1: \documentclass[12pt]{article}
2:
3: \baselineskip = 14pt
4: \textwidth = 6.0in
5: \textheight = 8.5in
6: \topmargin -0.25truein
7: \oddsidemargin 0.30truein
8: \evensidemargin 0.30truein
9: \raggedbottom
10: \def\beq{\begin{equation}} \def\eeq{\end{equation}}
11: \def\bea{\begin{eqnarray}} \def\eea{\end{eqnarray}} \def\nn{\nonumber}
12: \def\noi{\noindent}
13: \def\lsim{\raise0.3ex\hbox{$<$\kern-0.75em\raise-1.1ex\hbox{$\sim$}}}
14: \def\gsim{\raise0.3ex\hbox{$>$\kern-0.75em\raise-1.1ex\hbox{$\sim$}}}
15: %\renewcommand{\thefootnote}{\fnsymbol{footnote}}
16:
17:
18: %\usepackage{axodraw}
19: \usepackage{epsfig}
20: \usepackage{wrapfig}
21:
22: \begin{document}
23: \setlength{\unitlength}{1mm}
24:
25: \begin{titlepage}
26:
27: \begin{flushright}
28: LPT-Orsay 03-99\\
29: IPPP/03/76\\
30: DCPT/03/152\\
31: \end{flushright}
32: \vspace{1.cm}
33:
34: \begin{center}
35: \vbox to 1 truecm {}
36: {\large \bf Isolated photon + jet photoproduction as a tool to
37: constrain the gluon distribution in the proton and the photon}
38: \par \vskip 3 truemm
39: \vskip 1 truecm {\bf M. Fontannaz$^{(a)}$, G. Heinrich$^{(b),}$\footnote{Address
40: after December 1, 2003: II. Institut f\"ur Theoretische Physik,
41: Universit\"at Hamburg,
42: Luruper Chaussee 149,
43: 22761 Hamburg, Germany.}}
44: \vskip 3 truemm
45:
46: {\it $^{(a)}$ Laboratoire de Physique Th\'eorique, UMR 8627 CNRS,\\
47: Universit\'e
48: Paris XI, B\^atiment 210, 91405 Orsay Cedex, France}\\
49:
50: \vskip 3 truemm
51:
52: {\it $^{(b)}$ IPPP, Department of Physics, University of Durham, \\Durham DH1
53: 3LE, England}
54:
55: \vskip 2 truecm
56:
57: \normalsize
58:
59: \begin{abstract}
60: We analyse how the reaction $\gamma \,p \to \gamma$ + jet + X
61: can serve to constrain the gluon distributions.
62: Our results are based on a code of partonic event generator type
63: which includes full NLO corrections.
64: We conclude that there are phase space domains in
65: which either the gluon in the photon or the gluon in the proton give
66: important contributions to the cross section, which should be
67: observable in HERA experiments.
68: \end{abstract}
69:
70: \vspace{3cm}
71:
72: \end{center}
73:
74: \end{titlepage}
75:
76: \section{Introduction}\label{intro}
77:
78:
79: Over the past years, the
80: ZEUS~\cite{Breitweg:1999su,Chekanov:2001aq}
81: and H1~\cite{h1} collaborations
82: at HERA have been able to observe the photoproduction of
83: large-$p_{T}$ photons, and the comparisons of data with existing NLO QCD
84: predictions~\cite{Gordon:1997yt,Fontannaz:2001ek,Fontannaz:2001nq,Krawczyk:2001tz,Zembrzuski:2003nu}
85: appear successful. In photoproduction
86: reactions, a quasi-real photon, emitted at small angle from the
87: electron, interacts with a parton from the proton. The photon can
88: either participate directly in the hard scattering or be resolved
89: into a partonic system, in which case the parton stemming from the
90: photon takes part in the hard interaction. Therefore photoproduction is
91: a privileged reaction to measure or constrain the parton distributions
92: in the photon and in the proton. In this paper we shall investigate the
93: possibility to constrain both the gluon in the photon and the gluon in
94: the proton by looking at the production of a large-$p_{T}$ photon
95: and a jet.
96: With the aim of enhancing the contribution of processes involving
97: initial gluons, we will explore various kinematical domains in detail.
98: We shall show that there are kinematical configurations which are
99: dominated either by the gluon in the photon, or by the gluon in the
100: proton, and which should be accessible to experiment. \par
101:
102: The photoproduction of large-$p_{T}$ particles and jets has a long
103: story\,; it offers interesting tests of QCD and gives access to the
104: measurement of the initial state parton distributions and the final
105: state fragmentation functions (for a review, see e.g.~\cite{klasen}).
106: Reactions involving
107: large-$p_{T}$ jets and/or hadrons have been copiously observed
108: because of their large cross sections, whereas the production of prompt
109: photons has been measured only more recently and the statistical errors are
110: still rather large. However, this latter reaction has advantages with respect
111: to jet or hadron production. Indeed a large transverse momentum is
112: necessary to unambiguously define a jet and to avoid too large
113: hadronisation
114: and underlying events corrections. The theoretical predictions
115: also are subject to uncertainties due to scale
116: variations. All these effects are quite sizeable, even for
117: $E_{T}^{\rm jet} > 21$~GeV \cite{Adloff:2003nr}.
118: The photoproduction of
119: large-$p_{T}$ hadrons also suffers from sizeable theoretical uncertainties
120: coming from a large sensitivity to scale variations
121: and from the fragmentation functions which are not very accurately
122: measured \cite{Kniehl:2000hk,Fontannaz:2002nu}.
123: In the case of a large-$p_{T}$ photon, there
124: is of course no problem due to jet definition, hadronisation
125: or inaccurate fragmentation functions (only isolated photons are observed).
126: Even more importantly, the theoretical predictions are reasonably stable under
127: scale variations. Therefore the photoproduction of prompt photons
128: appears as an ideal reaction to test QCD and measure the
129: non-perturbative inputs. \par
130:
131: However one must keep in mind two reserves, one of experimental
132: and another one of theoretical nature. First, the measured cross sections
133: are small and become rapidly very low at peripheral regions
134: of the phase space which might be physically interesting.
135: Moreover, the detection of a photon
136: among the huge amount of large-$p_{T}$ $\pi^0$'s is not an easy task.
137: Second, only isolated photons are observed, the isolation criterion
138: being that very little hadronic energy is contained in a cone
139: surrounding the photon. This has the advantage of reducing the
140: fragmentation component of the cross section, but at the same time this
141: isolation can eliminate events containing too much
142: hadronic energy coming from the underlying event in the cone,
143: an effect which cannot be taken into account in the NLO calculations.
144: This point has been discussed in
145: \cite{Fontannaz:2001nq} and studied by the H1
146: collaboration \cite{h1,Lemrani:2003mj}.\par
147:
148: In order to constrain the kinematics of the different subprocesses,
149: it is important to observe a large-$p_{T}$ jet in association with
150: the photon, which introduces uncertainties in the comparison between
151: data and theory due to hadronisation and underlying event phenomena.
152: However, the effect of these phenomena can be considerably reduced if
153: the transverse momentum of the jet, which is not well measured, is not
154: used to constrain the kinematics, but only its rapidity. This is done
155: by using the variables $x_{LL}$ \cite{Aurenche:2000nc,Fontannaz:2001nq}
156: instead of the commonly used $x_{obs}$.
157: % initially defined by the ZEUS collaboration \cite{}.
158: We will make an extensive use of the variables $x_{LL}$ for the
159: proton and the photon in order to constrain
160: the kinematical region relevant for the observation of the gluon
161: distributions. \par
162:
163: Our paper is organised in the following way.
164: In Section 2, we discuss theoretical issues related to the subsequent
165: numerical studies,
166: such as photon isolation, suitable observables to study the parton
167: distributions and the importance of asymmetric cuts on the minimum transverse
168: energies. Section 3 contains the numerical results, where first the sensitivity
169: to the gluon content of the proton is studied. Then we turn to the
170: gluon distribution in the real photon before we conclude in Section 4.
171:
172:
173: \section{Theoretical framework}
174:
175:
176: As the general framework of the calculation already has been described in
177: detail in~\cite{Fontannaz:2001ek,Fontannaz:2001nq},
178: we will sketch the method only briefly here and focus instead on
179: issues related to the gluon distributions.
180:
181: In photoproduction, the electron acts like a source of
182: quasi-real photons whose spectrum can be described by the
183: Weizs\"acker-Williams approximation, which we use in the following form
184: \begin{equation}
185: f^e_{\gamma}(y) = \frac{\alpha_{em}}{2\pi}\left\{\frac{1+(1-y)^2}{y}\,
186: \ln{\frac{Q^2_{\rm max}(1-y)}{m_e^2y^2}}-\frac{2(1-y)}{y}\right\}\;.
187: \label{ww}
188: \end{equation}
189: As already mentioned, the quasi-real photon then either takes part
190: {\em directly} in the
191: hard scattering process, or it acts as a composite object, being a
192: source of partons which take part in the hard subprocess.
193: The latter mechanism is referred to as {\em resolved} process and
194: is parametrised by the parton distributions in the photon
195: $F_{a/\gamma}(x^{\gamma},Q^2)$. Thus the distribution of partons
196: of type "$a$" in the electron is a convolution
197: \begin{equation}
198: F_{a/e}(x_e,M)=\int_0^1 dy \,dx^{\gamma}\,f^e_{\gamma}(y) \,
199: F_{a/\gamma}(x^{\gamma},M)\,\delta(x^{\gamma}y-x_e)
200: \label{resolved}
201: \end{equation}
202: where in the "direct" case the parton $a$ is the photon itself, i.e.
203: $F_{a/\gamma}(x^{\gamma},M)=\delta_{a\gamma}\delta(1-x^{\gamma})$.
204: The parton distributions in the photon $F_{a/\gamma}(x,Q^2)$
205: behave like $\alpha/\alpha_s(Q^2)$ for large $Q^2$. Therefore
206: the additional power of $\alpha_s$ contained in the "resolved"
207: component as compared to the "direct" one is compensated.
208: This means that the NLO corrections to the resolved component
209: can be numerically sizeable and have to be taken into account.
210:
211: The cross section can symbolically be written as
212: \begin{eqnarray}
213: &&d\sigma^{e p \to \gamma \, j}(P_e,P_p,P_{\gamma},P_{j})=\nn\\
214: &&\sum_{a,b}\int dx_e\int d x_p\,
215: F_{a/e}(x_e,{M})F_{b/p}(x_p,{M})\{d\hat\sigma^{\rm{dir}}\;+
216: \;d\hat\sigma^{\rm{frag}}\}\label{sigma}\\
217: %\quad + {\small {\cal O}(1/Q^{\nu})}\\
218: &&\nn\\
219: &&d\hat\sigma^{\rm{dir}}=d\hat\sigma^{ab\to \gamma\,j}
220: (x_a,x_b,P_{\gamma},P_{j},\mu,M,M_F)\nn\\
221: &&d\hat\sigma^{\rm{frag}}=\sum_c\int dz\,D_{\gamma/c}(z,{M_F})
222: d\hat\sigma^{ab\to c\,j}
223: (x_a,x_b,P_{\gamma}/z,P_{j},\mu,{M},{M_F})\nn
224: \end{eqnarray}
225: where we have split the hard scattering cross sections $\hat\sigma$
226: explicitly into a "direct" and a "fragmentation" part in order to
227: point out that there are two contributions to the "prompt photon"
228: in the final state: The "direct" one, where the final state photon
229: is produced directly in the hard interaction, and the one where the
230: photon stems from the fragmentation of a large-$p_{T}$
231: quark or gluon in the final state.
232: This fragmentation process is described by
233: the fragmentation functions $D_{\gamma/c}(z,{M_F})$.
234: At next-to-leading order,
235: the distinction between "direct" and "fragmentation" becomes
236: scheme dependent because the final state collinear singularity
237: appearing in a
238: "direct" process like $\gamma g \rightarrow \gamma q\bar q$
239: when a quark becomes collinear to the photon is
240: absorbed at the fragmentation scale $M_F$ into the "bare" fragmentation
241: functions, and where to attribute the finite parts is a matter of
242: choice of the factorisation scheme. In our calculation we use the
243: ${\overline{\rm MS}}$ scheme.
244:
245: Note that the cross section (\ref{sigma}) depends on three scales,
246: the renormalisation scale $\mu$, the initial state factorisation scale $M$,
247: and the fragmentation scale $M_F$, and that it can be considered as
248: consisting of 4 categories of subprocesses, depending on whether there is a
249: "direct" photon in the initial and/or final state: 1. direct direct,
250: 2. resolved direct, 3. direct fragmentation, 4. resolved fragmentation.
251: Each of these contributions consists of several partonic subprocesses
252: at NLO and is strongly scale dependent. Only in the sum this
253: scale dependence cancels to a large extent, and only the sum
254: can be considered as a physical quantity. This has to be kept in mind
255: when the contributions of certain subprocesses only will be considered
256: below, in order to estimate the contribution of gluon initiated processes.
257: The study of particular subprocesses can be very useful to
258: get an idea of the underlying parton dynamics, but cannot be considered
259: as precise quantitative statements because of the scale and scheme
260: dependence outlined above.
261:
262: We would like to emphasize that we calculated the full NLO
263: corrections to all four categories of subprocesses.
264: We also included the quark loop box contribution $\gamma g\to \gamma g$
265: which is NNLO from a naive $\alpha_s$ power counting point of view,
266: but as the process $\gamma g\to \gamma g$ does not exist
267: at tree level, the box is the "leading order" diagram
268: for this process, and its numerical contribution is quite
269: sizeable~\cite{Fontannaz:2001ek}.
270: %In \cite{Krawczyk:2001tz,Zembrzuski:2003nu}, the asymptotic behaviour
271: %$\sim 1/\alpha_s(Q^2)$ of the photon parton distributions
272: %and of the photon fragmentation functions is not considered as
273: %a compensation of the factor $\alpha_s$ present in the LO
274: %resolved respectively fragmentation contributions.
275: The calculation presented in \cite{Krawczyk:2001tz,Zembrzuski:2003nu}
276: has no higher order corrections to the resolved-direct, direct-fragmentation
277: and resolved-fragmentation contributions. It
278: only contains the higher order corrections to the
279: direct-direct part, and the box contribution is also included.
280:
281:
282: \subsection*{Photon Isolation}
283: In prompt photon measurements, the experimental challenge consists in
284: the separation of prompt photon events from the large background of
285: secondary photons produced by the decay of light mesons, predominantly
286: $\pi^0$ mesons. The latter, when produced at high energy, decay into two
287: almost collinear photons which cannot be resolved in the calorimeter.
288: However, they are in general accompanied by hadronic energy
289: and thus this background can be suppressed by isolation cuts.
290:
291: Commonly a cone isolation criterion is used, defined in the following
292: way\footnote{A more sophisticated criterion has been
293: proposed in \cite{frixione}, in which the veto on accompanying
294: hadronic transverse energy is the more severe, the closer the corresponding
295: hadron is to the photon direction. It has been designed to make the
296: fragmentation contribution vanish completely, in an infrared safe way,
297: but is less straightforward to implement experimentally.}:
298: A photon is isolated if, inside a cone centered around the photon direction
299: in the rapidity and azimuthal angle plane, the amount of hadronic transverse
300: energy
301: $E_T^{had}$ deposited is smaller than some value $E_{T,\rm{max}}$ fixed by the
302: experiment:
303: \begin{equation}\label{criterion}
304: %\left.
305: \begin{array}{rcc}
306: \left( \eta - \eta^{\gamma} \right)^{2} + \left( \phi - \phi^{\gamma} \right)^{2}
307: & \leq & R_{\mathrm{exp}}^{2} \\
308: E_T^{had} & \leq & E_{T,\rm{max}}\;.
309: \end{array}
310: %\right\}
311: \end{equation}
312: Following the HERA conventions, we used
313: $E_{T,\rm{max}}=\epsilon \,p_T^{\gamma}$ with
314: $\epsilon=0.1$ and $R_{\mathrm{exp}}$ = 1.
315: Of course isolation not only reduces the background from
316: secondary photons, but also
317: substantially reduces the fragmentation components, such that the total cross
318: section depends very little on the fragmentation functions.
319:
320: But another, undesired, effect of isolation is a partial suppression of
321: the direct contribution. Indeed, the hadronic transverse energy
322: $E_T^{had}$ deposited in the cone may stem from the soft underlying event
323: due to the spectator-spectator collisions. This renders the effective
324: isolation cut more stringent and leads to a decrease of the cross section,
325: included the direct contributions. The simulation of this effect, discussed
326: in \cite{Fontannaz:2001ek}, requires a good knowledge of the hadron
327: distributions in the underlying event, which is asymmetric in $\gamma\,p$
328: collisions\,; it has been studied in recent H1
329: publications \cite{h1,Lemrani:2003mj}.
330:
331:
332: \subsection{Suitable observables to study the parton distributions}
333:
334: As observables which serve to reconstruct the longitudinal
335: momentum fraction of the parton stemming from the
336: proton respectively photon, it is common to use
337: \beq
338: x_{obs}^{\rm{p}}=\frac{p_T^\gamma\,e^{ \eta^\gamma}+
339: E_T^{\rm jet}\,e^{\eta^{\rm jet}}}{2E^{\rm{p}}}\;,\;
340: x_{obs}^{\gamma}=\frac{p_T^\gamma\,e^{- \eta^\gamma}+
341: E_T^{\rm jet}\,e^{-\eta^{\rm jet}}}{2E^{\gamma}} \;.
342: \label{xobs}
343: \eeq
344: However, as the measurement of $E_T^{\rm jet}$ can be a
345: substantial source of systematic errors at low $E_T$ values,
346: we propose a slightly different variable which does not depend on
347: $E_T^{\rm jet}$,
348: \beq
349: x_{LL}^{\rm{p},\gamma}=\frac{p_T^\gamma\,(e^{\pm \eta^\gamma}+e^{\pm
350: \eta^{\rm jet}})}{2E^{\rm{p},\gamma}}\;.
351: \label{xll}
352: \eeq
353: At leading order, for the non-fragmentation contribution, the variables
354: $x_{obs}$ and $x_{LL}$ coincide, and they are also equal to the
355: "true" partonic longitudinal momentum fraction, i.e. the argument
356: of the parton distribution function.
357: At NLO, the real corrections involve 3 partons in the final state (with
358: transverse momenta $p_{T3}, p_{T4}, p_{T5}$), one of
359: which -- say parton 5 -- is unobserved. Therefore, $x_{obs}$ and $x_{LL}$
360: will be different at NLO.
361:
362: %\begin{figure}[htb]
363: \begin{wrapfigure}[20]{r}[0.cm]{10.7cm}
364: \vspace*{-1.3cm}
365: %\begin{center}
366: \epsfig{file=xobsll.eps,height=10.7cm}
367: %\end{center}
368: \caption{Comparison of $x_{LL}$ and $x_{obs}$ for the proton and the photon.
369: The photon and jet rapidities and transverse energies have been integrated
370: in the range
371: $-2\leq \eta^{\gamma,{\rm jet}}\leq 4$, $p_{T}^{\gamma}\geq 6$\,GeV,
372: $E_{T}^{\rm jet}\geq 5$\,GeV, $\sqrt s=318$\,GeV.}
373: \label{xobsll}
374: \end{wrapfigure}
375: The difference between $x_{LL}$ and $x_{obs}$ is
376: rather small in the proton case, as shown in Fig.~\ref{xobsll}\,a).
377: In the photon case, $x_{LL}^\gamma$ and $x_{obs}^\gamma$
378: are very similar in the region $0\,\lsim \,x^\gamma\,\lsim \,0.85$.
379: However, for $x^\gamma$ close to one, there are important differences
380: between $x_{LL}^\gamma$ and $x_{obs}^\gamma$, the former leading
381: to a smoother distribution $d\sigma/dx^\gamma$ if the size of the bins around
382: $x^\gamma=1$ is not chosen too small
383: (see Fig.~\ref{xobsll}\,b).
384:
385: \clearpage
386:
387:
388: \subsection{Asymmetric cuts}
389:
390: It is well
391: known\,\cite{Frixione:1997ks, Aurenche:2000nc, Fontannaz:2001nq} that
392: symmetric cuts on the minimum transverse energies of dijets or a photon plus
393: a jet should be avoided as they amount to including
394: a region where the fixed order
395: perturbative calculation shows infrared sensitivity.
396: As explained in detail in~\cite{Fontannaz:2001nq}, the problem stems
397: from terms $\sim \log^2(|1-p_T^{\gamma}/E^{\rm jet}_{T,\rm{min}}|)$
398: which become large as $p_T^{\gamma}$ approaches $E^{\rm jet}_{T,\rm{min}}$,
399: the lower cut on the jet transverse energy.
400: Therefore the partonic NLO cross section has a singular behaviour at
401: $p_{T}^{\gamma}=E^{\rm jet}_{T,\rm{min}}$, which is displayed in
402: Fig.~\ref{etmin}. Of course, analogously, there are
403: $ \log^2(|1-E_T^{\rm jet}/p^{\gamma}_{T,\rm{min}}|)$ terms which become
404: large for $E_T^{\rm{jet}}\to p^{\gamma}_{T,\rm{min}}$, see Fig.~\ref{largebin}.
405: %the same is true if we fix
406: %$E^{\gamma}_{T,\rm{min}}$ and consider the limit
407: %$E_T^{\rm{jet}}\to E^{\gamma}_{T,\rm{min}}$.
408:
409: \begin{wrapfigure}[20]{r}[0.cm]{9cm}
410: \vspace*{-1.cm}
411: %\begin{figure}[htb]
412: %\begin{center}
413: \mbox{\epsfig{file=jetmin7.eps,height=9cm}}
414: %\end{center}
415: \caption{Logarithmic singularity in $d\sigma/dp_{T}^{\gamma}$ at
416: $p_{T}^{\gamma}=E^{\rm jet}_{T,\rm{min}}$. }
417: \label{etmin}
418: \end{wrapfigure}
419: The comparison with data can be done in two ways.
420: First, if one wants to display e.g. the differential cross section
421: $d\sigma/dE_T^{\rm{jet}}$ while
422: $p_{T,\rm{min}}^{\gamma}$ lies within the considered $E_T^{\rm{jet}}$ range,
423: the binning in
424: $E_T^{\rm{jet}}$ must be chosen large enough to average over the logarithmic
425: singularity which is integrable.
426: For instance, the binning
427: $E_T^{\rm{jet}}-\Delta\leq p_{T,\rm{min}}^{\gamma}\leq E_T^{\rm{jet}}+\Delta$
428: with $\Delta=0.5$\,GeV for the bin around $p_{T,\rm{min}}^{\gamma}=7$\,GeV
429: in Fig.~\ref{largebin} should lead to a correct
430: average of the theoretical singularity and allow for a comparison with data.
431: Obviously, $d\sigma/dp_T^{\gamma}$ will not exhibit a problem as long as
432: $E_{T,\rm{min}}^{\rm{jet}}<p_{T,\rm{min}}^{\gamma}$ since the critical point
433: $p_{T}^{\gamma}=E_{T,\rm{min}}^{\rm{jet}}$ will not be reached in this case.
434:
435: Second, one often would like to have a more inclusive cross section such as
436: $d\sigma/d\eta^\gamma$, obtained by integrating the differential cross
437: section over $p_T^{\gamma}$ and $E_T^{\rm{jet}}$.
438: In this case one should not choose
439: $p_{T,\rm{min}}^{\gamma}=E^{\rm jet}_{T,\rm{min}}$ as this amounts to
440: integrating the spectrum of Fig.~\ref{largebin} to the right-hand side of
441: $p_{T,\rm{min}}^{\gamma}$ only and thus to picking up only the singular negative
442: contribution to the NLO cross section, without the compensation coming
443: from the positive contribution to the left of $p_{T,\rm{min}}^{\gamma}$.
444: As a result, the theoretical prediction, although being finite,
445: is infrared sensitive as a consequence of choosing symmetric cuts.
446: This point has been discussed in detail in ref.~\cite{Fontannaz:2001nq}.
447: \clearpage
448:
449: \begin{wrapfigure}[19]{r}[0.cm]{8.cm}
450: \vspace*{-0.8cm}
451: \epsfig{file=etcuts_largebin.eps,height=8cm}
452: \caption{The logarithmic singularity in $d\sigma/dE_{T}^{\rm jet}$ at
453: $E^{\rm jet}_{T}=p_{T,\rm{min}}^{\gamma}$ is averaged over by choosing
454: the binning large enough.}
455: \label{largebin}
456: \end{wrapfigure}
457:
458: It is also illustrated in Fig.~\ref{etajet},
459: where we consider the cross section $d\sigma/d\eta^{\rm jet}$ in a kinematic
460: range studied by H1.
461: To exhibit the effect of different cuts on $E^{\rm jet}_{T}$,
462: we vary $E^{\rm jet}_{T,\rm{min}}$ to take the values 4, 4.5 or 5 GeV,
463: while $p_{T,\rm{min}}^{\gamma}$ has been fixed to 5 GeV, and display
464: direct and resolved parts separately.
465: %\footnote{
466: Note that the leading order prediction is the same
467: for all three values of $E^{\rm jet}_{T,\rm{min}}$ as $E^{\rm jet}_{T}$
468: cannot become smaller than $p_{T,\rm{min}}^{\gamma}$ at leading order.
469: One observes that in the
470: direct part at small rapidities, the higher order corrections are
471: large and negative.
472: In the case of symmetric cuts, they are even negative in the resolved part
473: at small rapidities, and the theoretical prediction depends strongly
474: on the small change in $E^{\rm jet}_{T,\rm{min}}$ from
475: 5 GeV to 4.5 GeV, whereas away from the symmetric cut region, the
476: prediction is much more stable under small changes of
477: $E^{\rm jet}_{T,\rm{min}}$.
478: %This illustrates once again the importance of avoiding symmetric cuts
479:
480: \begin{wrapfigure}[20]{l}[0.cm]{10.2cm}
481: \vspace*{-0.5cm}
482: \epsfig{file=h1etajet.eps,height=10.2cm}
483: \caption{Magnitude of higher order corrections to direct/resolved parts separately}
484: \label{etajet}
485: \end{wrapfigure}
486:
487: In summary, we repeat that for a successful comparison of data and NLO theory,
488: one has to either ensure to stay away from IR singular domains or to consider
489: suitably averaged observables.
490: In what concerns the cuts on $p_{T}^{\gamma}$ and $E_{T}^{\rm jet}$,
491: this amounts to choosing asymmetric cuts adapted to the observable.
492: Therefore we disagree with the statement made in~\cite{Chyla:2003in} that
493: asymmetric cuts are not superior to symmetric ones.
494:
495:
496: \clearpage
497:
498: \section{Numerical results}
499:
500: Our studies are based on the program
501: {\tt EPHOX}\footnote{The program together with detailed documentation
502: is available at\\
503: http://wwwlapp.in2p3.fr/lapth/PHOX\_FAMILY/main.html.}, which is a
504: partonic Monte Carlo event generator.
505: Unless stated otherwise, we use the following input for our numerical
506: results:
507: A center of mass energy $\sqrt{s}=318$\,GeV with $E_e=27.5$\,GeV and
508: $E_p=920$\,GeV is used. The cuts on the minimum transverse energies
509: of photon and jet are $E_T^{\rm jet}>5$\,GeV, $p_T^{\gamma}>6$\,GeV.
510: The rapidities have been
511: integrated over in the domain $-2\leq \eta^\gamma,\eta^{\rm jet}\leq 4$
512: unless stated otherwise.
513: For the parton distributions in the
514: proton we take the MRST01~\cite{mrst01} parametrisation, for the photon we
515: use AFG04\footnote{This parametrisation has been used in
516: ref.~\cite{Fontannaz:2002nu} under the name AFG02.}~\cite{Aurenche:1994in}
517: distribution functions and BFG~\cite{Bourhis:1998yu} fragmentation functions.
518: We take $n_f=4$ flavours, and for $\alpha_s(\mu)$ we use an exact
519: solution of the two-loop renormalisation group
520: equation, and not an expansion in log$(\mu/\Lambda)$.
521: The default scale choice is $M=M_F=\mu=p_T^{\gamma}$.
522: Jets are defined using the $k_T$-algorithm~\cite{ktalgo}.
523: The rapidities refer to the $e\,p$
524: laboratory frame, with the HERA
525: convention that the proton is moving towards positive rapidity.
526:
527:
528: \subsection{The gluon distribution in the proton}
529:
530: As explained already in Section \ref{intro}, an accurate knowledge of
531: the gluon distribution in the proton is very important
532: at the LHC because of its large gluon luminosity.
533: At present the error on important cross sections at the LHC stemming
534: from the gluon pdfs is about 5-7\%, but can be up to 20\%
535: in certain cases.
536: Therefore our aim is to find a region where
537: 1) the sensitivity to the gluon in the proton is enhanced,
538: 2) $x^{\rm p}$ is rather large,
539: 3) the uncertainty stemming from the poorly known gluon in the {\it photon}
540: is minimised.
541:
542: \subsection*{Cut on $x^\gamma$}
543: Requirement 3) can be assured most easily by imposing a lower cut
544: $x^\gamma_{min}$ on $x^\gamma$
545: because at small $x^\gamma$ the gluon in the photon is large.
546: On the other hand, large values of $x^\gamma$, corresponding
547: mainly to direct initial state photons, correspond to small
548: values of $x^{\rm p}$ at a fixed $p_T$ value, according to eq.~(\ref{xll}).
549: Therefore 2) can be achieved by a cut $x^\gamma_{max}$ on $x^\gamma$.
550: In prompt photon production, the contribution from the process
551: $\gamma\,(direct) + g^p \to \gamma\,(direct)\,+$\,jet is rather small
552: anyway because this process does exist only at NLO.
553: Therefore the subprocess $q^\gamma+g^p \to \gamma\,+$\,jet is the one
554: which should dominate in the region fulfilling the requirements 1)\,--\,3),
555: and our aim is to enhance the contribution of this subprocess.
556: To this aim we investigated how cuts on $x^\gamma$ act in this respect,
557: and found that the cut $0.05< x_{LL}^{\gamma} <0.95$
558: maximally enhances the sensitivity to the gluon in the proton while
559: keeping the contribution from $g^\gamma$ negligible,
560: as shown in Fig.~\ref{figcutxga}.
561: Note that the contributions `$g^p$ only' and `$g^\gamma$ only' are not
562: disjunct, they both contain the subprocess $g^p+g^\gamma\to \gamma+$\,jet.
563: This subprocess does not exist at leading order, but beyond LO its
564: contribution is non-zero and in fact negative.
565: This can clearly be seen in Fig.~\ref{figcutxga}:
566: The cut $0.05< x^{\gamma}_{LL} <0.95$ actually
567: enhances the total value of the $g^p$-initiated part of the cross
568: section, because the lower limit $0.05< x^{\gamma}_{LL}$ removes mainly
569: the $g^p+g^\gamma$ initiated part.
570: %to a large extent.
571: The fact that the $g^p+g^\gamma$ contribution
572: is negative also reminds us that the
573: individual subprocesses are unphysical, such that these considerations
574: can be viewed only as qualitative reflections of the
575: underlying parton dynamics.
576:
577: \begin{figure}[htb]
578: \begin{center}
579: \epsfig{file=xp_cutxga_glull.eps,height=12.5cm}
580: \end{center}
581: \caption{Effect of a cut on $x^\gamma_{LL}$ to enhance the contribution of
582: $g^p$ initiated subprocesses }
583: \label{figcutxga}
584: \end{figure}
585:
586: Fig.~\ref{figsubeta} shows in detail, as a function of the photon rapidity,
587: how the requirement
588: $0.05< x^{\gamma}_{LL} <0.95$ suppresses the direct photon contribution
589: and enhances the relative importance of the subprocess
590: $g^p+q^\gamma\to \gamma+$\,jet, especially in the region
591: $\eta^{\gamma}\,\lsim \,1$.
592: %The $\gamma_{direct}g^p$ contribution being negative in the
593: %backward region, we again see an {\it enhancement} after the cut.
594: \begin{figure}[htb]
595: \begin{center}
596: \mbox{\epsfig{file=sub_eta.eps,height=12.5cm}}
597: \end{center}
598: \caption{ Relative importance of different subprocesses as a function of
599: photon rapi\-dity. The contribution from the direct photon gets suppressed
600: by the cut $x^{\gamma}_{LL} <0.95$\,. }
601: \label{figsubeta}
602: \end{figure}
603:
604:
605: \subsection*{Rapidity cuts}
606: Another possibility to enhance the contribution of the
607: $q^\gamma g^p$ initiated subprocesses is to impose rapidity cuts.
608: Whereas $x_{obs,LL}$ are variables where rapidity and energy measurements
609: enter, using
610: rapidity cuts only is very straightforward experimentally.
611: At small rapidities
612: the Compton process $\gamma \,q^p\to \gamma\, q$ dominates.
613: Further to the forward region, the importance of
614: $q^\gamma g^p$ initiated subprocesses increases, while at
615: very large rapidities the gluon in the photon also plays a role.
616: Therefore, one can also meet requirements 1) to 3) by restricting the
617: photon and jet rapidities to positive values. Fig.~\ref{xp_rapcuts}
618: shows that the relative contribution
619: of $g^p$-initiated processes increases from about 35\%
620: of the total in the full rapidity range
621: $-2<\eta^{\gamma,\rm{jet}}<4$ (Fig.~\ref{xp_rapcuts}\,a),
622: to about 48\% in the region $0<\eta^{\gamma,\rm{jet}}<4$,
623: while the contribution from the gluon
624: in the photon is still small, as can be seen from Fig.~\ref{xp_rapcuts}\,b.
625: Cutting further (e.g. $1<\eta^{\gamma,\rm{jet}}<4$) only reduces
626: the cross section substantially and introduces a larger
627: uncertainty from the gluon in the photon.
628:
629: \begin{figure}[htb]
630: \begin{center}
631: \mbox{\epsfig{file=xp_rapcuts_final.eps,height=12.5cm}}
632: \end{center}
633: \caption{Restricting the rapidities to positive values enhances the relative
634: contribution of $g^p$-initiated processes.}
635: \label{xp_rapcuts}
636: \end{figure}
637: Comparing the two methods, we find that the cut $0.05< x^{\gamma}_{LL} <0.95$
638: reduces the total cross section only by about 31\%,
639: enhancing the contribution from the gluon in the proton
640: from about 35\% of the total to 56\% of the total.
641: The rapidity cut $0<\eta^{\gamma,\rm{jet}}<4$ reduces the cross section by
642: about 70\% as compared to the range $-2<\eta^{\gamma,\rm{jet}}<4$.
643:
644:
645: %\clearpage
646:
647: \subsection*{Scale dependence}
648:
649: Fig.~\ref{xp_scal} shows that the NLO cross section
650: $d\sigma/dx^p_{LL}$ is very stable under scale changes.
651: %\begin{wrapfigure}[22]{r}[0.cm]{10cm}
652: \begin{figure}[htb]
653: \begin{center}
654: %\vspace*{-0.8cm}
655: \epsfig{file=xp_scal_ori.eps,height=13cm}
656: \caption{Scale dependence of the cross section $d\sigma/dx^p_{LL}$.}
657: \label{xp_scal}
658: \end{center}
659: \end{figure}
660: %\end{wrapfigure}
661: \newpage
662: In Fig.~\ref{mrs4}, we
663: show the predictions obtained with different
664: parametri\-sations of parton distribution functions for the proton,
665: where we chose the set CTEQ6M~\cite{cteq6}
666: and two different sets of MRST01~\cite{mrst01}, the default set and the set
667: MRST01J, which gives better agreement
668: with the Tevatron high-$E_T$ inclusive jet data due to a "bump" in the gluon
669: distribution at large $x$.
670:
671: %\begin{wrapfigure}[20]{r}[0.cm]{13cm}
672: \begin{figure}[htb]
673: \begin{center}
674: \epsfig{file=mrs4.eps,height=13cm}
675: \caption{Dependence of the cross section $d\sigma/dx^p_{LL}$ on different
676: parton distribution functions for the proton.}
677: \label{mrs4}
678: \end{center}
679: \end{figure}
680: %\end{wrapfigure}
681: Comparing Figs.~\ref{xp_scal} and \ref{mrs4}, we notice
682: that the differences in the cross sections due to different
683: parton distribution functions are of the
684: order of the variations due to the scale changes.
685: However, this situation can be somewhat improved:
686: Fig.~\ref{pdfscut} shows that the cut $0.05< x_{LL}^{\gamma} <0.95$
687: makes the differences between various
688: parametrisations more pronounced
689: (as it enhances the gluon initiated
690: contribution to the cross section),
691: especially in the region
692: $x_{LL}^{\rm p}\,\lsim\, 0.02$.
693: On the other hand, the variation
694: of the cross section due to scale changes
695: in the region $x_{LL}^{\rm p}\,\lsim\, 0.015$
696: %of the order of the numerical error, so
697: is not increased
698: by the presence of the cut on $x_{LL}^{\gamma}$.
699: Therefore, the reaction
700: $\gamma \,{\rm p}\to \gamma$\,+\,jet\,+\,X could indeed
701: be useful to further constrain the gluon in the proton
702: in the range $x_{LL}^{\rm p}\,\lsim\, 0.015$.
703: However, it is also clear that
704: data with very high statistics are needed
705: to distinguish between different parametrisations.
706: \begin{figure}[b]
707: \begin{center}
708: \epsfig{file=xp_pdfs_cutxga.eps,height=14.5cm}
709: \caption{The cut $0.05< x_{LL}^{\gamma} <0.95$ enhances the gluon contribution
710: and thus the differences between the parton distribution functions
711: in the region $x_{LL}^{\rm p}\,\lsim\, 0.015$, while it does not
712: affect the stability with respect to scale changes.}
713: \label{pdfscut}
714: \end{center}
715: \end{figure}
716: \clearpage
717:
718: \subsection{The gluon content of the photon}
719:
720: The photoproduction of large-$p_T$ jets, hadrons and photons
721: are privileged reactions to explore the gluon content of the
722: resolved photon, which is hardly observable in $\gamma\,\gamma^*$ DIS.
723: However, as discussed in the introduction, the scale dependence of the hadron
724: and jet production cross sections is not negligible, whereas the photon cross
725: section is more stable. This fact should allow us a more accurate determination
726: of the gluon distribution in the photon, $g^\gamma(x^{\gamma},Q^2)$.
727: %\begin{wrapfigure}[23]{l}[0.cm]{10cm}
728: %\vspace*{-0.5cm}
729: \begin{figure}[htb]
730: \begin{center}
731: \epsfig{file=rapm2to4.eps,height=11.5cm}
732: \caption{Magnitude of different subprocesses over the full photon rapidity
733: range. The jet rapidities have been integrated over
734: $-2<\eta^{\rm jet}<4$, and $E_T^{\rm jet}>5$\,GeV,
735: $p_T^{\gamma}>6$\,GeV.}
736: \label{fulleta}
737: \end{center}
738: \end{figure}
739:
740: In Fig.~\ref{fulleta} we display the various contributions to the cross section
741: $d\sigma/d\eta^\gamma$. The gluon distribution $g^\gamma(x^{\gamma},Q^2)$ only contributes at
742: small values of $x^{\gamma}$, corresponding to large values of $\eta^\gamma$, and we
743: shall try, by various cuts, to enhance the re\-la\-tive contribution of this
744: component.
745:
746:
747: In Fig.~\ref{cutxp} we see that the direct contribution,
748: corresponding to $x_{LL}^{\gamma}$ close to one, does not screen
749: the contribution initiated by the gluon in the photon.
750: But at smaller values of $x_{LL}^{\gamma}$, the `background' coming from other
751: subprocesses, such as $q^{\gamma}g^p\to q g$, is large.
752: Exactly this fact has been exploited to enhance the gluon from the proton
753: by restricting $x_{LL}^{\gamma}$ to intermediate values,
754: see Fig.~\ref{figcutxga}.
755: Now we would like to enhance the gluon from the {\it photon}, and therefore we
756: impose a lower cut on $x_{LL}^{\rm p}$ in order to reduce the
757: contributions from the gluon in the proton.
758: However, this cut has no effect at very small values of
759: $x_{LL}^{\gamma}$, where the gluon in the photon is most visible.
760: Therefore, contrary to the situation for the gluon in the proton
761: treated in the previous subsection,
762: cuts on the photon and jet rapidities are more effective than cuts on
763: $x_{LL}^{\rm p}$ to enhance the gluon in the photon,
764: as shown in Fig.~\ref{cutrap}.
765: %\begin{wrapfigure}[23]{r}[0.cm]{11.5cm}
766: %\vspace*{-0.5cm}
767: \begin{figure}[htb]
768: \begin{center}
769: \epsfig{file=cutxp.eps,height=11cm}
770: \caption{Effect of a lower cut on $x_{LL}^{\rm p}$ on the relative contribution
771: from the gluon in the photon.}
772: \label{cutxp}
773: \end{center}
774: \end{figure}
775: %\end{wrapfigure}
776: Fig.~\ref{cutrap}\,a) shows that if the photon and jet rapidities are
777: restricted to positive values, the resolved photon component is already
778: fairly large, but the gluon content of the latter is still small.
779: If we restrict the rapidities more to the forward region -- especially the
780: jet rapidity, which can be measured at larger angles -- the direct
781: photon contribution is almost completely suppressed, and the gluon
782: contribution makes up almost 40\% of the total, as shown in
783: Fig.~\ref{cutrap}\,b). Imposing even more severe cuts only decreases
784: the cross section further without increasing the gluon content
785: substantially, as can be seen from Fig.~\ref{cutrap}\,c).
786: Therefore the rapidity cut $\eta^{\gamma}>0.5,\,\eta^{\rm jet}>1.5$
787: seems to be the optimal compromise between enhancement of the gluon content
788: and reduction of the cross section.
789:
790: \begin{figure}[htb]
791: \begin{center}
792: \mbox{\epsfig{file=xgam_rapcuts_new.eps,height=12cm}}
793: \end{center}
794: \caption{Effect of rapidity cuts to enhance the contribution
795: from the gluon in the photon.}
796: \label{cutrap}
797: \end{figure}
798:
799: Note that the lower cuts on the transverse momenta are rather large,
800: $p_T^\gamma>6$\,GeV, $E_T^{\rm jet}>5$\,GeV.
801: One can increase the cross section by choosing lower $p_T$ cuts, as
802: shown in Fig.~\ref{xgamscal}.
803: This figure also shows the scale dependence of
804: $d\sigma/dx_{LL}^{\gamma}$ in the presence of the cuts
805: $\eta^\gamma>0,\,\eta^{\rm jet}>0$ respectively
806: $\eta^\gamma>0.5,\,\eta^{\rm jet}>1.5$.
807: The behaviour of the cross section $d\sigma/dx_{LL}^{\gamma}$,
808: which varies by $\pm$\,8\,\% under the scale changes, is less good than
809: the behaviour of $d\sigma/dx_{LL}^{\rm p}$ (see Fig.~\ref{xp_scal}).
810: However, one should keep in mind that the distribution $g^\gamma$
811: is poorly known and that a determination of the latter with an accuracy of
812: $\pm$\,10\,\% would already be welcome.
813:
814:
815:
816: \begin{figure}[htb]
817: \begin{center}
818: \mbox{\epsfig{file=xgam454_scal.eps,height=13cm}}
819: \end{center}
820: \caption{Scale dependence of $d\sigma/dx_{LL}^{\gamma}$ in the presence of
821: forward rapidity cuts.}
822: \label{xgamscal}
823: \end{figure}
824:
825:
826: \clearpage
827:
828: \section{Conclusions}
829:
830: In this work we studied the possibility to measure the gluon
831: distribution in the proton and in the
832: photon by means of the reaction $\gamma$ + p $\to \gamma\, +$ jet + X.
833: This reaction is well suited for such a study because of
834: the stability of the theoretical prediction
835: under variations of the renormalisation and factorisation scales,
836: and because the prompt photon cross section does not suffer
837: from large uncertainties due to hadronisation in the final state.
838: \par
839:
840: We have shown that gluon induced subprocesses give important
841: contributions to the cross section in the $x$-ranges
842: $0 < x_{LL}^{\rm p} \ \lsim \ 0.1$ for the proton and
843: $0 < x_{LL}^{\gamma} \ \lsim \ 0.2$ for the photon.
844:
845: The effects of cuts on $x_{LL}^{\gamma}$ respectively
846: $x_{LL}^{\rm p}$, or on the
847: pseudo-rapidities $\eta^{\gamma}$ and $\eta^{\rm jet}$,
848: are investigated in detail. We found that a
849: judicious choice of cuts allows us to enhance the `signals', i.e.
850: the gluon induced subprocesses,
851: over the `background' stemming from other subprocesses,
852: to constitute up to $\sim 50\,$\% of the total cross section.
853:
854: However, the relevant cross sections are small, of the
855: order of 10 - 50 pb.
856: Clearly these numbers require a large
857: luminosity to obtain observable effects.
858:
859:
860:
861: \vspace*{8mm}
862:
863: {\bf\Large Acknowledgements}
864:
865: \medskip
866:
867: \noindent
868: GH would like to thank the LPT Orsay and the LAPTH Annecy for their
869: hospita\-li\-ty while part of this work has been completed,
870: and also G\"unter Grindhammer for encouragement
871: to pick up again the subject of asymmetric cuts.
872:
873: \vspace*{4mm}
874:
875:
876: \begin{thebibliography}{99}
877:
878: \bibitem{Breitweg:1999su}
879: J.~Breitweg {\it et al.} [ZEUS Collaboration],
880: %``Measurement of inclusive prompt photon photoproduction at HERA,''
881: Phys.\ Lett.\ B {\bf 472} (2000) 175
882: [hep-ex/9910045];\\
883: J.~Breitweg {\it et al.} [ZEUS Collaboration],
884: %``Observation of isolated high-E(T) photons in photoproduction at HERA,''
885: Phys.\ Lett.\ B {\bf 413} (1997) 201
886: [hep-ex/9708038].
887:
888: \bibitem{Chekanov:2001aq}
889: S.~Chekanov {\it et al.} [ZEUS Collaboration],
890: %``Study of the effective transverse momentum of partons in the proton using prompt photons in photoproduction at HERA,''
891: Phys.\ Lett.\ B {\bf 511} (2001) 19
892: [hep-ex/0104001].
893:
894: \bibitem{h1}H1 Collaboration, submitted to the Int. Europhysics Conference on
895: High Energy Physics, EPS03, July 2003, Aachen (Abstract 093), and
896: to the XXI Int. Symposium on Lepton and Photon Interactions, LP03,
897: August 2003, Fermilab.
898:
899: \bibitem{Gordon:1997yt}
900: L.~E.~Gordon,
901: %``Prompt photon plus jet photoproduction at HERA at next-to-leading order in QCD,''
902: Phys.\ Rev.\ D {\bf 57} (1998) 235
903: [hep-ph/9707464].
904:
905: \bibitem{Fontannaz:2001ek}
906: M.~Fontannaz, J.~P.~Guillet and G.~Heinrich,
907: %``Isolated prompt photon photoproduction at NLO,''
908: Eur.\ Phys.\ J.\ C {\bf 21} (2001) 303
909: [hep-ph/0105121].
910:
911: \bibitem{Fontannaz:2001nq}
912: M.~Fontannaz, J.~P.~Guillet and G.~Heinrich,
913: %``Is a large intrinsic k(T) needed to describe photon + jet photoproduction at HERA?,''
914: Eur.\ Phys.\ J.\ C {\bf 22} (2001) 303
915: [hep-ph/0107262].
916:
917: \bibitem{Krawczyk:2001tz}
918: M.~Krawczyk and A.~Zembrzuski,
919: %``Photoproduction of the isolated photon at DESY HERA in next-to-leading order QCD,''
920: Phys.\ Rev.\ D {\bf 64} (2001) 114017
921: [hep-ph/0105166].
922: \bibitem{Zembrzuski:2003nu}
923: A.~Zembrzuski and M.~Krawczyk,
924: %``Photoproduction of isolated photon and jet at the DESY HERA,''
925: hep-ph/0309308.
926:
927: \bibitem{klasen}
928: M.~Klasen,
929: %``Theory of hard photoproduction,''
930: Rev.\ Mod.\ Phys.\ {\bf 74} (2002) 1221
931: [hep-ph/0206169].
932:
933: \bibitem{Adloff:2003nr}
934: C.~Adloff {\it et al.} [H1 Collaboration],
935: %``Measurement of inclusive jet cross sections in photoproduction at HERA,''
936: DESY-02-225, hep-ex/0302034.
937:
938: \bibitem{Kniehl:2000hk}
939: B.~A.~Kniehl, G.~Kramer and B.~P\"otter,
940: %``Testing the universality of fragmentation functions,''
941: Nucl.\ Phys.\ B {\bf 597} (2001) 337
942: [hep-ph/0011155].
943:
944: \bibitem{Fontannaz:2002nu}
945: M.~Fontannaz, J.~P.~Guillet and G.~Heinrich,
946: %``A NLO calculation of the hadron-jet cross section in photoproduction reactions,''
947: Eur.\ Phys.\ J.\ C {\bf 26} (2002) 209
948: [hep-ph/0206202].
949:
950: \bibitem{Lemrani:2003mj}
951: R.~Lemrani [H1 Collaboration],
952: %``Prompt photon production at HERA,''
953: %By H1 Collaboration and ZEUS Collaboration
954: %(Rachid Lemrani for the collaboration). Aug 2003. 5pp.
955: {\it Talk given at 11th International Workshop on Deep Inelastic
956: Scattering (DIS 2003), St. Petersburg, Russia, 23-27 Apr 2003},
957: hep-ex/0308066;\\
958: %\bibitem{Lemrani-Alaoui:2003gy}
959: R.~Lemrani-Alaoui,
960: {\it ``Prompt photon production at HERA''}, Ph.D. thesis, \\
961: DESY-THESIS-2003-010, available at \\
962: {\it http://www-h1.desy.de/publications/theses\_list.html}.
963:
964: \bibitem{Aurenche:2000nc}
965: P.~Aurenche, L.~Bourhis, M.~Fontannaz and J.~P.~Guillet,
966: %``NLO Monte Carlo approach in 1 or 2 jets photoproduction,''
967: Eur.\ Phys.\ J.\ C {\bf 17} (2000) 413
968: [hep-ph/0006011].
969:
970: \bibitem{frixione}
971: S.~Frixione,
972: %``Isolated photons in perturbative {QCD},''
973: Phys.\ Lett.\ B {\bf 429} (1998) 369
974: [hep-ph/9801442].
975:
976: \bibitem{Frixione:1997ks}
977: S.~Frixione and G.~Ridolfi,
978: %``Jet photoproduction at HERA,''
979: Nucl.\ Phys.\ B {\bf 507} (1997) 315
980: [hep-ph/9707345].
981:
982: \bibitem{Chyla:2003in}
983: J.~Chyla and K.~Sedlak,
984: %``Dijet cross sections in e p collisions: Who is afraid of symmetric cuts?,''
985: hep-ph/0308116.
986:
987: \bibitem{mrst01} A.~D.~Martin, R.~G.~Roberts, W.~J.~Stirling and R.~S.~Thorne,
988: %``MRST2001: Partons and alpha(s) from precise deep inelastic scattering and Tevatron jet data,''
989: Eur.\ Phys.\ J.\ {\bf C23} (2002) 73
990: [hep-ph/0110215].
991:
992: \bibitem{Aurenche:1994in}
993: P.~Aurenche, J.~P.~Guillet and M.~Fontannaz,
994: %``Parton distributions in the photon,''
995: Z.\ Phys.\ C {\bf 64} (1994) 621;\\
996: P.~Aurenche, J.~P.~Guillet and M.~Fontannaz, new version of AFG,
997: publication in preparation.
998:
999: \bibitem{Bourhis:1998yu}
1000: L.~Bourhis, M.~Fontannaz and J.~P.~Guillet,
1001: %``Quark and gluon fragmentation functions into photons,''
1002: Eur.\ Phys.\ J.\ C {\bf 2} (1998) 529
1003: [hep-ph/9704447].
1004:
1005: \bibitem{ktalgo}
1006: %\bibitem{Catani:1993hr}
1007: S.~Catani, Y.~L.~Dokshitzer, M.~H.~Seymour and B.~R.~Webber,
1008: %``Longitudinally invariant K(t) clustering algorithms for hadron-hadron collisions,''
1009: Nucl.\ Phys.\ B {\bf 406} (1993) 187;\\
1010: S.~D.~Ellis and D.~E.~Soper,
1011: %``Successive combination jet algorithm for hadron collisions,''
1012: Phys.\ Rev.\ D {\bf 48} (1993) 3160
1013: [hep-ph/9305266].
1014:
1015: \bibitem{cteq6}
1016: J.~Pumplin, D.~R.~Stump, J.~Huston, H.~L.~Lai, P.~Nadolsky and W.~K.~Tung,
1017: %``New generation of parton distributions with uncertainties from global QCD analysis,''
1018: JHEP {\bf 0207} (2002) 012
1019: [hep-ph/0201195].
1020:
1021:
1022: \end{thebibliography}
1023:
1024: \end{document}
1025:
1026:
1027:
1028:
1029:
1030:
1031:
1032: