cm}

447: \put(-27,172){\makebox(0,0)[tl]{\large (a)}}

448: \put(164,122){\makebox(0,0)[tl]{\large (b)}}

449: ~\epsfig{file=Figures/figure5.eps,height=6cm}

450: \end{center}

451: \caption{Cross-sections (a) $d\sigma/dp_T$ for $J/\psi$ mesons from

452: beauty decays at the Tevatron and (b) beauty production at LEP. The

453: data are compared to the different NLO QCD calculations. }

454: \label{fig:b_tev_lep}

455: \end{figure}

456: 

457: As well as NLO calculations, full final-state Monte Carlos can also

458: give a reasonable description of most heavy flavour cross-sections

459: which have been measured in photoproduction and other high-energy

460: processes. Given the importance of heavy flavour production

461: cross-sections, particularly beauty, in identifying interesting

462: processes at LHC and a future linear collider, detailed validation of

463: such models across all available measurements is of considerable

464: interest and importance, and is

465: ongoing~\cite{gutsche,Butterworth:2004mu,*Butterworth:2002ts,heralhc}.

466: 

467: \subsection{Charmonium production}

468: \label{sec:ineljpsi}

469: 

470: The $J/\psi$ meson was discovered over 30 years ago and thereby

471: confirmed the existence of the charm quark. However, although the

472: particle has been extensively measured, the nature of its production

473: is poorly understood. Quasi-elastic production of the $J/\psi$ and

474: other vector mesons was briefly discussed in

475: Section~\ref{sec:totalxsec}. The mechanism by which $J/\psi$ mesons are

476: ``inelastically'' produced is expected to be predictable in QCD. The

477: inelastic domain arises from the presence of a large transfer of

478: energy between the projectile and the target particle.

479: 

480: In the so-called colour-singlet model~\cite{Berger:1980ni}, $c\bar{c}$

481: pairs produced in a colour-singlet (CS) state are identified with physical

482: $\psi$ states. This model has been extended in the context of

483: non-relativistic

484: QCD~\cite{Bodwin:1994jh,*Cho:1995vh,*Cacciari:1996dg,*Ko:1996xw} to

485: account for the production of $c\bar{c}$ pairs which are not

486: colourless. The charm quark pair can emit gluons (and hence change

487: colour) until the pair is colourless. This model is referred to as the

488: colour-octet (CO) model. Measurements by the CDF collaboration in

489: $p\bar{p}$ collisions~\cite{prl:79:572} indicate that the

490: colour-singlet model greatly underestimates the production rate of

491: $\psi$ mesons, by factors of between 10 and 80. However, it has been

492: claimed that calculations including colour-octet contributions can

493: account for this difference.  A test of this claim is the requirement

494: that the colour-octet matrix elements extracted from $\psi$ cross-section 

495: measurements in $p\bar{p}$ collisions can also account for

496: photoproduction data.

497: 

498: The cross-sections $d\sigma/dp_T^2$ and $d\sigma/dz$ have been

499: measured in photoproduction~\cite{Chekanov:2002at,Adloff:2002ex},

500: where $z$, the ``inelasticity'', is the fraction of the incoming

501: photon's energy carried by the $J/\psi$ meson. The distributions of

502: these two variables are shown in \mbox{Fig.~\ref{fig:jpsi_hera}}

503: compared with theoretical calculations incorporating the models

504: mentioned previously. In Fig.~\ref{fig:jpsi_hera}a, a prediction from

505: the colour-singlet model alone \mbox{(LO, CS)} clearly does not

506: describe the data being a factor of over 10 below at high

507: $p_T^2$. After the inclusion of NLO

508: corrections~\cite{Kramer:1994zi,*Kramer:1995nb}, however, the theory

509: agrees well with the data. This calculation is shown again in

510: Fig.~\ref{fig:jpsi_hera}b for the inelasticity variable and again

511: describes the data reasonably well. At low $z$, the NLO prediction is

512: below the data although in this region more background from

513: e.g. resolved photons is expected. Also shown are calculations which

514: have both colour-singlet and colour-octet contribution, shown as

515: \mbox{(LO, CS+CO)}\cite{Kramer:2001hh}. They also describe the data

516: well. However, other models incorporating colour-octet

517: processes~\cite{Beneke:1999gq,*Kniehl:1998qy} give different

518: results~\cite{Chekanov:2002at}. The amount of colour-octet in

519: different calculations was extracted from the same $p\bar{p}$ data;

520: thus it is evident these calculations suffer from large uncertainties

521: from other sources, and hence a lack of predictive power.

522: 

523: In summary, there seems to be a real issue with the theory here.  The

524: very large NLO corrections to the CS calculations indicate, at least

525: naively, poor convergence of the perturbative expansion. These

526: corrections have not been calculated for the CS+CO model; the LO

527: predictions for this model differ substantially from each other. Due

528: to the large theoretical uncertainties affecting all predictions the

529: presence of colour-octet processes in photoproduction cannot be

530: excluded or definitely confirmed, and the matrix elements extracted

531: from hadroproduction are consistent with photoproduction data within

532: very large uncertainties. More precision comparisons with theory

533: require a better theoretical understanding of the uncertainties,

534: and/or the measurement of more accurately predicted variables, such as

535: perhaps angular distributions.

536: 

537: \begin{figure}

538: \begin{center}

539: ~\epsfig{file=Figures/dsigmadpt2-z+h.eps,height=7.5cm}

540: \put(-27,199){\makebox(0,0)[tl]{\large (a)}}

541: \put(174,199){\makebox(0,0)[tl]{\large (b)}}

542: ~\epsfig{file=Figures/dsigmadz-pt1-z+h.eps,height=7.5cm}

543: \end{center}

544: \caption{Cross-sections (a) $d\sigma/dp_T^2$ and (b) $d\sigma/dz$ for

545: $J/\psi$ photoproduction at HERA. Data from both H1 and ZEUS

546: collaborations are compared with predictions at (a) LO and NLO in the

547: framework of the colour-singlet model and (b) NLO in the framework of

548: the colour-singlet model and LO with the addition of the

549: colour-octet model.}

550: \label{fig:jpsi_hera}

551: \end{figure}

552: 

553: 

554: \subsection{Jet substructure and fragmentation functions} % (4 pages)

555: \label{sec:substructure}

556: 

557: Measurement of jet substructure allows the possibility of distinguishing 

558: jets originating from quarks and gluons due to their different radiation 

559: amplitudes. This permits a stringent test of QCD, provides information 

560: on the properties of quark- and gluon-initiated jets and, by tagging the 

561: final state, reveals details of the initial state and hence proton and 

562: photon structure (see Section~\ref{sec:nature_of_photon}). At 

563: sufficiently high jet transverse energy, where the influence of 

564: hadronisation becomes negligible, the internal structure of a jet is 

565: calculable in pQCD. Such calculations predict that gluon-initiated jets 

566: are broader than quark-initiated jets due to the larger colour charge of 

567: the gluon. The jet shape~\cite{prl:69:3615} and subjet

568: multiplicity~\cite{np:b383:419,*np:b421:545,*pl:b378:279,*jhep:9909:009}

569: have been used to study the internal structure of jets in 

570: photoproduction. The integrated jet shape, $\psi(r)$, using only those

571: particles belonging to the jet, is defined as the fraction of the jet

572: transverse energy that lies inside a cone in the $\eta-\phi$ plane of

573: radius, $r$, concentric with the jet axis:

574: 

575: \begin{equation}

576: \psi(r)  = \frac{E_T(r)}{E_T^{\rm jet}},

577: \label{eq:jet_shape}

578: \end{equation}

579: 

580: where $E_T(r)$ is the transverse energy within the given cone of radius 

581: $r$. The mean integrated jet shape, $\langle \psi(r) \rangle$, is defined 

582: as the averaged fraction of the jet transverse energy inside the cone

583: $r$:

584: 

585: \begin{equation}

586: \langle \psi(r) \rangle = \frac{1}{N_{\rm jets}} \sum_{\rm jets} \frac{E_T(r)}{E_T^{\rm jet}},

587: \label{eq:int_jet_shape}

588: \end{equation}

589: 

590: where $N_{\rm jets}$ is the total number of jets in the sample.

591: 

592: The internal structure of a jet

593: is expected to depend mainly on the type of primary parton from which

594: it originated and to a lesser extent on the particular hard scattering

595: process. This expectation is supported by comparisons such as that

596: shown in Fig.~\ref{fig:jet_shapes}. Here it is seen that in those

597: processes which at LO give rise only to quark jets ($e^+e^-$

598: annihilation, DIS), the jets are narrow, and very similar to each other. In

599: photoproduction at these transverse energies, direct processes

600: dominate. However, the presence of QCD Compton and resolved photon

601: processes means that some gluon jets are present in the final state at LO,

602: and the average jet shape is correspondingly broader. In $p\bar{p}$

603: events, where the gluon admixture is even stronger, the average jet

604: shape is broader still.

605: 

606: \begin{figure}[ht]

607: \begin{center}

608: ~\epsfig{file=Figures/jet_shapes.eps,height=11cm}

609: \end{center}

610: \caption[*]{Measured jet shapes in $e^+e^-$~\cite{Akers:1994wj},

611: DIS~\cite{Breitweg:1998gf}, photoproduction~\cite{epj:c2:61} and

612: $p\bar{p}$~\cite{Abe:1992wv,*Abachi:1995zw}.}

613: \label{fig:jet_shapes}

614: \end{figure}

615: 

616: The measured mean integrated jet shape for $E_T^{\rm jet} > 17$\,GeV

617: in different regions of $\eta^{\rm jet}$ is shown in

618: Fig.~\ref{fig:q_g_shapes}~\cite{Chekanov:2004kz}. The jets broaden as

619: $\eta^{\rm jet}$ increases. Leading-logarithmic parton-shower

620: predictions from {\sc Pythia} for resolved plus direct processes and

621: gluon- and quark-initiated jets are compared to the data. The

622: description of the data is generally good, although the data are

623: somewhat broader than the predictions. From the comparison with the

624: predictions of gluon- and quark-initiated jets, it is seen that the

625: measured jets are quark-like for $-1 < \eta^{\rm jet} <0$ and become

626: increasingly gluon-like as $\eta^{\rm jet}$ increases. Jets become

627: narrower as $E_T^{\rm jet}$ increases (not shown) consistent with the

628: dominance of direct processes, and hence quark jets, at higher

629: $E_T^{\rm jet}$. These effects have also been seen in measurements of

630: the subjet multiplicity~\cite{Chekanov:2004kz}.

631: 

632: \begin{figure}[ht]

633: \begin{center}

634: ~\epsfig{file=Figures/DESY-04-072_1.eps,height=13cm}

635: \end{center}

636: \caption{Measured mean integrated jet shape in different regions of 

637: $\eta^{\rm jet}$. Predictions of the {\sc Pythia} Monte Carlo program 

638: for quark (dot-dashed lines), gluon (dashed lines) and all (solid lines) 

639: jets are shown. The fractional difference of the data to the predictions 

640: of {\sc Pythia} for all jets is also shown (open dots).}

641: \label{fig:q_g_shapes}

642: \end{figure}

643: 

644: The differences in gluon- and quark-initiated jets are used to select samples 

645: enriched in such jets to study their properties and the dynamics of the 

646: hard subprocesses in more detail. The two samples of jets were selected on a 

647: statistical basis and classified as follows:

648: 

649: \begin{itemize}

650: 

651: \item gluon-enriched sample (``broad jets''), defined as those jets with 

652:       $\psi(r=0.3) < 0.6$;

653: 

654: \item quark-enriched sample (``narrow jets''), defined as those jets with 

655:       $\psi(r=0.3) > 0.8$.

656: 

657: \end{itemize}

658: 

659: Non-overlapping regions were chosen to suppress migration effects and the 

660: cuts values represent a compromise between purity and statistics. In dijet 

661: events, samples with two broad jets, two narrow jets and one broad and one 

662: narrow jet can also be selected. In Fig.~\ref{fig:q_g_xsecs}, cross-sections 

663: for broad and narrow jets (Fig.~\ref{fig:q_g_xsecs}a and~\ref{fig:q_g_xsecs}b) 

664: and events with two broad or two narrow jets (Fig.~\ref{fig:q_g_xsecs}c 

665: and~\ref{fig:q_g_xsecs}d) are shown.

666: 

667: \begin{figure}[ht]

668: \begin{center}

669: ~\epsfig{file=Figures/gp_shape1811jsncol.eps,height=7cm}

670: \put(-27,177){\makebox(0,0)[tl]{\large (a)}}

671: \put(174,177){\makebox(0,0)[tl]{\large (b)}}

672: ~\epsfig{file=Figures/gp_shape1821jsnncol.eps,height=7cm}

673: ~\epsfig{file=Figures/gp_shape1841jsncol.eps,height=7cm}

674: \put(-27,177){\makebox(0,0)[tl]{\large (c)}}

675: \put(174,177){\makebox(0,0)[tl]{\large (d)}}

676: ~\epsfig{file=Figures/gp_shape1831jsncol.eps,height=7cm}

677: \end{center}

678: \caption{Cross-sections (a) $d\sigma/d\eta^{\rm jet}$ and 

679: (b) $d\sigma/dE_T^{\rm jet}$ for broad and narrow jets and 

680: (c) $d\sigma/d|\cos\theta^*|$ and (d) $d\sigma/dx_\gamma^{\rm obs}$ for 

681: events with two broad jets or two narrow jets. The data are compared with 

682: {\sc Pythia MC} predictions and in (a) for {\sc Pythia} expectations for 

683: quark and gluon jets.}

684: \label{fig:q_g_xsecs}

685: \end{figure}

686: 

687: The measured cross-sections for broad and narrow jets exhibit a different 

688: behaviour: the $\eta^{\rm jet}$ distribution for broad jets increases up to 

689: the highest $\eta^{\rm jet}$ value measured whereas the distribution for 

690: narrow jets peaks at $\eta^{\rm jet} \sim 0.7$; and the distribution in 

691: $E_T^{\rm jet}$ for narrow jets has a harder spectrum than for broad jets. 

692: The Monte Carlo prediction, normalised to the total cross-section, gives 

693: a good description of the narrow-jet sample but gives a poorer description 

694: of the broad-jet sample. The {\sc Pythia} programme predicts that the 

695: broad-jet event sample consists of 15\% $gg$, 50\% $gq$ and 35\% $q\bar{q}$. 

696: The selection of narrow jets gives a reasonably pure sample of quark-initiated 

697: jets; the {\sc Pythia} programme predicts 62\% $q\bar{q}$, 34\% $qg$ and 

698: 4\% $gg$. In Fig.~\ref{fig:q_g_xsecs}a, predictions for quark-initiated jets 

699: describe the narrow-jet sample well, as expected due to its high quark purity. 

700: The prediction for gluon-initiated jets describes the broad-jet sample less 

701: well due to the higher impurity from quark-initiated jets. However, the 

702: result supports the expectation that the broad- and narrow-jet samples are 

703: dominated by gluon-  and quark-initiated jets, respectively.

704: 

705: Dijet cross-sections in which either two broad or two narrow jets are

706: tagged are shown in Fig.~\ref{fig:q_g_xsecs}c

707: and~\ref{fig:q_g_xsecs}d. As these jets are indistinguishable, only

708: the absolute value of $\cos\theta^*$ can be measured (in contrast to

709: the case when one broad and one narrow jet are tagged, as in

710: Fig.~\ref{fig:costheta}). The measured and predicted cross-sections

711: were normalised at $|\cos\theta^*| = 0.1$. The cross-section for

712: events with two broad jets exhibits a faster rise to high

713: $\cos\theta^*$. The predictions from the {\sc Pythia} Monte Carlo

714: programme give an adequate description of the data.  For events with

715: two broad jets, {\sc Pythia} predicts the parton final state to

716: consist of 16\% $gg$, 52\% $qg$ and 32\% $qq$. As in the sample with only

717: one broad jet, the impurity is relatively large. The sample with two

718: narrow jets is a much purer sample of quark jets; {\sc Pythia}

719: predicts the parton final state to consist of 71\% $qq$, 28\% $qg$ and

720: 1\% $gg$. The differences in the measured distributions can be

721: understood in terms of the dominant two-body processes: the resolved

722: subprocess $q_\gamma g_p \to qg$, mediated by gluon exchange for the

723: broad-broad dijet sample and the direct subprocess $\gamma g \to

724: q\bar{q}$, mediated by quark exchange for the narrow-narrow dijet

725: sample. The dominance of resolved and direct processes for broad-broad

726: and narrow-narrow dijet events, respectively is also confirmed by the

727: cross-section $d\sigma/dx_\gamma^{\rm obs}$. The narrow-narrow dijet

728: sample is strongly peaked at high $x_\gamma^{\rm obs}$ whereas the

729: broad-broad dijet sample is roughly flat in $x_\gamma^{\rm obs}$.

730:  

731: %%{\bf Single particle fragmentation functions and fractions. Not sure.\\

732: %    Inclusive has not so much to say.\\

733: %     Charm function and fractions are not published.}

734: 

735: 

736: \subsection{Multijet production and colour coherence}

737: 

738: Jet photoproduction has so far mostly been discussed in terms of the

739: LO diagrams (Fig.~\ref{fig:feyn}) where two jets are produced. In

740: reality, a combination of QCD radiation and hadronisation will mean

741: that there is a distribution of energy around the final state outside

742: the two leading jets. This may be in the form of low energy particles

743: from soft processes, or further high $\ETJ$ jets. The distribution of

744: these jets, and to some extent even of softer particles, is predicted

745: by QCD.

746: 

747: The three-jet photoproduction cross-section has been measured by the

748: ZEUS collaboration~\cite{pl:b443:394}. The measurement was made for

749: three-jet masses above 50~GeV and with a requirement that at least two

750: jets satisfy $\ETJ > 6$~GeV while the third has $\ETJ > 5$~GeV. QCD

751: calculations at ${\cal O}(\alpha\as^2)$ (which is LO for three jet

752: production) are in good agreement with the data. In addition, since

753: this cross-section is dominated by configurations in which the third

754: jet has substantially lower $\ETJ$ than the two leading jets,

755: leading-logarithmic parton-shower models are expected to describe the

756: cross-sections rather well. They do indeed, as may be seen in

757: Fig.~\ref{fig:3jet}. An interesting feature here is that coherent

758: gluon radiation must be implemented in the Monte Carlo in order to

759: reproduce the shape of the $\psi_3$ distribution (the angle, in the

760: three-jet centre-of-mass system, between the plane containing the leading jet and the

761: beam, and the plane containing the three jets). For coherent

762: radiation, these planes tend to line up such that the region $\psi_3

763: \approx \pi/2$ is depopulated. If a large amount of incoherent

764: radiation is allowed (see appendix), this effect is destroyed.

765: 

766: \begin{figure}

767: \centering

768: \leavevmode

769: ~\epsfig{file=Figures/desy98162cedar.eps,height=8cm}

770: \put(-234,191){\makebox(0,0)[tl]{\large (a)}}

771: \put(-29,191){\makebox(0,0)[tl]{\large (b)}}

772: \caption{(a) The cross-section $d\sigma / d\psi_3$  

773: and (b) the area-normalised distribution of $\psi_3$.  

774: The correlated systematic uncertainty due to the jet

775: energy-scale is shown as the shaded band in (a). The solid

776: histogram shows the default {\sc Pythia} prediction. In (a) and (b) 

777: the dotted and dashed histograms show the predictions of {\sc Herwig} 

778: and of {\sc Pythia} with colour coherence switched off, respectively.

779: \label{fig:3jet}}

780: \end{figure}

781: 

782: \subsection{Energy flow, rapidity gaps and forward particles} % (6 pages)

783: \label{sec:eflow_rapidity_gaps}

784: 

785: As discussed in Section~\ref{sec:totalxsec}, total hadronic cross-sections 

786: in the limit of large $s/t$ increase by some small (but

787: positive!) power of $s$. In Regge theory this is an exchange of vacuum

788: quantum numbers, known as the Pomeron. This is presumably an emergent

789: behaviour of the QCD Lagrangian, but making a connection with the

790: partonic language of perturbative QCD is a challenge which continues

791: to throw up surprises in both data and theory\footnote{See

792: \cite{forshaw:1997:pomeron} for a discussion and

793: bibliography.}.

794: 

795: The high energy regime probed at HERA and the Tevatron means that the

796: large $s/t$ limit may be approached even when $t$ is sufficiently

797: large for perturbative techniques to be applied. It is also possible,

798: even when the momentum exchange, $t$, with the proton is small, for

799: relatively high-mass final states to be produced, which opens up the

800: phase space for short distance, high energy scale processes in the

801: diffractively produced system. These two distinct cases both offer the

802: possibility of describing some or all of ``Pomeron'' exchange in terms

803: of perturbative quarks and gluons. Both have been measured in high

804: energy photoproduction.

805: 

806: \subsubsection{Hard colour singlet exchange}

807: 

808: In configurations where the two leading jets are separated by a large

809: rapidity interval, it is possible in high energy photoproduction to

810: approach the regime where $t \approx \pt^2$ is of the order a few tens

811: of GeV, but $\hat{s}$, the partonic centre-of-mass energy squared, is

812: a few hundred GeV. Thus the scattering will be

813: dominated by the exchange of vacuum quantum numbers, but yet the high

814: $t$ means that pQCD calculations should be applicable. The lowest

815: order colour-singlet exchange in QCD is a pair of gluons. However,

816: such a diagram is subject to large logarithmic corrections, which are

817: resummed in the BFKL~\cite{jetp:45:199,sovjnp:28:822} approach. This is

818: a topic in QCD with a very extensive literature (see for example,

819: references within~\cite{forshaw:1997:pomeron}), but for our purposes

820: here it is sufficient to say that there is a possible enhancement in

821: the cross-section for colour-singlet exchange between jets separated

822: by a large rapidity interval in hadronic collisions.

823: 

824: If no colour is exchanged between the hadrons (in the case of

825: photoproduction, the proton and a hadronic photon), colour coherence

826: leads to a suppression of QCD radiation, and thus energy flow, between

827: the jets. Hence the large rapidity interval is very likely to be a

828: rapidity gap. This is the experimental signature searched for by the

829: HERA collaborations. A key observable is the ``gap fraction''

830: measured as a function of the rapidity interval, $f(\Delta\eta)$. This

831: is the ratio of the cross-section for dijet events with a rapidity

832: gap between the jets (however the gap is defined) to the total number

833: of dijet events.

834: 

835: In the first measurement~\cite{pl:b369:55}, a gap was defined by the

836: requirement that no particle with transverse energy $>300$~MeV lay in

837: the interval between the jets. In ``normal'' QCD events, the

838: probability of zero radiation into the gap falls exponentially as the

839: interval increases, and thus so does $f(\Delta\eta)$. This is seen in

840: the first bins of Fig.~\ref{fig:gbj1}a and \ref{fig:gbj1}b. However,

841: for colour singlet exchange no such suppression is expected, and

842: indeed the final bin is higher than one would expect given the fall in

843: the first three. A simple fit to a flat fraction for colour singlet

844: exchange combined with the exponential fall gives good description of

845: the data. This was the first evidence for strongly interacting

846: colour-singlet exchange in jet photoproduction. Such measurements have

847: also been made at the

848: Tevatron~\cite{Abachi:1995gz,*Abbott:1998jb,*Abe:1994de,*Abe:1997ie}

849: and show similar evidence for such processes in $p\bar{p}$ events,

850: albeit at a much lower rate.

851: 

852: \begin{figure}

853: \centering

854: \leavevmode

855: ~\epsfig{file=Figures/desy95194cedar.eps,height=8cm}

856: \put(-234,161){\makebox(0,0)[tl]{\large (a)}}

857: \put(-29,161){\makebox(0,0)[tl]{\large (b)}}

858: \caption[*]{ZEUS gap fraction measurement (solid points). 

859: In (a) the gap fraction is compared

860: to a fit to the sum of an exponential (non-colour-singlet) and a flat

861: (assumed colour singlet) contribution. In (b) the gap fraction is

862: redisplayed and compared to {\sc Pythia} and {\sc Herwig}

863: non-diffraction MC, and to a {\sc Herwig} sample which includes a

864: colour singlet exchange component from a BFKL-based

865: calculation~\cite{pl:b284:123}.

866: \label{fig:gbj1}

867: }

868: \end{figure}

869: 

870: The main problem with this measurement, apart from the limited

871: statistics and rapidity reach available at the time, is that the

872: method used to define a rapidity gap is not infrared safe. That is, a

873: definition of activity in terms of particles is not amenable to

874: calculation in perturbative QCD. A more robust definition can be made

875: in terms of energy flow within the interval, and by using minijets

876: rather than particles to calculate this energy

877: flow~\cite{Cox:1999dw,*wyatt,*mhs,Oderda:1999kr,Oderda:1999ta}.  Such

878: a definition was used by the H1 collaboration in a subsequent

879: measurement~\cite{Adloff:2002em}. In this analysis, the $\kt$ jet

880: algorithm~\cite{np:b406:187} was run over the final state in the

881: inclusive mode. This produces a list of ``jets'', some of which have

882: very low transverse energy and consist of very few particles.

883: However, since the algorithm is infrared safe, quantities based on

884: these objects are, in principle at least, calculable. The highest

885: $\ETJ$ pair define the rapidity interval, as usual. The cross-sections

886: and gap fraction in the paper are defined in terms of $\etgap$, the

887: transverse energy of all the jets within the rapidity interval between

888: the jets. The requirement $\etgap < E_T^{\rm cut}$ was applied, for

889: values of $E_T^{\rm cut}$ ranging between 0.5~GeV and 2~GeV.  For all

890: values of $E_T^{\rm cut}$ studied, (Fig.~\ref{fig:h1gbj}) there is an

891: excess of gap events at large rapidity intervals compared to the

892: expected fall-off from normal QCD events.

893: 

894: \begin{figure}[ht]

895: \begin{center}

896: ~\epsfig{file=Figures/desy02023cedara.eps,height=12cm}

897: \put(-176,310){\makebox(0,0)[tl]{\large (a)}}

898: \put(-25,310){\makebox(0,0)[tl]{\large (b)}}

899: \put(-176,170){\makebox(0,0)[tl]{\large (c)}}

900: \put(-25,170){\makebox(0,0)[tl]{\large (d)}}

901: \end{center}

902: \caption{H1 gap fraction measurement (solid points) differential in

903: $\Delta \eta$. Gap events are defined for four values of

904: $\etgap$, shown in (a-d). The gap fractions are compared to the

905: prediction of \herwig~(dashed line) and \pythia~(solid line).  The

906: dotted line shows \herwig~+ BFKL colour singlet exchange (see

907: text).\label{fig:h1gbj}}

908: \end{figure}

909: 

910: The exponential fall of the gap fraction expected for

911: non-colour-singlet processes is also seen in the non-colour singlet

912: {\sc Herwig} and {\sc Pythia} curves shown in Fig.~\ref{fig:gbj1}b and

913: in Fig.~\ref{fig:h1gbj}. The data are also compared to {\sc

914: Herwig} predictions~\cite{Butterworth:1995jf} implementing a

915: leading-logarithmic BFKL-based calculation of hard colour-singlet

916: exchange between two partons as computed by Mueller and

917: Tang~\cite{pl:b284:123}. In the limit $\Delta\eta \gg 1$ the 

918: cross-section for \mbox{quark--quark} scattering may be approximated as

919: \begin{equation}

920: \frac{d \sigma(q q \to q q)}{d \hat t} \approx (C_F \as)^4

921: \frac{2\pi^3}{\hat t^2} \frac{\exp(2 \omega_0 y)}{(7 \as C_A

922: \zeta(3) y)^3}

923: \label{mullertang}

924: \end{equation} 

925: where 

926: \begin{equation}

927: \omega_0 = C_A (4 \ln 2/\pi) \as.

928: \label{intercept}

929: \end{equation}

930: and

931: \begin{equation}

932: y = \Delta\eta = \ln \left( \frac{\hat{s}}{-\hat t} \right)

933: \end{equation}

934: Here, $1+\omega_0$ is the perturbative Pomeron intercept, $C_F =

935: \frac{4}{3}$ is the usual colour factor for \mbox{quark--quark} scattering,

936: $C_A$ is the number of colours, $-t \approx (\ETJ)^2$, and $\zeta$ is

937: the Riemann $\zeta$-function. In the leading-logarithm approximation, the 

938: values of $\as$ in Eqs.\ref{mullertang} and \ref{intercept} are free parameters. 

939: In the H1 analysis~\cite{Adloff:2002em}, where the exact

940: rather than asymptotic form of Eq.\ref{mullertang} was

941: used~\cite{Cox:1999dw}, these were both set to 0.18. This corresponds

942: to a choice of Pomeron intercept of $1 + \omega_0 = 1.48$, which was

943: used for the asymptotic {\sc Herwig} curves shown here. For both 

944: ZEUS~(Fig.~\ref{fig:gbj1}) and H1 (Fig.~\ref{fig:h1gbj}), good

945: agreement with the data is obtained for this model.

946: 

947: For this kind of measurement in a real experiment with finite rapidity

948: and non-zero $\etgap$, the large logarithms in rapidity, or $x$, which

949: are summed in the BFKL approach are not necessarily the only, or even

950: the most important, terms which should be included. There are also

951: terms with logarithms in $\etgap/\ETJ$ to be considered. Such

952: logarithms have been resummed for various jet

953: definitions~\cite{Oderda:1999kr,Oderda:1998en,Berger:2001ns,Appleby:2002ke,*Appleby:2003sj},

954: and there is progress towards an approach which sums all important

955: logarithms of both types~\cite{Forshaw:2005sx}.

956: 

957: Two other photoproduction processes which give access to hard colour

958: singlet exchange are the high-$|t|$ diffractive photoproduction of vector

959: mesons, and of photons.

960: 

961: The main advantage that diffractive photoproduction of vector mesons

962: at high $|t|$ has over jet photoproduction is the experimental

963: cleanliness of the measurement. At HERA, for example, the final state

964: consists of; the decay products of the vector meson, an electron

965: scattered through a very small angle (which may escape down the beam

966: pipe), and a low-mass hadronic system in the forward (proton)

967: direction. The $\q2$ is constrained to be low by the low

968: scattering-angle of the electron, and thus $t$ may be determined from

969: the vector meson decay products.  Theoretically, the unknown

970: fragmentation and hadronisation effects involved in the formation of

971: jets are absent, but the transition probability of a photon to a

972: vector meson is introduced.

973: 

974: Calculations of high-$|t|$ vector meson exchange based upon BFKL

975: resummation have been

976: performed~\cite{zfp:c68:137,pl:b375:301,pl:b478:101,Enberg:2003jw}.

977: In this approach, as with the calculations of rapidity gaps between

978: jets, a ladder of gluons (``hard Pomeron'') couples to a single parton

979: in the proton at one end. At the other end things are very different;

980: the photon splits into a $q\bar{q}$ pair. The probability to form a

981: vector meson is calculated from the overlap between the photon and

982: vector meson wavefunctions caused by the coupling to the two gluons at

983: the top of the ladder to the dipole formed by the $q\bar{q}$ pair.

984: The cross-section is predicted to fall as $1/|t|^N$, and to rise

985: strongly with the photon-proton centre of mass energy, $W_{\gamma p}$.

986: The calculations are able to describe the measured magnitude of the

987: cross-sections for $\rho$, $\phi$ and $J/\psi$

988: production~\cite{epj:c14:213,Chekanov:2002rm,Aktas:2003zi}. They also

989: describe the dependence on $W_{\gamma p}$ for $J/\psi$ production for

990: $|t| > 5$~GeV~\cite{Aktas:2003zi}.

991: 

992: The transition of a quasi-real photon (helicity -1 or +1) to a vector

993: meson (helicity -1, 0 or +1) is characterised by three independent

994: helicity flip amplitudes. These may be extracted from measurements of

995: the angular distribution of the vector meson decay

996: products~\cite{Chekanov:2002rm}. In lowest order QCD, $N = 4$ for transversely

997: polarized mesons, and $N = 3$ for longitudinally polarized mesons. For

998: massless quarks this helicity amplitude is suppressed, since the

999: photon only couples to $q\bar{q}$ pairs of even chirality. 

1000: 

1001: However, there are approximations involved at this stage since

1002: for light quarks the formation of the $q\bar{q}$ dipole cannot be

1003: perturbatively calculated, and nor can the vector meson wave

1004: function. With some flexibility introduced to allow for such effects,

1005: the calculations can also describe the $|t|$ dependence of the

1006: data~\cite{Poludniowski:2003yk}, as well as the helicity structure of $J/\psi$

1007: production (where the charm mass adds reliability to the perturbative

1008: approximations). They still fail to describe the helicity structure of

1009: $\rho$ and $\phi$ production. Nevertheless, the agreement with the

1010: cross-section measurements is further evidence that some understanding

1011: of diffraction has been obtained within QCD.

1012: 

1013: High-$t$ photon production shares the advantages of vector meson

1014: production, with the added advantage that the hard subprocess is

1015: completely calculable in pQCD - there is no need for a Vector Meson

1016: form factor. The disadvantage is that the cross-section is

1017: smaller. However, given the enhancements expected from the large $s/t$

1018: logarithms, the process was predicted to be observable at

1019: HERA~\cite{Ginzburg:1996vq,*Evanson:1999zb,*Ivanov:1998jw,*Cox:1999kv,*Hoyer:2000mb},

1020: and has indeed been observed~\cite{beckingham,*Favart:2003kw},

1021: providing further compelling evidence, from photoproduction, for the

1022: understanding of diffraction within QCD.

1023: 

1024: Understanding hard colour singlet exchange is not only of great

1025: theoretical importance for our understanding of high energy QCD, but

1026: is of importance for critical measurements at hadron colliders. In

1027: particular, both forward ({\it i.e.} high rapidity) jet rates and a

1028: suppression of QCD radiation between jets will play a key part in

1029: measurements of $WW$ scattering (see for

1030: example~\cite{atlastdr,*atlasptdr,*cmstdr} and references therein). This

1031: is an important search channel for the Standard Model Higgs, and is

1032: also the process where perturbative unitarity is violated within the

1033: LHC energy range in the absence of a light Higgs.

1034: 

1035: Photoproduction has not has its final say on this topic. Preliminary

1036: jet and $J/\psi$ data with higher statistics have already been

1037: presented~\cite{ZEUS-prel-04-001,*ZEUS-prel-03-018}. In addition,

1038: there is a possibility of extending the rapidity reach of the

1039: jet measurement by using the ``forward plug

1040: calorimeter''\cite{Bamberger:1999xz,Butterworth:1995jf} which was

1041: installed during part of HERA I data taking at ZEUS.

1042: 

1043: \subsubsection{Forward rapidity gaps and forward particles}

1044: 

1045: The other class of rapidity gap events observed in photoproduction is

1046: that in which the gap is not within the jet system, but is between the

1047: jets and the forward (proton) direction. In fact, in these cases the

1048: proton remains intact, or dissociates into a low-mass baryonic

1049: state. Thus, an alternative signature for these processes is a tag of

1050: a fast forward going proton or neutron. In these events, the moment

1051: transferred across the gap is small, and so the coupling between the

1052: exchanged colour singlet and the proton is not perturbatively

1053: calculable. However, in a subset of events there can be a large

1054: momentum scale involved in the scattering between the photon and the

1055: exchanged colour singlet, and thus partonic language may be useful for

1056: discussing some aspects of the event. In particular, one can assume

1057: factorisation, and discuss conditional parton distributions in the

1058: proton, {\it e.g.} diffractive parton distributions.  Often a further

1059: assumption is made in which the exchange is treated as a particle with

1060: a momentum distribution and its own parton densities (Regge

1061: factorisation for diffraction, or the one pion exchange model for

1062: forward neutron production). While these models are

1063: known~\cite{Chekanov:2005vv}, or expected, to fail in detail, they

1064: remain a useful approximation over a wide range of phase space.

1065: 

1066: The distribution of the momentum fraction $x_L$ for forward protons

1067: and neutrons is shown in Fig.~\ref{fig:xl}. In the case of protons, a

1068: clear peak due to diffractive processes is seen at $x_L \approx 1$. In

1069: general, those processes where the proton or low-mass dissociated

1070: system carries more than around 99\% of the initial proton energy are

1071: dominated by diffraction (this may even be seen as a definition of

1072: diffraction). In the case of neutrons, no such diffractive peak is

1073: seen, though the distribution does peak at lower values, around $x_L =

1074: 0.7$, as expected from single pion exchange models.

1075: 

1076: \begin{figure}[htp]

1077: \begin{center}

1078: ~\epsfig{file=Figures/ln_lp.eps,height=8cm}

1079: \put(-240,176){\makebox(0,0)[tl]{\large (a)}}

1080: \put(-35,176){\makebox(0,0)[tl]{\large (b)}}

1081: \caption[*]{\label{fig:xl} The distribution $x_L$, of fraction of the

1082: incoming proton momentum carried by a tagged forward 

1083: (a) neutron~\cite{Chekanov:2002pf} 

1084: and (b) proton~\cite{epj:c2:237}. Note the different horizontal scales!}

1085: \end{center}

1086: \end{figure}

1087: 

1088: Of particular interest is the use of photoproduction to study

1089: factorisation and the breaking of factorisation in diffractive

1090: processes. There is a factorisation theorem for diffraction in

1091: DIS~\cite{Collins:1997sr,*Collins:2001ga}; that is, one can define and

1092: use a set of diffractive parton densities for the proton and use them

1093: in different diffractive DIS processes as one would with the normal

1094: inclusive PDFs. However, this explicitly does not apply in

1095: hadron-hadron interactions because of the possibility of soft

1096: rescattering between the hadrons. At least naively then, one might

1097: expect direct photoproduction to exhibit factorisation, since there is

1098: only one incoming hadron, and resolved photoproduction to exhibit

1099: factorisation breaking, since it looks like a hadron-hadron collision.

1100: It is important to remember, however, that the terms resolved and

1101: direct photoproduction are short hand for kinematic extremes, the no

1102: unique separation between the two processes is possible (see

1103: Section~\ref{sec:dual}). Specifically then, a lower fraction of

1104: diffractive processes might be expected at low $\xgo$ compared to high

1105: $\xgo$.

1106: 

1107: Dijet photoproduction in association with a forward rapidity gap has

1108: been measured~\cite{cpaper:ichep2004:177,*cpaper:ichep2004:249} and

1109: compared to LO Monte Carlo models as well as NLO QCD

1110: calculations~\cite{Klasen:2001sg,np:b507:315,*np:b467:399,*np:b507:295}, 

1111: using diffractive PDFs extracted from DIS data. The LO Monte Carlos, which 

1112: in this case do not include any remnant-remnant interactions, describe the 

1113: shape of the data distributions well without any need for a rescattering 

1114: correction. However, in the NLO calculations, agreement is only seen if a 

1115: rescattering correction of around 0.5 is applied. Additionally, it seems 

1116: that this must be applied over all $\xgo$ values. The predicted 

1117: suppression~\cite{Kaidalov:2003xf} is around 0.34 for hadronic events. The 

1118: validity or not of QCD factorisation here is, however, complicated by the 

1119: uncertainty in the diffractive PDFs~\cite{dpdfs:levy,*dpdfs:newman} and 

1120: their application in dijet production in deep inelastic 

1121: scattering~\cite{cpaper:eps2005:342}. Fits to different data sets lead 

1122: to significantly different diffractive parton densities which when used 

1123: to calculate the diffractive dijet cross section in deep inelastic 

1124: scattering differ by a factor of two. Only by further understanding of 

1125: the inclusive diffractive data and the techniques used can this be resolved 

1126: and the question of factorisation breaking in jet production in hadron-hadron 

1127: collisions, photoproduction and deep inelastic scattering be addressed.

1128: 

1129: The leading neutron energy spectrum in photoproduction (as well as

1130: DIS) at HERA has been shown to be well described by single pion

1131: exchange~\cite{epj:c6:587,Chekanov:2002pf}. Some of the most

1132: interesting aspects of these interactions concern what is variously

1133: described as rescattering, absorption, or the multiple interaction

1134: probability.

1135: 

1136: \subsubsection{Survival and rescattering}

1137: 

1138: In Section~\ref{sec:ue}, underlying event models were discussed, with

1139: a particular emphasis on multiple hard scatters. However, soft

1140: scattering may also be eikonalised ({\it i.e.} repeated and

1141: exponentiated). For instance, multiple soft Pomeron exchange preserves

1142: unitarity in Regge based models which make no reference to partonic

1143: scattering. The natural ``soft'' counterpart of multiple parton

1144: scattering is soft rescattering, or absorption, which will affect the

1145: forward proton or neutron rate even if not enough momentum transfer is

1146: involved to throw particles into a detector and destroy a

1147: (pseudo)rapidity gap.

1148: 

1149: In hard multiple scattering models in hadron-hadron collisions, the

1150: presence of a hard scale is correlated to an increased probability of

1151: multiple interactions, since both are more likely in central

1152: collisions.

1153: 

1154: In inclusive photoproduction, where there is no hard scale, the photon

1155: is dominantly hadronic, and the forward neutron rate is expected to be

1156: reduced by soft rescattering. In DIS, two things are different. First

1157: there is a hard scale, and second, the photon is small, (that is,

1158: pointlike at the scale of the interaction). Obviously in DIS these are

1159: both due to the large photon virtuality. However, the statements are

1160: not equivalent.  Because it is pointlike, the photon has no remnant to

1161: undergo rescattering. The forward neutron rate is correspondingly

1162: higher, even though if the photon were considered as a hadron, one

1163: might expect a hard scale to increase rescattering.

1164: 

1165: In charm photoproduction~\cite{Chekanov:2004dk}, there is again a hard scale,

1166: provided by the charm mass. Some contribution from the hadronic photon

1167: is expected to be present, but this is suppressed with respect to the

1168: inclusive case, at least for inclusive dijet charm

1169: events~\cite{epj:c6:67}. There is no evidence for rescattering in these

1170: events, with the measured neutron fraction of $9 \pm 1\%$ being in

1171: good agreement with the DIS rate, and inconsistent with the rate for

1172: inclusive photoproduction. 

1173: 

1174: Finally one can consider dijet photoproduction. Here a hard scale is

1175: present, but one can select between hadronic and pointlike photons

1176: using the $\xgo$ variable.  Measurements of dijet photoproduction as a

1177: function of $\xgo$ help clarify the situation. All jet cross-sections

1178: self-evidently involve a hard scale, but the $\xgo$ variable allows

1179: the ``size'' of the photon to be deconvoluted from this, in that

1180: larger, hadronic photons are more likely to give rise to events with

1181: low $\xgo$. The H1 measurement~\cite{epj:c41:273} of the ratio of

1182: forward-neutron-tagged to inclusive dijet photoproduction in 

1183: Fig.~\ref{fig:ln_jets} confirms

1184: that, independent of the scale, there is a lower neutron rate in

1185: resolved photoproduction events, as expected from rescattering models

1186: in hadronic collisions.

1187: \begin{figure}[ht]

1188: \begin{center}

1189: ~\epsfig{file=Figures/d04-247f8a.eps,height=6.8cm}

1190: \put(-27,182){\makebox(0,0)[tl]{\large (a)}}

1191: \put(168,182){\makebox(0,0)[tl]{\large (b)}}

1192: ~\epsfig{file=Figures/d04-247f8b.eps,height=6.8cm}

1193: ~\epsfig{file=Figures/d04-247f8c.eps,height=6.8cm}

1194: \put(-27,182){\makebox(0,0)[tl]{\large (c)}}

1195: \put(168,182){\makebox(0,0)[tl]{\large (d)}}

1196: ~\epsfig{file=Figures/d04-247f8d.eps,height=6.8cm}

1197: \end{center}

1198: \caption{The ratio of the cross section for dijet photoproduction with a

1199:          leading neutron to that for inclusive dijet photoproduction as 

1200:          a function of (a) $E_T^{\rm jet}$, (b) $\eta^{\rm jet}$, 

1201:          (c) $x_\gamma^{\rm jet}$ and (d) $x_p^{\rm jet}$. The Monte Carlo 

1202:          predictions are shown for {\sc Pythia} for inclusive production 

1203:          and either {\sc Pythia} or {\sc Rapgap} with one-pion exchange 

1204:          for neutron production.}

1205: \label{fig:ln_jets}

1206: \end{figure}

1207: 

1208: One major source of interest in the measurements of forward proton and

1209: neutron production, and the understanding of rescattering effects, is

1210: that exclusive diffractive production may be an important search

1211: channel for new physics at

1212: LHC~\cite{Schafer:1990fz,*Bialas:1991wj,*Lu:1994ys,*Cudell:1995ki,*Khoze:1997dr,*Khoze:2000cy,*Khoze:2001xm,*DeRoeck:2002hk,*Cox:2003xp,*Boonekamp:2004nu,*Cox:2005if,*Boonekamp:2005yi}. The

1213: phenomenological predictions require a good understanding of

1214: diffractive processes, particularly diffractive PDFs and factorisation

1215: breaking.  Comparison to high energy photoproduction data, as well as

1216: to $p\bar{p}$ data, is a critical factor in developing and

1217: demonstrating this understanding.

1218: 

1219: