1: \documentclass[onecolumn,showpacs,nobibnotes,nofootinbib]{revtex4}
2: %,fleqn
3: \usepackage{amsmath,amssymb}
4: \usepackage{graphicx}
5: %%\usepackage{subfigure}
6:
7: \newcommand{\N}{{\cal{N}}}
8: \newcommand{\x}{\mathbf{x}}
9: \newcommand{\y}{\mathbf{y}}
10: \newcommand{\z}{\mathbf{z}}
11: \newcommand{\vk}{\mathbf{k}}
12: \newcommand{\vkp}{\mathbf{k}_0}
13: \newcommand{\vq}{\mathbf{q}}
14: \newcommand{\vri}{\mathbf{r}}
15: \newcommand{\vrh}{\boldsymbol{\rho}}
16: \newcommand{\vrp}{\boldsymbol{\rho}_0}
17: \newcommand{\cb}{\mathbf{b}}
18: \newcommand{\hyper}{{}_2{\rm F}_1}
19: \newcommand{\cc}[1]{\bar{#1}}
20: \newcommand{\abs}[1]{\left|#1\right|}
21: \newcommand{\BesselJ}[2]{J_{#1}\left(#2\right)}
22: \newcommand{\la}{\label}
23:
24: \begin{document}
25:
26: \title{Traveling waves and geometric scaling at non-zero momentum transfer}
27: \author{C. Marquet}
28: \email{marquet@spht.saclay.cea.fr}
29: \author{R. Peschanski}
30: \email{pesch@spht.saclay.cea.fr}
31: \author{G. Soyez\footnote{on leave from the fundamental theoretical physics
32: group of the University of Li\`ege.}}
33: \email{gsoyez@spht.saclay.cea.fr}
34: \affiliation{SPhT \footnote{URA 2306, unit\'e de recherche associ\'ee au CNRS.},
35: CEA Saclay, B\^{a}t. 774, Orme des Merisiers,
36: F-91191 Gif-Sur-Yvette Cedex, France}
37: \pacs{11.55.-m, 13.60.-r}
38:
39: \begin{abstract}
40: We extend the search for traveling-wave asymptotic solutions of the non-linear
41: Balitsky-Kovchegov (BK) saturation equation to
42: non-forward dipole-target amplitudes. Making use of conformal invariant
43: properties of the Balitsky-Fadin-Kuraev-Lipatov (BFKL)
44: kernel, we exhibit traveling-wave solutions in momentum space in the region
45: where the momentum transfer $q$ is smaller than the
46: characteristic scale $Q$ of the projectile. We prove geometric scaling in the
47: variable $Q/q\Omega_s(Y)$ where $\Omega_s(Y)$ has the
48: same energy dependence as in the forward analysis.
49: %The obtained asymptotic wavefront in momentum space is free from problem caused
50: % by the perturbative tail seen in impact parameter space. Consequences for
51: %phenomenology are drawn.
52: \end{abstract}
53:
54: \maketitle
55:
56: \section{Introduction}\label{sec:intro}
57:
58: Geometric scaling \cite{Stasto:2000er} is an interesting phenomenological
59: feature of high energy deep-inelastic scattering (DIS). It
60: is expressed as a scaling property of the virtual photon-proton cross section,
61: namely
62: $\sigma^{\gamma^*}(Y,Q)=\sigma^{\gamma^*}\!\left({Q}/{Q_s(Y)}\right). $ $Q$ is
63: the virtuality of the photon, $Y$ the total rapidity
64: and $Q_s(Y)$ an increasing function of $Y$, called the saturation scale
65: \cite{endnote17}. On the theory side of the problem, it is
66: convenient to work within the QCD dipole picture of DIS \cite{Nikolaev:1991ja}.
67: %In the leading logs approximation of perturbative QCD at high $Y$, the cross
68: %section factories as
69: %\[ %begin{equation}\label{eq:gammaN}
70: %\sigma^{\gamma^*}(Y,Q)
71: % = \int_0^\infty {\rho}\,d\rho\int_0^1 dz\,|\psi(z,{\rho};Q)|^2\, \N(Y,{\rho})\
72: .
73: %\] %end{equation}
74: %$\psi(z,{\rho}Q)$ is the photon wave function on a $q\bar q$ dipole of size
75: %$\rho$, and $z$ is the longitudinal momentum fraction of the photon carried by
76: %the quark. $\N(Y,\rho)$ is the dipole-target forward scattering amplitude.
77: In this framework, the geometric scaling
78: \begin{equation}\label{eq:defscaling}
79: \N(Y,\rho)=\N\left({\rho}\ {Q_s(Y)}\right)
80: \end{equation}
81: appears to be a genuine property of the conveniently-normalised dipole-target
82: forward scattering amplitude $\N(Y,\rho)$, where
83: $\rho$ is the size of the dipole and $Y$ its rapidity. In fact, it is known
84: that, if this amplitude verifies the non-linear
85: Balitsky-Kovchegov (BK) saturation equation \cite{Balitsky,Kovchegov}, this
86: property is a mathematical consequence of the
87: high-energy behaviour of the solution in terms of traveling waves
88: \cite{Munier1}.
89: More precisely, the equation is shown to admit traveling-wave solutions which
90: translate directly in terms of the geometric scaling
91: property (\ref{eq:defscaling}).
92:
93: However, the BK equation is written for the dipole-target amplitude in the full
94: $2$-dimensional transverse coordinate plane. Hence,
95: it is supposed to give solutions for the dipole amplitude $\N(Y,\rho,b)$ where
96: $b$ is the impact-parameter of the dipole projectile
97: with respect to the target. Saturation as a function of impact parameter has
98: been studied either phenomenologically
99: \cite{Munier:2001nr}, in the framework of models \cite {Kowalski:2003hm}, from
100: a semi-classical approach \cite{Bondarenko:2003ym},
101: from numerical studies of the BK equation
102: \cite{Golec-Biernat:2003ym,Gotsman:2004ra} or from an analytic point of view
103: \cite{Ikeda:2004zp}. However, referring to solutions of the BK equation, the
104: perturbative tail in impact parameter shows up fastly
105: in the BK evolution \cite{Golec-Biernat:2003ym}. It appears difficult to be
106: consistent in a region where the confinement is expected
107: to dominate (see discussions in Refs.\cite{Ferreiro:2002kv,Kovner:2001bh}).
108:
109: In the present paper we want to tackle the problem of non-forward amplitudes in
110: the saturation regime from a new point of view,
111: motivated by (and by extension of) the mathematical properties of non-linear
112: evolution equations, used in Refs.\cite{Munier1} for
113: the forward amplitude. In the present work, we shall stick to the framework of
114: the BK equation but our method could be extended to
115: different cases with the same mathematical features.
116:
117: In Refs.\cite{Munier1}, it was shown that the BK equation for $\N(Y,\rho)$ can
118: be considered to lie in the same universality class
119: than a well-known equation, namely the Fisher or Kolmogorov-Petrovsky-Piscounov
120: (F-KPP) equation \cite{KPP}. It has been shown
121: \cite{Bramson} that these equations admit asymptotic traveling-wave solutions
122: whose physical meaning \cite{Munier1} is nothing else
123: than the geometric scaling property.
124:
125: More generally, if some general features that we shall now point out are
126: realized \cite{ebert}, one can infer the existence and the
127: form \cite{Munier1} of traveling wave solutions only from the knowledge of the
128: linear part of the kernel \cite{brunet}. Let us quote
129: these general features, taking as an example the general structure of the F-KPP
130: equations. It is known that they are governed by
131: three types of terms: a ``diffusion term'', a ``growth term'' and a non-linear
132: ``damping term''. In particular, suppose the
133: ``damping term'' to be absent, the equation is restricted to its linear part and
134: leads to an exponential rise of the solution with
135: ``time'' together with diffusion in ``space''. This is indeed characteristic of
136: the BFKL kernel which governs the linear part of
137: the BK equation. In more general cases, called ``pulled front'' cases in the
138: literature \cite{ebert}, the presence of similar
139: ingredients, together with a specific property of initial conditions being
140: steep enough to carry along the critical regime of
141: traveling waves, induces the existence and the form of traveling wave solutions.
142: The key point is that the main features of these
143: solutions can be determined only from the knowledge of the solutions of linear
144: part of the equation, which is an easier task than
145: looking directly for solutions of a non-linear equation.
146:
147: Our starting point is the knowledge of the exact solutions of the BFKL equation
148: for dipole-dipole scattering, which have been obtained using the conformal
149: invariance of the BFKL kernel \cite{Lipatov86,NP,Navelet:1997tx,LipatovReview}.
150: Our aim is the following: making use of the powerful mathematical properties
151: of conformal symmetry, allowing to formulate exact solutions of the BFKL equation
152: with impact parameter and/or momentum transfer, we can construct the solutions of
153: the linear part of the BK equation. We shall then look for the
154: kinematical domain where the ingredients allowing for the existence
155: and derivation of traveling waves are present. Namely,
156: schematically (these points shall be discussed more completely in the next
157: section):
158: \begin{itemize}
159: \item A solution of the linear part of the equation with an exponential growth,
160: \item A non linear damping effect due to unitarity,
161: \item A steep enough initial condition.
162: \end{itemize}
163: By contrast with previous approaches of non-forward amplitudes, we shall not
164: try to get solutions in the full phase space (which
165: after all is not necessarily dominated by saturation effects) but shall
166: concentrate on kinematical regions where the mathematical
167: requirements for traveling waves are fulfilled.
168:
169: The plan of our study is the following. In section \ref{sec:bindep}, we recall
170: in detail the derivation of the traveling-wave speed
171: ({\em i.e.} the saturation-scale energy dependence) and of the wave front ({\em
172: i.e.} the amplitude) in the case of the forward
173: amplitude. In section \ref{sec:bdep}, we derive the solutions of the non-forward
174: BFKL dipole-dipole amplitudes in various
175: representations : full coordinate space, a mixed one where the external
176: particles are coordinate space dipoles interacting with a
177: given momentum transfer $q,$ and full momentum space where the external
178: particles have given momentum $k,k_0.$ In the next section
179: \ref{sec:tw}, we construct the solutions of the linear part of the BK equation
180: and select the kinematical domains and
181: representations in which one meets the abovementionned requirements and we
182: derive the corresponding properties. Finally, in section
183: \ref{sec:ccl}, we summarise our results and derive their main consequence,
184: namely geometric scaling at nonzero transfer. We show
185: how it provides a solution to the puzzling problem related to the confinement
186: scale. We point out interesting phenomenological
187: consequences of our solutions.
188:
189:
190: \section{Traveling waves in the forward case}\label{sec:bindep}
191:
192: \begin{figure}
193: \begin{center}
194: \includegraphics{chi.eps}
195: \end{center}
196: \caption{Critical exponent $\gamma_c$ for the full BFKL
197: kernel.}\label{fig:critqcd}
198: \end{figure}
199:
200: Let us recall how traveling waves and geometric scaling emerge from the study of
201: the BK equation for forward amplitudes.
202:
203: In the large-$N_c$ approximation, it is well-known that the high-energy
204: behaviour of the dipole forward scattering amplitude off a
205: large ({\em e.g.} nuclear) target follows the Balitsky-Kovchegov equation
206: \cite{Balitsky,Kovchegov} where we neglect the impact
207: parameter dependence. This equation can be put into the form \cite{Kovchegov}
208: \begin{equation}\label{eq:bkb}
209: \partial_Y \N(Y,k) =
210: \bar{\alpha}\chi(-\partial_L)\N(Y,k)-\bar{\alpha}\N^2(Y,k),
211: \end{equation}
212: where
213: \[
214: \chi(\gamma) = 2\psi(1)-\psi(\gamma)-\psi(1-\gamma)
215: \]
216: is the BFKL kernel and
217: \[ %begin{equation}\label{eq:fourier}
218: \N(Y,k)=\int_0^{\infty} \frac{d\rho}{\rho}J_0(k\rho)\,\N(Y,\rho)
219: \] %end{equation}
220: can be interpreted as the density of gluons in momentum space. Here, $L =
221: \log(k^2/k_0^2)$ with $k_0$ some fixed scale. The starting
222: point of Ref.\cite{Munier1} is that, if we expand the BFKL kernel to second
223: order around $\gamma=\scriptstyle{\frac{1}{2}}$, this
224: equation is equivalent, up to a change of variable, to the
225: Fisher-Kolmogorov-Petrovsky-Piscounov (F-KPP) equation \cite{KPP}
226: \begin{equation}\label{eq:KPP}
227: \partial_t u(t,x) = \partial^2_x u(t,x) + u(t,x) - u^2(t,x)
228: \end{equation}
229: with $t\propto Y$. Looking to the different terms in the right hand side of
230: (\ref{eq:KPP}), one can identify a ``diffusion term''
231: ($\partial_x^2 u$), an ``expansion term'' ($u$) and a ``damping term'' ($-u^2$),
232: as explained in the introduction.
233:
234: This equation has been extensively studied and it has been proven that it admits
235: traveling-wave solutions \cite{Bramson} {\em i.e.},
236: at large time, the solution can be written
237: \[
238: u(t,x)\underset{t\rightarrow +\infty}{\sim}f(x-m(t))
239: \]
240: with
241: \[ %begin{equation}\label{eq:velocity}
242: m(t)=v t- w \log t+ z t^{-1/2} + {\cal O}(1/t)\ ,
243: \] %end{equation}
244: where the constants $v,w,z$ can be determined \cite{ebert} from the properties
245: of the linear regime. If the initial condition
246: behaves like $e^{-\beta x}$ with $\beta > \beta_c = 1$, these coefficients
247: acquire critical values, whatever the value of $\beta$
248: is. In particular, the speed $v$ takes the critical value $v_c=2$. The three
249: coefficients are the only ``critical'' ones since they
250: do not depend on the specific form of the non-linear damping.
251: % and they were derived \cite{Munier1} for the BK equation.
252:
253: In Ref.\cite{Munier1}, it has been shown that this kind of feature is expected
254: to be much more general. Let us for instance show
255: the properties of the critical regime, when the linear part of the evolution
256: admits a superposition of waves for solution. One
257: writes:
258: \begin{equation}\label{eq:waves}
259: u(t,x)
260: = \int_{c-i\infty}^{c+i\infty} \frac{d\gamma}{2i\pi} u_0(\gamma) e^{-\gamma x +
261: \omega(\gamma) t}
262: = \int_{c-i\infty}^{c+i\infty} \frac{d\gamma}{2i\pi} u_0(\gamma)
263: e^{-\gamma(x_{wf}+v t) + \omega(\gamma) t},
264: \end{equation}
265: where $\omega(\gamma)$ is the Mellin transform of the linear kernel and
266: $x_{wf}=x-vt$ is the position relative to the wavefront.
267: Then, the non-linear term drive the solution to the critical behaviour
268: corresponding to traveling waves at large time
269: \begin{equation}\label{eq:travwaves}
270: u(t,x) \underset{t\to\infty}{\sim} e^{-\gamma_c x_{wf}}.
271: \end{equation}
272: The critical exponent $\gamma_c$ corresponds to a wave having a minimal phase
273: velocity equal the group velocity (see {\em e.g.}
274: \cite{GLR}):
275: \begin{equation}\label{eq:speed}
276: v_c = \frac{\omega(\gamma_c)}{\gamma_c} = \left.\partial_\gamma
277: \omega(\gamma)\right|_{\gamma_c}.
278: \end{equation}
279: The appearance of traveling waves \eqref{eq:travwaves} with critical velocity
280: given by \eqref{eq:speed} only depends on a few
281: general conditions:
282: \begin{itemize}
283: \item $u=0$ is an unstable fix point of the equation, and $u=1$ is a stable fix
284: point.
285: \item the linearised evolution equation admits solutions of the form
286: \eqref{eq:waves}. It generates a growth of the solution and
287: non-linearities damp the solution and saturate it. % to 1.
288: \item The initial condition is steep enough. If $u_0(x) \sim e^{-\gamma_0 x}$,
289: this means that we want $\gamma_0 > \gamma_c$.
290: \end{itemize}
291:
292: In particular, in the case of the $b$-independent BK equation, $\omega(\gamma) =
293: \bar\alpha\chi(\gamma)$ and we find $\gamma_c
294: \approx 0.6275$, as represented in Fig.\ref{fig:critqcd}. As noticed in
295: \cite{Munier1} we expect $\gamma_0 = 1$ due to QCD colour
296: transparency and, therefore, the QCD evolution in rapidity will asymptotically
297: reach the traveling wave regime \eqref{eq:travwaves}
298: which in turn corresponds to geometric scaling
299: \begin{equation}\label{eq:scaling}
300: \N(Y,k) = \N\left(\frac{k^2}{Q_s^2(Y)}\right)
301: \end{equation}
302: where the saturation scale $Q_s^2(Y)$ grows as $k_0^2e^{v_cY}$ and the critical
303: speed $v_c$ is $4.8834\,\bar\alpha$.
304:
305: In this paper we investigate in which kinematical regions one has similar
306: properties from the BK equation in full phase-space. We
307: shall first discuss under which circumstances the BFKL dynamics leads to
308: solution of the form \eqref{eq:waves}. We shall then show
309: how we can obtain solutions for the linear BK equation and how non linearities
310: lead to the formation of traveling waves and to
311: geometric scaling.
312:
313: \section{Linear BFKL dynamics at nonzero momentum transfer}\label{sec:bdep}
314:
315: If we now take into account the $b$-dependence in the BK equation, we have to
316: study the asymptotic solutions of
317: \begin{equation}\label{eq:bk}
318: \partial_Y \N(\x,\y) = \frac{\bar{\alpha}}{2\pi}\int d^2z
319: \frac{(\x-\y)^2}{(\x-\z)^2(\z-\y)^2}
320: \left[\N(\x,\z)+\N(\z,\y)-\N(\x,\y)-\N(\x,\z)\N(\z,\y) \right],
321: \end{equation}
322: where $\N(\x, \y)$ is the conveniently-normalised dipole-target scattering
323: amplitude. $\x$ and $\y$ are then the transverse space
324: coordinates of the quark and antiquark constituting the dipole. This equation
325: does not depend explicitly of the target whose centre
326: of mass defines the origin. This non-linear equation corresponds to resumming
327: QCD fan diagrams in the leading-logarithmic
328: approximation \cite{Kovchegov}.
329:
330: Our main tool is to use the knowledge of the exact BFKL solutions
331: \cite{Lipatov86,NP,Navelet:1997tx} for dipole-dipole scattering.
332: We shall study the amplitude $f(\vrh_1, \vrh_2, \vrh_{1'}, \vrh_{2'})$ (see
333: Fig.\ref{fig:coord}) where we identify $\vrh_1$ and
334: $\vrh_2$ with $\x$ and $\y$ . We impose that one dipole is bigger than the other
335: ($\rho = \abs{\vrh_1-\vrh_2} \ll \rho_0 =
336: \abs{\vrh_{1'}-\vrh_{2'}} \ll 1/\Lambda_{QCD}$).
337: Indeed, if we want to use the dipole-dipole amplitude in order to study the
338: dipole-target amplitude within the BK framework, we need
339: to extract solutions of the linear part of the BK equation \eqref{eq:bk} from
340: the solutions of the BFKL equation. This is done by
341: integrating out the target impact factor, and gives relevant results provided
342: the convolution with the large dipole factorises from
343: the smaller one. This separation criterium closely depends on the conformal
344: invariance properties of the BFKL equation which is one
345: of the basic properties of the non-forward linear BFKL dynamics
346: \cite{Lipatov86}.
347:
348: Since our discussion may lead to different theoretical and phenomenological
349: features, we shall perform our analysis in different
350: spaces: the coordinate space, the mixed representation where we keep the dipole
351: sizes in coordinate space but use the momentum
352: transfer $q$ instead of the impact parameter, and the full momentum space.
353:
354:
355: \subsection{Coordinate space}\label{sec:coord}
356:
357: \begin{figure}
358: \includegraphics{space.eps}
359: \caption{Relevant variables in the collision of two dipoles viewed in the
360: transverse plane.}\label{fig:coord}
361: \end{figure}
362:
363: Let us start from the well-known BFKL amplitude \cite{Lipatov86}
364: \[ %begin{equation}\label{eq:bfklampl}
365: {\cal A}(s,q^2) = is \int\frac{d\omega}{2i\pi} e^{\omega Y} f_\omega(q^2),
366: \] %end{equation}
367: where $s=e^Y$ is the centre-of-mass energy squared and $q$ is the momentum
368: transfer.
369: %We keep only the dominant component in energy which corresponds to vanishing
370: %conformal spin \cite{Lipatov86}, $n=0$ or $\gamma=\frac{1}{2}+i\nu$),
371: One writes \cite{Lipatov86}
372: \begin{eqnarray}
373: f_\omega(q^2)\delta^{(2)}(\vq-\vq') & = & \int \prod_i
374: \frac{d^2\rho_i}{(2\pi)^2}\,
375: e^{i\frac{1}{2}(\vrh_1+\vrh_2).\vq}\,\Phi_P(\vrh,\vq) f_\omega(\vrh_1, \vrh_2,
376: \vrh_{1'},
377: \vrh_{2'})e^{-i\frac{1}{2}(\vrh_{1'}+\vrh_{2'}).\vq'}\,\Phi_T(\vrp,\vq'),
378: \nonumber\\
379: f_\omega(\vrh_1, \vrh_2, \vrh_{1'}, \vrh_{2'}) & = & \int_{-\infty}^\infty
380: \frac{\nu^2\,d\nu}{2\pi(\nu^2+1/4)^2}\frac{1}{\omega-\bar\alpha \chi(\nu)} \int
381: d^2r\,
382: E^\nu(\vrh_{1}\!-\!\vri,\vrh_{2}\!-\!\vri)\bar E^\nu(\vrh_{1'}\!-\!\vri,
383: \vrh_{2'}\!-\!\vri) \label{eq:fomegarrb}
384: \end{eqnarray}
385: where
386: \[
387: E^\nu(\vrh_1,\vrh_2)=\left(\frac{|\vrh_1-\vrh_2|}{|\vrh_1||\vrh_2|}\right)^{1+2i
388: \nu}
389: \]
390: are the conformal eigenfunctions of the BFKL kernel with vanishing conformal
391: spin quantum number ($n=0$ in the usual terminology
392: \cite{Lipatov86}). It corresponds to the dominant contribution at high energy.
393: $\Phi_P$ (resp. $\Phi_T$) are the impact factors
394: describing the coupling of the BFKL pomeron to the projectile (resp. target),
395: depending on the momentum transfer $\vq$ and on the
396: dipole sizes $\vrh\! =\! \vrh_1\! -\! \vrh_2$ and
397: $\vrp\!=\!\vrh_{1'}\!-\!\vrh_{2'}$.
398:
399: The Dirac distribution which appears in the definition of $f_\omega(q^2)$ comes
400: from the fact that, the global system being invariant
401: under translation, the expression in the right-hand-side does not depend on
402: $\vrh_1\!+\!\vrh_2\!+\!\vrh_{1'}\!+\!\vrh_{2'}$. To be
403: more precise, the relevant variables of this problem are the sizes of the two
404: dipoles, $\vrh$ and $\vrp$, and the impact parameter $
405: \cb\!=\!\frac{1}{2}(\vrh_1\!+\!\vrh_2\!-\!\vrh_{1'}\!-\!\vrh_{2'})$, as shown on
406: Fig.\ref{fig:coord}. Combining these expressions
407: and using\footnote{In the following, we shall use indifferently $\gamma$ or
408: $\nu$ as superscripts, {\em e.g.} $E^\nu$ and $E^\gamma$
409: denote the same function.} $\gamma=\frac{1}{2}+i\nu$ instead of $\nu$, we obtain
410: the following expression for the amplitude
411: \begin{equation}\label{eq:amplrrb}
412: {\cal A}(s,q^2) = \frac{is}{(2\pi)^6} \int
413: d^2b\,d^2\rho\,d^2\rho_0\,e^{i\vq.\cb}\Phi_P(\vrh,
414: \vq)\Phi_T(\vrp,\vq)\int_{\frac{1}{2}-i\infty}^{\frac{1}{2}+i\infty}
415: \frac{d\gamma}{2i\pi}\,e^{\bar\alpha\chi(\gamma)Y}f^\gamma(\vrh,\vrp,\cb),
416: \end{equation}
417: where we have introduced the function $f^\gamma$ as the result of the
418: integration over $\vri$ in \eqref{eq:fomegarrb}
419: \[
420: f^\gamma(\vrh,\vrp,\cb) =
421: \frac{-\left(\gamma-\frac{1}{2}\right)^2}{\gamma^2(1-\gamma)^2}\int
422: d^2r\,E^\gamma(\vrh_{1}\!-\!\vri,\vrh_{2}\!-\!\vri)\bar
423: E^\gamma(\vrh_{1'}\!-\!\vri, \vrh_{2'}\!-\!\vri).
424: \]
425: Using the complex representation for the two-dimensional vectors\footnote{The
426: complex representation of the vector $\x=(x_1,x_2)$ is
427: \[
428: x=x_1\!+ ix_2, \qquad \bar x=x_1\!-ix_2.
429: \]
430: In this section, we shall explicitly use $\abs{x}$ when the modulus of the
431: vector has to be considered.\label{foot:cplx}} (in this
432: section, we keep bold characters for vectors and ordinary ones for complex
433: numbers), it has been shown \cite{LipatovReview, NP} that
434: the result of this integration is
435: \[
436: f^\gamma(\rho,\rho_0,b) = \frac{c_\gamma}{\gamma^2(1-\gamma^2)}
437: \abs{z}^{2\gamma}\hyper(\gamma,\gamma;2\gamma;z)\hyper(\gamma,\gamma;2\gamma;\cc
438: {z}) + (\gamma \leftrightarrow 1-\gamma),
439: \]
440: where $\hyper$ is the Gauss hypergeometric function \cite{math}, the pre-factor
441: \begin{equation}\label{eq:cgamma}
442: c_\gamma = \pi \,
443: 2^{1-4\gamma}\frac{\Gamma(\gamma)\,\Gamma\left(\frac{3}{2}-\gamma\right)}{\Gamma
444: (1-\gamma)\Gamma\left(\gamma-\frac{1}{2}\right)}\,,
445: \end{equation}
446: and
447: \begin{equation}\label{eq:ratio}
448: z = \frac{\rho_{12}\rho_{1'2'}}{\rho_{11'}\rho_{22'}} \equiv
449: \frac{4\rho\rho_0}{4b^2-(\rho-\rho_0)^2}
450: \end{equation}
451: is the anharmonic ratio. This is a remarkable property due to conformal
452: invariance that the BFKL pomeron exchange in coordinate
453: space depends on $\rho$, $\rho_0$ and $b$ only through the anharmonic ratio $z$.
454:
455: Since we are interested in the situation where one small dipole of size $\rho$
456: scatters on a larger one of size $\rho_0$, {\em i.e.}
457: $\abs{\rho}\ll\abs{\rho_0}$, we can expand $f^\gamma(\rho, \rho_0, b)$ in series
458: of $\rho/\rho_0$. In that limit, the hypergeometric function
459: goes to 1 and we obtain
460: \begin{equation}\label{eq:rrb}
461: f^\gamma(\rho, \rho_0, b) \approx \frac{c_\gamma}{\gamma^2(1-\gamma)^2}
462: \abs{\frac{4\rho\rho_0}{4b^2-\rho_0^2}}^{2\gamma} + (\gamma
463: \leftrightarrow 1-\gamma).
464: \end{equation}
465: Before going any further, one has to point out that the corrections to this
466: expression are of order $(\rho/\rho_0)^2$ but reduce to
467: $(\rho/\rho_0)^4$ if we integrate over the phase $\phi$ of $\rho$ (see
468: Fig.\ref{fig:coord} for the kinematics).
469:
470: The expression \eqref{eq:rrb} still depends on the angle $\phi_0$ between
471: $\rho_0$ and $b$. Since this is not relevant for
472: phenomenological studies, it is interesting to integrate this result over the
473: phases $\phi$ and $\phi_0$ of $\rho$ and $\rho_0$ (one
474: may assume that $b$ is real). One obtains
475: \begin{equation}\label{eq:rrb_mean}
476: \left\langle f^\gamma(\rho,\rho_0,b)\right\rangle_{\phi,\phi_0}
477: = \frac{c_\gamma}{\gamma^2(1-\gamma)^2}
478: \left(\frac{4\abs{\rho}\abs{\rho_0}}{\abs{\abs{\rho_0}^2-4\abs{b}^2}}
479: \right)^{2\gamma}P_{\gamma-1}\left(\frac{\abs{\rho_0}^4
480: +16\abs{b}^4}{\abs{\abs{\rho_0}^4-16 \abs{b}^4}}\right)
481: + (\gamma \leftrightarrow 1-\gamma),
482: \end{equation}
483: where %, here, $\rho$, $\rho_0$ and $b$ represent the modulus of the vectors and
484: $P_{\gamma-1}(x)$ is the Legendre function of the first kind \cite{math}.
485:
486: Finally, if we additionally take $b$ going to zero in this result, we obtain
487: \begin{equation}\label{eq:rrbb}
488: \left\langle f^\gamma(\rho,\rho_0,0)\right\rangle_{\phi,\phi_0}
489: = \frac{1}{\gamma^2(1-\gamma)^2}\left(c_\gamma
490: \abs{\frac{4\rho}{\rho_0}}^{2\gamma}
491: + c_{1-\gamma} \abs{\frac{4\rho}{\rho_0}}^{2-2\gamma}\right)
492: \end{equation}
493: which exhibits the same power behaviour as in the forward case.
494:
495:
496: \subsection{Mixed space}\label{sec:mixed}
497:
498: It is useful to introduce a mixed space representation \cite{Lipatov86} in terms
499: of $\rho$, $\rho_0$ and $q$. One obtains from
500: \eqref{eq:amplrrb}
501: \begin{equation}\label{eq:amplrrq}
502: {\cal{A}}(s,q^2) = \frac{is}{(2\pi)^4}\int d^2\rho\, d^2\rho_0\, \Phi_P(\vrh,
503: \vq)\Phi_T(\vrp,\vq)\int_{-\infty}^\infty
504: \frac{d\nu}{2\pi}\,f_{q}^\nu(\vrh, \vrp)\,e^{\bar\alpha\chi(\nu)Y}
505: \end{equation}
506: where
507: \begin{equation}\label{eq:fqgamma}
508: f_q^\nu(\vrh, \vrp) = \int
509: \frac{d^2b}{(2\pi)^2}\,e^{i\vq.\cb}\,f^\nu(\vrh,\vrp,\cb) =
510: \frac{\abs{\vrh}\abs{\vrp}}{16(\nu^2+1/4)^2}
511: \bar E_q^\nu(\vrp) E_q^\nu(\vrh)
512: \end{equation}
513: acquires a factorised form with
514: \[
515: E_q^\nu(\vrh)=\frac{4c_\gamma}{\pi^2|\vrh|}\int d^2z\
516: e^{i\vq.\z}E^\nu\left(\z+\frac{\vrh}{2},\z-\frac{\vrh}{2}\right).
517: \]
518: Using again the complex representation for the two-dimensional vectors, one
519: finds (see \cite{NP})
520: \[
521: E_q^\nu(\rho) = \abs{q}^{2i\nu} 2^{-6i\nu} \Gamma^2(1-i\nu)
522: \left\lbrack
523: \BesselJ{-i\nu}{\frac{\bar
524: q\rho}{4}}\BesselJ{-i\nu}{\frac{q\bar\rho}{4}}
525: - \BesselJ{ i\nu}{\frac{\bar q\rho}{4}}\BesselJ{
526: i\nu}{\frac{q\bar\rho}{4}}
527: \right\rbrack \quad\text{and}\qquad
528: \bar{E}_q^\nu(\rho_0) = E_q^{-\nu}(\rho_0)\,,
529: \]
530: where $\BesselJ{\alpha}{z}$ is the Bessel function of the first kind.
531: Using $\gamma = \frac{1}{2}+i\nu$ instead of $\nu$, the $(\nu \leftrightarrow
532: -\nu)$ symmetry turns into a $(\gamma \leftrightarrow
533: 1-\gamma)$ symmetry and we finally get
534: \begin{eqnarray}
535: f_q^\gamma(\rho, \rho_0) =
536: \frac{\abs{\rho\rho_0}}{16 \gamma^2(1-\gamma)^2}
537: \Gamma^2\left(\frac{3}{2}-\gamma\right)
538: \Gamma^2\left(\frac{1}{2}+\gamma\right) \label{eq:bfkl_rrq}
539: &&\left\lbrack
540: \BesselJ{\frac{1}{2}-\gamma}{\frac{\bar q\rho}{4}}
541: \BesselJ{\frac{1}{2}-\gamma}{\frac{q\bar \rho}{4}}
542: - \BesselJ{\gamma-\frac{1}{2}}{\frac{\bar q\rho}{4}}
543: \BesselJ{\gamma-\frac{1}{2}}{\frac{q\bar \rho}{4}}
544: \right\rbrack\\\times&& \left\lbrack
545: \BesselJ{\gamma-\frac{1}{2}}{\frac{\bar q\rho_0}{4}}
546: \BesselJ{\gamma-\frac{1}{2}}{\frac{q\bar \rho_0}{4}}
547: - \BesselJ{\frac{1}{2}-\gamma}{\frac{\bar q\rho_0}{4}}
548: \BesselJ{\frac{1}{2}-\gamma}{\frac{q\bar \rho_0}{4}}
549: \right\rbrack\,.\nonumber
550: \end{eqnarray}
551: Let us notice that this expression has the crucial feature that it is {\em
552: factorised} in $\rho$ and $\rho_0$ by contrast with the
553: $b$ representation. This property, inherent to the fact that we use the momentum
554: transfer $q$ instead of the impact parameter $b$,
555: will be of prime importance for obtaining the solutions of the BK equation.
556:
557: As explained before, we are interested in the situation where one dipole is much
558: larger than the other one ($\rho_0\gg\rho$). In
559: order to expand the Bessel functions in series of their respective arguments, we
560: shall consider three situations: $\rho_0\gg\rho\gg
561: 1/q$, $\rho_0\gg1/q\gg\rho$ and $1/q\gg\rho_0\gg\rho$. The results then follow
562: from the asymptotic expansion of the Bessel functions
563: \begin{eqnarray}
564: \BesselJ{\mu}{z}\BesselJ{\mu}{\bar z}-\BesselJ{-\mu}{z}\BesselJ{-\mu}{\bar z}
565: & \stackrel{\abs{z}\ll 1}{\longrightarrow} &
566: \frac{1}{\Gamma^2(1+\mu)}\abs{\frac{z}{2}}^{2\mu}
567: -
568: \frac{1}{\Gamma^2(1-\mu)}\abs{\frac{z}{2}}^{-2\mu}, \label{eq:BesselSmall}\\
569: & \stackrel{\abs{z}\gg 1}{\longrightarrow} & \frac{-2}{\pi\abs{z}} \sin(\mu
570: \pi)\cos\left[2\Re e(z)\right].\label{eq:BesselLarge}
571: \end{eqnarray}
572:
573: Using these expressions we obtain the following results for the three relevant
574: cases.
575: \begin{itemize}
576: \item \underline{Case 1}: $1/q\gg\rho_0\gg\rho$.\\
577: Using \eqref{eq:bfkl_rrq} and \eqref{eq:BesselSmall}, we find
578: \begin{eqnarray*}
579: f_q^\gamma(\rho,\rho_0) & = & \frac{\abs{\rho\rho_0}}{16 \gamma^2(1-\gamma)^2}
580: \Gamma^2\left(\frac{3}{2}-\gamma\right)
581: \Gamma^2\left(\frac{1}{2}+\gamma\right) \label{eq:bfkl_bbq}\\
582: &\times&\left\lbrack
583: \frac{1}{\Gamma^2\left(\frac{3}{2}-\gamma\right)}\abs{\frac{\rho
584: q}{8}}^{1-2\gamma}
585: - \frac{1}{\Gamma^2\left(\frac{1}{2}+\gamma\right)}\abs{\frac{\rho
586: q}{8}}^{2\gamma-1}
587: \right\rbrack \left\lbrack
588: \frac{1}{\Gamma^2\left(\frac{1}{2}+\gamma\right)}\abs{\frac{\rho_0
589: q}{8}}^{2\gamma-1}
590: - \frac{1}{\Gamma^2\left(\frac{3}{2}-\gamma\right)}\abs{\frac{\rho_0
591: q}{8}}^{1-2\gamma}
592: \right\rbrack.
593: \end{eqnarray*}
594: In particular, if we take $q\to 0$ in this expression, we recover
595: \begin{equation}
596: f_q^\gamma(\rho,\rho_0) \stackrel{q\to 0}{\longrightarrow}
597: \frac{\abs{\rho\rho_0}}{16
598: \gamma^2(1-\gamma)^2}\left(\abs{\frac{\rho}{\rho_0}}^{1-2\gamma} +
599: \abs{\frac{\rho}{\rho_0}}^{2\gamma-1}\right).
600: \end{equation}
601:
602: \item \underline{Case 2}: $\rho_0\gg1/q\gg\rho$.\\
603: Due to the asymptotic expansion \eqref{eq:BesselLarge}, the $\rho_0$ part will
604: not depend on $\gamma$ anymore and we obtain
605: \begin{equation}\label{eq:rrq_case2}
606: f_q^\gamma(\rho,\rho_0) =
607: \frac{-1}{2\pi}\frac{\cos(\gamma\pi)}{\gamma^2(1-\gamma)^2} \cos\left[\frac{\Re
608: e(\rho_0\bar
609: q)}{2}\right]\frac{1}{\abs{q}^2}
610: \left[\Gamma^2\left(\frac{3}{2}-\gamma\right)\abs{\frac{\rho q}{8}}^{2\gamma}-
611: \Gamma^2\left(\frac{1}{2}+\gamma\right)\abs{\frac{\rho
612: q}{8}}^{2-2\gamma}\right].
613: \end{equation}
614: Note that this amplitude still depends on the angle $\psi_0$ between $q$ and
615: $\rho_0$. If we integrate out this angle together with
616: the dependence on the angle $\psi$ between $q$ and $\rho$, we have
617: \begin{equation}\label{eq:rrq_case2_noangle}
618: \left\langle f_q^\gamma(\rho,\rho_0) \right\rangle_{\psi,\psi_0} =
619: \frac{-\cos(\gamma\pi)}{\gamma^2(1-\gamma)^2}
620: \cos\left(\frac{\abs{\rho_0
621: q}}{2}-\frac{\pi}{4}\right)\sqrt{\frac{1}{\pi^3\abs{\rho_0 q^3}}}
622: \left[\Gamma^2\left(\frac{3}{2}-\gamma\right)\abs{\frac{\rho q}{8}}^{2\gamma}-
623: \Gamma^2\left(\frac{1}{2}+\gamma\right)\abs{\frac{\rho
624: q}{8}}^{2-2\gamma}\right].
625: \end{equation}
626:
627: \item \underline{Case 3}: $\rho_0\gg\rho\gg 1/q$.\\
628: In this case, both the $\rho$- and $\rho_0$-dependent parts of
629: \eqref{eq:bfkl_rrq} involve expansion \eqref{eq:BesselLarge} which
630: gives
631: \begin{equation}\label{eq:rrq_case3}
632: f_q^\gamma(\rho,\rho_0) =
633: \frac{-4}{\pi^2}\frac{\cos^2(\gamma\pi)}{\gamma^2(1-\gamma)^2}
634: \Gamma^2\left(\frac{1}{2}+\gamma\right)\Gamma^2\left(\frac{3}{2}-
635: \gamma\right)\frac{1}{\abs{q}^2}\cos\left[\frac{\Re e(\rho\bar
636: q)}{2}\right]\cos\left[\frac{\Re e(\rho_0\bar q)}{2}\right]
637: \end{equation}
638: and, averaging over the angles $\psi$ and $\psi_0$,
639: \begin{equation}\label{eq:rrq_case3_noangle}
640: \left\langle f_q^\gamma(\rho,\rho_0) \right\rangle_{\psi,\psi_0} =
641: \frac{-16}{\pi^4}\frac{\cos^2(\gamma\pi)}{\gamma^2(1-\gamma)^2}
642: \Gamma^2\left(\frac{1}{2}+\gamma\right)\Gamma^2\left(\frac{3}{2}
643: -\gamma\right)\frac{1}{\abs{\rho\rho_0q^4}}
644: \cos\left(\frac{\abs{\rho q}}{2}-\frac{\pi}{4}\right)
645: \cos\left(\frac{\abs{\rho_0 q}}{2}-\frac{\pi}{4}\right).
646: \end{equation}
647: \end{itemize}
648:
649: %Before going to the momentum space, we have to note that
650: %\eqref{eq:rrq_case2_noangle} and \eqref{eq:rrq_case3_noangle} are not
651: %positive defined.
652:
653:
654:
655: \subsection{Momentum space}\label{sec:mom}
656:
657: The situation in momentum space is very similar to the one in mixed space.
658: If we still neglect the contribution of higher conformal
659: spins, we find
660: \begin{equation}\label{eq:amplkkq}
661: {\cal A}(s,q^2) = is \int d^2k\, d^2k_0\, \Phi_P(\vk, \vq)\Phi_T(\vkp,
662: \vq)\int_{-\infty}^\infty \frac{d\nu}{2\pi}\, e^{\bar
663: \alpha\chi(\nu)Y} f^\nu(\vk, \vkp, \vq),
664: \end{equation}
665: where $\Phi_P$ and $\Phi_T$ are the impact factors in momentum space
666: \[ %begin{equation}\label{eq:impactkkq}
667: \Phi_P(\vk, \vq)=\int \frac{d^2\rho}{(2\pi)^2}\,
668: e^{-i(\vk-\vq/2).\vrh}\,\Phi_P(\vrh, \vq),
669: \] %end{equation}
670: and $f^\nu(\vk,\vkp, \vq)$ is the Fourier transform of $f_q^\nu(\vrh, \vrp)$:
671: \begin{equation}\label{eq:fkkqdef}
672: f^\nu(\vk,\vkp,\vq) = \frac{1}{(2\pi)^4}\frac{1}{(\nu^2+1/4)^2}
673: \int d^2\rho\, e^{i(\vk-\vq/2).\vrh}\, \frac{\abs{\vrh}}{4} E_q^\nu(\vrh)
674: \int d^2\rho_0\, e^{i(\vkp-\vq/2).\vrp}\, \frac{\abs{\vrp}}{4} \bar
675: E_q^\nu(\vrp).
676: \end{equation}
677: The detailed calculation of this integral is presented in Appendix
678: \ref{ap:fourier} and the relation with conformal invariance is
679: developed in Appendix \ref{ap:sl2c}. The final expression, using again the
680: complex representation (see footnote \ref{foot:cplx}) and
681: replacing $\nu$ by $\gamma=\frac{1}{2}+i\nu$, is
682: \begin{eqnarray}
683: f^\gamma(k,k_0,q)
684: & = & \frac{-1}{(2\pi)^4}\frac{\sin^2(\gamma\pi)}{16\gamma^2(1-\gamma)^2}
685: \Gamma^2\left(\frac{1}{2}+\gamma\right)\Gamma^2\left(\frac{3}{2}-\gamma\right)
686: \left(\frac{4}{\abs{kk_0}}\right)^3\nonumber\\
687: &\times&\left\lbrack
688: \frac{\Gamma^2(1+\gamma)}{\Gamma^2\left(\frac{1}{2}+\gamma\right)}
689: \abs{\frac{q}{4k}}^{2\gamma-1}
690: \hyper\left(\gamma,1+\gamma;2\gamma;\frac{q}{k}\right)
691: \hyper\left(\gamma,1+\gamma;2\gamma;\frac{\bar q}{\bar k}\right)
692: - (\gamma\leftrightarrow 1-\gamma)
693: \right\rbrack\nonumber\\
694: &\times&\left\lbrack
695: \frac{\Gamma^2(1+\gamma)}{\Gamma^2\left(\frac{1}{2}+\gamma\right)}
696: \abs{\frac{q}{4k_0}}^{2\gamma-1}
697: \hyper\left(\gamma,1+\gamma;2\gamma;\frac{q}{k_0}\right)
698: \hyper\left(\gamma,1+\gamma;2\gamma;\frac{\bar q}{\bar k_0}\right)
699: - (\gamma\leftrightarrow 1-\gamma)
700: \right\rbrack.\label{eq:fkkpq}
701: \end{eqnarray}
702: As in the mixed space representation, this expression has the remarkable
703: property of being {\em factorised} in $k$ and $k_0$. This
704: appears from the fact that we use the momentum transfer $q$ and is lost if we go
705: back to impact-parameter $b$.
706:
707: We can now consider these expressions in the case where one dipole is much harder
708: than the other, corresponding to $k\gg k_0$. Thanks
709: to the fact that $f^\gamma$ is factorised in $q/k$ and $q/k_0$, we again need to
710: consider three situations: $k\gg k_0\gg q$, $k\gg
711: q\gg k_0$ and $q\gg k\gg k_0$. When $q$ is small compared to $k$, it is
712: sufficient to consider
713: \[
714: \hyper\left(\gamma,1+\gamma;2\gamma;\frac{q}{k}\right) \stackrel{\abs{k}\gg
715: \abs{q}}{\approx} 1\,.
716: \]
717: The behaviour at large $q$ is more complicated. We show in Appendix
718: \ref{ap:asymp} that, averaging over the angle $\theta$ between
719: $k$ and $q$, we have
720: \begin{eqnarray}
721: \left\langle \frac{\Gamma^2(1+\gamma)}{\Gamma^2\left(\frac{1}{2}+\gamma\right)}
722: \abs{\frac{q}{4k}}^{2\gamma-1}
723: \hyper\left(\gamma,1+\gamma;2\gamma;\frac{q}{k}\right)
724: \hyper\left(\gamma,1+\gamma;2\gamma;\frac{\bar q}{\bar k}\right)
725: - (\gamma\leftrightarrow 1-\gamma)
726: \right\rangle_{\!\theta}\nonumber\\
727: \stackrel{\abs{k}\ll \abs{q}}{\longrightarrow}
728: 4\gamma^2(1-\gamma)^2\frac{\cos(\gamma\pi)}{\sin(\gamma\pi)}\abs{\frac{q}{k}}^{-
729: 3}\,\log\abs{\frac{q}{k}}.\label{eq:2f1exp}
730: \end{eqnarray}
731: Before considering the three cases in more details, one may note that the
732: $q^{-3}$ behaviour could have been anticipated from the
733: fact that, in that limit, the Fourier transform of $E_q^\nu(\rho)$ does not
734: depend on $k$. Note however the logarithmic term arising
735: from the hypergeometric function.
736:
737: Let us now consider the results for the three different cases.
738: \begin{itemize}
739: \item \underline{Case 1}: $k\gg k_0\gg q$.\\
740: In this case, both parts of the expression show a power behaviour,
741: \begin{eqnarray}
742: f^\gamma(k,k_0,q) & = & \frac{-1}{\pi^4}
743: \frac{\sin^2(\gamma\pi)}{4\gamma^2(1-\gamma)^2}\Gamma^2\left(\frac{3}{2}
744: -\gamma\right)\Gamma^2\left(\frac{1}{2}+\gamma\right)
745: \frac{1}{\abs{kk_0}^3}\label{eq:kkq_case1}\\
746: &\times& \left\lbrack
747: \frac{\Gamma^2\left(1+\gamma\right)}{\Gamma^2\left(\frac{1}{2}+\gamma\right)}
748: \abs{\frac{q}{4k}}^{2\gamma-1}
749: - \frac{\Gamma^2\left(2-\gamma\right)}{\Gamma^2\left(\frac{3}{2}-\gamma\right)}
750: \abs{\frac{q}{4k}}^{1-2\gamma}\right\rbrack
751: \left\lbrack
752: \frac{\Gamma^2\left(1+\gamma\right)}{\Gamma^2\left(\frac{1}{2}+\gamma\right)}
753: \abs{\frac{q}{4k_0}}^{2\gamma-1}
754: - \frac{\Gamma^2\left(2-\gamma\right)}{\Gamma^2\left(\frac{3}{2}-\gamma\right)}
755: \abs{\frac{q}{4k_0}}^{1-2\gamma}\right\rbrack\nonumber\\
756: & = &
757: \frac{1}{(2\pi)^2}\frac{1}{\abs{kk_0}^3}
758: \left\lbrack\abs{\frac{k}{k_0}}^{1-2\gamma}-\frac{\sin^2(\gamma\pi)}
759: {16\pi^2\gamma^2(1-\gamma)^2}\frac{\Gamma^4(1+\gamma)\Gamma^2\left(\frac{3}{2}
760: -\gamma\right)}{\Gamma^2\left(\frac{1}{2}+\gamma\right)}\abs{\frac{q^2}
761: {16kk_0}}^{2\gamma-1}+(\gamma\to 1-\gamma)\right\rbrack.\nonumber
762: \end{eqnarray}
763:
764: In the distributed product, we clearly see that, taking the limit $q\to 0$, we
765: recover an expression depending only on the ratio
766: between the momenta of the two dipoles
767: \begin{equation}\label{eq:kkq_case10}
768: f^\gamma(k,k_0,q)\stackrel{q\to
769: 0}{\longrightarrow}\frac{1}{(2\pi)^2}\frac{1}{\abs{kk_0}^3}\left(\abs{\frac{k}
770: {k_0}}^{1-2\gamma}+\abs{\frac{k}{k_0}}^{2\gamma-1}
771: \right).
772: \end{equation}
773:
774: \item \underline{Case 2}: $k\gg q\gg k_0$.\\
775: At intermediate values of $q$, the dependence on $k_0$ goes away and we recover
776: an analytic power behaviour in $\gamma$ which, as we
777: shall see in the next section, will lead to traveling waves and geometric
778: scaling
779: \begin{eqnarray}
780: f^\gamma(k,k_0,q) & = &
781: \frac{-1}{2\pi^4}\sin(2\gamma\pi)\Gamma^2\left(\frac{3}{2}-\gamma\right)\Gamma^2
782: \left(\frac{1}{2}+\gamma\right)\nonumber\\
783: && \log\abs{\frac{q}{k_0}}\frac{1}{\abs{kq}^3}\left\lbrack
784: \frac{\Gamma^2\left(1+\gamma\right)}{\Gamma^2\left(\frac{1}{2}+\gamma\right)}
785: \abs{\frac{q}{4k}}^{2\gamma-1} -
786: \frac{\Gamma^2\left(2-\gamma\right)}{\Gamma^2\left(\frac{3}{2}-\gamma\right)}
787: \abs{\frac{q}{4k}}^{1-2\gamma}
788: \right\rbrack.\label{eq:kkq_case2}
789: \end{eqnarray}
790:
791: \item \underline{Case 3}: $q\gg k\gg k_0$.\\
792: This limit gives
793: \begin{equation}\label{eq:kkq_case3}
794: f^\gamma(k,k_0,q) =
795: \frac{-4}{\pi^4}\cos^2(\gamma\pi)\Gamma^2\left(\frac{3}{2}-\gamma\right)\Gamma^2
796: \left(\frac{1}{2}+\gamma\right)\gamma^2(1-\gamma)^2
797: \frac{1}{\abs{q}^6}\log\abs{\frac{q}{k_0}}\log\abs{\frac{q}{k}}.
798: \end{equation}
799: %Note that this expression is positive since, for $\gamma = \frac{1}{2}+i\nu$,
800: %we have $-\cos^2(\gamma\pi) = \sinh^2(\nu\pi)$.
801:
802: \end{itemize}
803:
804: \section{Traveling waves at nonzero transfer}\label{sec:tw}
805:
806: In this section, we investigate the consequences of our formul\ae\ for linear
807: BFKL dynamics on asymptotic solutions when the
808: non-linear effects of the BK equation are switched on. The key point is that
809: linear BFKL dynamics can provide solutions for the
810: linear part of the BK equation. Indeed, when factorisation between the target
811: and the projectile is fulfilled, the impact factor of
812: the target can be integrated out keeping a linear evolution governed by the same
813: kernel. It is thus a solution of the linear part of
814: the BK equation which should depend only on the kinematic variables of the
815: projectile.
816:
817: In the next step, the asymptotic solutions of the BK equation can be deduced
818: from the knowledge of its linear part, using quite
819: general arguments which were recalled in Sections \ref{sec:intro} and
820: \ref{sec:bindep}.
821: For each of the kinematical configurations introduced in the previous section,
822: we check whether the conditions for the existence of
823: traveling waves are fulfilled and, in these cases, derive their expression.
824:
825: Let us first concentrate on the full momentum representation \eqref{eq:amplkkq}.
826: We define the function
827: \begin{equation}\label{eq:bkfkq}
828: f(\vk,\vq) = \int d^2k_0\,\Phi_T(\vk_0,\vq)\int\frac{d\gamma}{2i\pi}\,e^{\bar
829: \alpha\chi(\gamma)Y} f^\gamma(\vk,\vk_0,\vq),
830: \end{equation}
831: depending only on $\vk$ and $\vq$. It is remarkable that in this representation,
832: the exact factorisation property of
833: $f^\gamma(\vk,\vk_0,\vq)$ (see \eqref{eq:fkkqdef}) gives rise to the simple
834: expression
835: \begin{equation}\label{eq:kkqfkq}
836: f(\vk,\vq) = \int\frac{d\gamma}{2i\pi}\,e^{\bar
837: \alpha\chi(\gamma)Y}\phi^\gamma(\vq)\,f^\gamma(\vk,\vq),
838: \end{equation}
839: where
840: \[
841: \phi^\gamma(\vq) = \int d^2k_0\, \frac{d^2\rho_0}{(2\pi)^2}\,
842: e^{i(\vk_0-\vq/2).\vrp}\,\frac{\abs{\vrp}}{4}\bar
843: E_q^\gamma(\vrp)\Phi_T(\vk_0,\vq)
844: \]
845: factorises out the target dependence, defining the initial condition, and
846: \begin{eqnarray}
847: \lefteqn{f^\gamma(\vk, \vq)
848: = \int \frac{d^2\rho}{(2\pi)^2}\,
849: e^{i(\vk-\vq/2).\vrh}\,\frac{\abs{\vrh}}{4}E_q^\gamma(\vrh)}\\
850: &&=
851: \frac{2^{4-6\gamma}\,\Gamma^2\!\left(\frac{3}{2}-\gamma\right)}{(2\pi)^2}
852: \sin(\gamma\pi)\frac{\abs{q}^{2\gamma-1}}{\abs{k}^3} %\nonumber\\
853: %& \times &
854: \left\lbrack
855: \frac{\Gamma^2\!\left(1+\gamma\right)}{\Gamma^2\!\left(\frac{1}{2}+\gamma\right)
856: }\abs{\frac{q}{4k}}^{2\gamma-1}\hyper\!\left(\gamma+
857: 1,\gamma;2\gamma;\frac{q}{k}\right)\hyper\!\left(\gamma+1,\gamma;2\gamma;
858: \frac{\bar q}{\bar k}\right)-(\gamma\to 1-\gamma)
859: \right\rbrack\nonumber
860: \end{eqnarray}
861: is calculated in Appendix \ref{ap:sl2c} and enters in formula \eqref{eq:fkkpq}.
862: It is obvious from the structure of \eqref{eq:bkfkq}, a linear superposition of
863: eigenfunctions of the BFKL kernel, that it provides
864: a natural basis for the solutions of the linear part of the BK equation in
865: momentum space.
866:
867: Therefore, we look for the kinematical domains where $f^\gamma(\vk, \vq)$ takes
868: the appropriate exponential behaviour, corresponding
869: to the wave superposition property \eqref{eq:waves} discussed in section
870: \ref{sec:bindep}. It will lead to traveling-wave solutions
871: for the full BK equation.
872:
873: We observe that, both in the cases 1 and 2 (formul\ae\ \eqref{eq:kkq_case10} and
874: \eqref{eq:kkq_case2}), {\em i.e.} in the limit
875: $k\!\gg\!q$, the solution of the linear part of the BK equation can be recast,
876: using the symmetry $\gamma \leftrightarrow 1-\gamma$,
877: under the form
878: \[
879: f(\vk,\vq) = \frac{1}{k^2}
880: \int\frac{d\gamma}{2i\pi}\,\phi^\gamma(\vq)\,e^{\bar\alpha\chi(\gamma)Y-\gamma
881: L},
882: \]
883: where
884: \[
885: L=\log\left(\frac{k^2}{q^2}\right)
886: \]
887: expresses the leading power-like behaviour at large $k/q$ and all remaining
888: factors in $\gamma$ and $\vq$ has been reabsorbed in
889: $\phi^\gamma(\vq)$. This clearly emphasises the fact that the region of interest
890: is $k\!\gg\!q$, independently of $k_0$. By
891: contrast, in the case 3 treating the large-$q$ expression, we see that equation
892: \eqref{eq:kkq_case3}, though being factorised, does
893: not fulfil the same property. We thus do not expect traveling-wave solutions in
894: this limit.
895:
896: Let us now include the effect of the nonlinear term in the BK equation. Due to
897: colour transparency ($f(\vk, \vq)\sim k^{-2}$ for
898: $k\gg q$), one can apply the general arguments exposed in Section
899: \ref{sec:bindep}. Therefore, the solution of the BK equation
900: reaches asymptotically the traveling-wave structure and exhibits geometric
901: scaling. The saturation scale
902: \begin{eqnarray}\label{eq:newqs}
903: Q_s^2(Y) & = & q^2 \Omega_s^2(Y) \nonumber \\[-3mm]
904: &&\\
905: &\sim& q^2
906: \exp\left[\bar\alpha\frac{\chi(\gamma_c)}{\gamma_c}Y-\frac{3}{2\gamma_c}\log(Y)
907: \right]\nonumber
908: \end{eqnarray}
909: has the same rapidity dependence as in the forward case but is proportional to
910: the momentum transfer. The critical exponent
911: $\gamma_c$ is still given by equation \eqref{eq:speed}. The asymptotic form of
912: the amplitude, {\em i.e.} the wavefront, can be
913: written at small $q/k$
914: \begin{equation}\label{eq:newfront}
915: f(\vk,\vq) \stackrel{Y\to\infty}{\sim}
916: \frac{1}{k^2}\,\phi^{\gamma_c}(\vq)\,\log\left(\frac{k^2}{q^2\Omega_s^2(Y)}
917: \right)\,\abs{\frac{k^2}{q^2\Omega_s^2(Y)}}^{-\gamma_c}\,
918: \exp\left[-\frac{1}{2\bar\alpha\chi''(\gamma_c)Y}\log^2\left(
919: \frac{k^2}{q^2\Omega_s^2(Y)}\right)\right].
920: \end{equation}
921:
922: It is interesting to see how we recover the forward limit $\vq\to 0$ recalled in
923: Section \ref{sec:bindep}. This can be done
924: remarking that formula \eqref{eq:kkq_case1}, obtained for BFKL dynamics in the
925: case 1 ($k\gg k_0\gg q$), introduces and additional
926: factor $(q/k_0)^{-2\gamma}$. By recombination of the $q/k$ and $q/k_0$ factors,
927: a different factorisation appear where $k_0$
928: substitute to $q$ as the reference scale, as clearly seen in
929: \eqref{eq:kkq_case10}. Inserting \eqref{eq:kkq_case10} in
930: \eqref{eq:bkfkq}, by straightforward algebra, it is easy to see that the result
931: can be cast under the form
932: \[
933: f(\vk,\vq\to 0) = \frac{1}{k^2}
934: \int\frac{d\gamma}{2i\pi}\,\phi^\gamma(Q_T)\,e^{\bar\alpha\chi(\gamma)Y-\gamma
935: L_0},
936: \]
937: where
938: \[
939: L_0=\log\left(\frac{k^2}{Q_T^2}\right),
940: \]
941: and $Q_T$ is a scale typical for the target, defining the initial condition for
942: the traveling-wave solution. This corresponds to the
943: solutions of equation \eqref{eq:bkb}.
944:
945: This discussion can be translated to the amplitude in the mixed space
946: representation of section \ref{sec:coord} defining similarly
947: \[
948: f(\vrh,\vq) = \int \frac{d^2\rho_0}{(2\pi)^2}\,\Phi_T(\vrp,
949: \vq)\int\frac{d\gamma}{2i\pi}e^{\bar\alpha\chi(\gamma)Y}f_q^\gamma(\vrh,
950: \vrp),
951: \]
952: where $f_q^\gamma(\vrh, \vrp)$, given by \eqref{eq:fqgamma}, is the solution of
953: the BFKL equation in the mixed representation.
954: As for the case of the momentum space, the factorisation of the target and
955: projectile dependences in the BFKL amplitude is the
956: important property, leading to traveling waves when we include the nonlinear
957: effects in the limit $\rho \ll 1/q$.
958:
959: The situation is not as simple in coordinate space. Indeed, it is hard to
960: introduce an amplitude depending only on $\vrh$ and $\cb$
961: from equation \eqref{eq:amplrrb}. This is closely related to the fact that
962: $f^\gamma(\vrh, \vrp,\cb)$, depending only on the
963: anharmonic ratio, cannot be factorised. In the limit where the BK equation is
964: valid ($\rho_0\!\gg\!b,\rho$), its linear part gives
965: solutions of the form (see \eqref{eq:rrbb}):
966: \[
967: f(\vrh,\cb)=\int\frac{d\gamma}{2i\pi}
968: \frac{2c_\gamma }{\gamma^2(1-\gamma)^2}\,e^{\bar\alpha\chi(\gamma)Y}\,
969: \left(\frac{4\rho}{\rho_0}\right)^{2\gamma},
970: \]
971: where $\rho_0$ is the typical size of the target. As explained in Section
972: \ref{sec:bindep}, this leads to traveling-wave solutions
973: for the BK equation ${\cal N}(\rho,Y)\!=\!{\cal N}(\rho\ Q_s(Y))$, where all
974: trace of the impact parameter has disappeared. Hence,
975: it gives no information upon the $b$ dependence. We could instead consider
976: equation \eqref{eq:rrb_mean} when the scaling variable
977: $\rho\rho_0/|\rho_0^2-4b^2|$ is large, {\em i.e.} $\rho_0\sim 2b$. However, it
978: is not clear whether the BK equation makes sense
979: physically in that limit where the dipole hits the target in its peripheral
980: region ($\rho_0\sim 2b$). As explained above, these
981: difficulties arise from the fact that $\rho$ and $\rho_0$ are mixed
982: non-trivially in the anharmonic ratio \eqref{eq:ratio}. Hence,
983: an interpretation in term of the BK equation is problematic. By contrast, the
984: situation in mixed and momentum spaces does not suffer
985: this inconvenient, due to the nice factorisation property in $\rho$ and
986: $\rho_0$, or in $k$ and $k_0$.
987:
988: \section{Conclusions and perspectives}\label{sec:ccl}
989:
990: Let us first emphasise the main results of this paper. We show that
991: traveling-wave solutions of the forward BK equation can be
992: extended to the full equation including nonzero transfer $q$, provided that
993: $k\gg q$, where $k$ represents the scale of the
994: projectile. The saturation scale \eqref{eq:newqs} has the same energy dependence
995: than in the forward case but is now proportional to
996: $q$. When this scale is large w.r.t. the scale, {\em e.g.} $Q_T$, characterising
997: the target, the saturation scale thus becomes
998: independent of the details of the target. But, when the momentum transfer $q$ is
999: small w.r.t. the target scale, $Q_T$ substitutes to
1000: $q$ in the saturation scale as expected from the forward case analysis.
1001: Our results show that non-forward BKFL evolution with momemtum transfer $q$ in a
1002: presence of saturation at the scale $\Omega_s(Y)$ is equivalent to forward BKFL
1003: evolution in the presence of saturation at the scale $q\Omega_s(Y)/Q_T.$
1004: Note that the similarity of a momentum transfer with an absorptive boundary in
1005: BFKL evolution has been noticed in Ref.\cite{mueltri}.
1006:
1007: As a consequence of the existence of traveling-wave solutions, we derive the
1008: asymptotic form of the solution, {\em i.e.} the
1009: wavefront \eqref{eq:newfront}, at small $q/k$. It appears as an expression of
1010: the same form as in the forward case with two
1011: modifications: the saturation scale is now the proportional to $q$ and the
1012: pre-factor, related to the initial conditions, is now
1013: $q$-dependent.
1014:
1015: From these considerations, we can predict that the BK equation implies the
1016: extension of geometric scaling at nonzero momentum
1017: transfer. Indeed, using formula \eqref{eq:amplkkq} and noting that, at large
1018: $k$, the impact factor of the projectile is expected to
1019: scale with the ratio $k/Q$ where $Q$ is a typical hard scale of the projectile.
1020: We see that the amplitude satisfies the geometric
1021: scaling under the form
1022: \begin{equation}
1023: {\cal{A}}(s,q^2,Q^2) \propto is\,\phi^{\gamma_c}(q)
1024: f\left(\frac{Q^2}{q^2\Omega^2_s(Y)}\right).
1025: \end{equation}
1026:
1027: All these results apply also in the mixed-space representation depending on the
1028: projectile size $\rho$ and the momentum transfer
1029: $q$. However, we find that these properties seem not to be easily expressed in
1030: terms of the impact parameter.
1031:
1032: Technically, the method used consists in three main steps. Firstly, we analyse
1033: the form of the BFKL solutions in full phase space
1034: and in different pertinent limits. Secondly, we use conformal-invariant
1035: properties of the BFKL dynamics to look for factorised
1036: solutions which then lead to solutions of the linear part of the BK equation.
1037: Thirdly, we infer the traveling-wave solutions,
1038: induced by nonlinear effects, in the kinematical regions and for initial
1039: conditions where the universality properties of the BK
1040: equation apply.
1041:
1042: Some comments are in order. Our formula \eqref{eq:newfront} suggests a solution
1043: to the puzzling problem of the compatibility of the
1044: BK equation with the confinement scale
1045: \cite{Kovner:2001bh,Ferreiro:2002kv,Bondarenko:2003ym,Golec-Biernat:2003ym,
1046: Gotsman:2004ra,Ikeda:2004zp}. The key point is to address
1047: the problem in momentum space where the non-perturbative dependence is
1048: factorised as clearly seen from the factor
1049: $\phi^{\gamma_c}(q)$ in equation \eqref{eq:newfront}, which may be characterised
1050: by a confinement scale. Fourier-transforming this
1051: result back to impact parameter space will break this factorisation property.
1052:
1053: The properties of the BK solutions at nonzero momentum transfer suggest to
1054: analyse the high-energy behaviour of suitable experimental
1055: processes. The electroproduction of $\rho$-mesons at nonzero transfer has been
1056: already analysed in impact parameter space
1057: \cite{Munier:2001nr} and our solutions of the BK equation incite to perform the
1058: analysis in momentum space.
1059:
1060: \begin{acknowledgments}
1061: The authors would like to thank Rikard Enberg for encouraging discussions,
1062: Krzysztof Golec-Biernat for reading the manuscript and Henri Navelet for
1063: correcting a crucial sign error. G.S. is funded by the National Funds for
1064: Scientific Research (Belgium).
1065: \end{acknowledgments}
1066:
1067: \begin{appendix}
1068:
1069: \section{Calculation of $f^\nu(\vk,\vkp,\vq)$}\label{ap:fourier}
1070:
1071: In this appendix, we compute the Fourier transform of
1072: \begin{equation}\label{eq:enuqdef}
1073: E_q^\nu(\rho) = 2^{3\mu} \abs{q}^{-2\mu} \Gamma^2(1+\mu)
1074: \left[J_\mu\left(\frac{\bar q \rho}{4}\right)J_\mu\left(\frac{q
1075: \bar\rho}{4}\right) - J_{-\mu}\left(\frac{\bar q
1076: \rho}{4}\right)J_{-\mu}\left(\frac{q \bar\rho}{4}\right) \right],
1077: \end{equation}
1078: with $\mu = -i\nu$. It is of course sufficient to compute the following integral
1079: \[
1080: I = \int d\rho\,d\phi\,\rho^{2+\alpha} e^{i\rho
1081: v\cos(\phi)}J_\mu\left(\frac{\abs{q}
1082: \rho}{4}e^{i(\phi-\psi)}\right)J_\mu\left(\frac{\abs{q}
1083: \rho}{4}e^{i(\psi-\phi)}\right),
1084: \]
1085: where $v = \abs{k-q/2}$ and $\phi$ (resp. $\psi$) is the angle between $k-q/2$
1086: and $\rho$ (resp. $q$). This integral has been
1087: regularised at infinity by introducing a factor $\rho^\alpha$ with
1088: $-2<\alpha<-3/2$. If we expand the Bessel functions in series and
1089: use
1090: \[
1091: \int_0^{2\pi} d\phi\,e^{i[\rho v \cos(\phi)-m\phi]} = 2\pi e^{i\pi m/2} J_m(\rho
1092: v),
1093: \]
1094: we find
1095: \[
1096: I = (2\pi)\sum_{j,k=0}^\infty
1097: \frac{\left(\frac{\abs{q}e^{i\psi}}{8}\right)^{\mu+2j}\left(
1098: \frac{\abs{q}e^{-i\psi}}{8}\right)^{\mu+2k}}{j!\, k!\, \Gamma(1+\mu+j)
1099: \Gamma(1+\mu+k)} \int_0^\infty d\rho\,\rho^{2+\alpha+2\mu+2j+2k}J_{2(j-k)}
1100: (\rho v).
1101: \]
1102: The integration over $\rho$ can be performed using
1103: \[
1104: \int_0^\infty dt\,t^{\beta-1}J_m(zt) =
1105: \frac{1}{2\pi}\left(\frac{2}{z}\right)^\beta\Gamma\left(\frac{\beta+m}{2}\right)
1106: \Gamma\left(\frac{\beta-m}{2}\right)\sin\left[\left(
1107: \frac{\beta-m}{2}\right)\pi\right].
1108: \]
1109: This leads to a factorisation of the integral in conformal blocs:
1110: \[
1111: I = \sin\left[\left(\frac{3+\alpha+2\mu}{2}\right)\pi\right] f(k,q)f(\bar k,
1112: \bar q),
1113: \]
1114: where, using the doubling formula,
1115: \begin{eqnarray*}
1116: f(k,q) & = & \sum_{j=0}^\infty
1117: \frac{1}{j!}\frac{\Gamma\left(\frac{3}{2}+\frac{\alpha}{2}+\mu+2j\right)}{\Gamma
1118: (1+\mu+j)}\left(\frac{\abs{q}
1119: e^{i\psi}}{8}\right)^{\mu+2j}\left(\frac{2}{v}\right)^{\frac{3}{2}+\frac{\alpha}
1120: {2}+\mu+2j} \\
1121: & = & \left(\frac{q}{4k-2q}\right)^\mu
1122: \left(\frac{2}{\abs{k-q/2}}\right)^{\frac{3}{2}+\frac{\alpha}{2}}
1123: \frac{\Gamma\left(\frac{3}{2}+\frac{\alpha}{2}+\mu\right)}{\Gamma(1+\mu)}
1124: \hyper\left(\frac{3}{4}+\frac{\alpha}{4}+\frac{\mu}{2},
1125: \frac{5}{4}+\frac{\alpha}{4}+\frac{\mu}{2};1+\mu;\left(\frac{q}{2k-q}\right)^2
1126: \right).
1127: \end{eqnarray*}
1128: The hypergeometric function can be simplified (see {\em e.g.} Ref.\cite{math}),
1129: and, taking $\alpha$ to zero, we finally obtain
1130: \begin{equation}
1131: I = -\cos(\mu\pi)\frac{\Gamma^2\left(\frac{3}{2}+\mu\right)}{\Gamma^2(1+\mu)}
1132: \left(\frac{2}{\abs{k}}\right)^3
1133: \abs{\frac{q}{4k}}^{2\mu}\hyper\left(\frac{3}{2}+\mu,\frac{1}{2}+\mu;1+2\mu;
1134: \frac{q}{k}\right)\hyper\left(\frac{3}{2}+\mu,\frac{1}{2}+\mu;1+2\mu;
1135: \frac{\bar{q}}{\bar{k}}\right).
1136: \end{equation}
1137: The final result follows from adding to $I$ the corresponding expression with
1138: $\mu\to-\mu$ and using \eqref{eq:enuqdef}.
1139:
1140: \section{Fourier transform of $E_q^\nu(\rho)$}\label{ap:sl2c}
1141:
1142: In this appendix, we show how the Fourier transform of $E_q^\nu(\vrh)$ can be
1143: calculated using conformal invariance. That will
1144: exhibit the link between $f^\nu(\vk,\vkp,\vq)$ and matrix elements of
1145: $SL(2,\mathbb{C})$ \cite{Janik:1999fk}. One has to deal with
1146: \[
1147: \int d^2\rho\, e^{i(\vk-\vq/2).\vrh} \frac{\abs{\vrh}}{4} E_q^\nu(\vrh)=
1148: \frac{c_\gamma}{\pi^2}\int d^2\rho_1 d^2\rho_2\
1149: e^{i\vq.\vrh_2}e^{i\vk.(\vrh_1-\vrh_2)}
1150: |\vrh_1-\vrh_2|^{2\gamma}|\vrh_1|^{-2\gamma}|\vrh_2|^{-2\gamma},
1151: \]
1152: where $c_\gamma$ is given by \eqref{eq:cgamma}.
1153:
1154: To compute this integral, one can switch to complex coordinates and perform the
1155: following changes of variables: first
1156: $\rho_2\!\rightarrow\!u\!=\!\rho_2/\rho_1,$ then
1157: $\rho_1\!\rightarrow\!w\!=\!\rho_1(\bar k\!+\!u(\bar q\!-\!\bar k))$, and
1158: finally
1159: $u\!\rightarrow\!v\!=\!u/(u\!-\!1).$ One obtains
1160: \[
1161: \int d^2\rho\, e^{i(\vk-\vq/2).\vrh} \frac{\abs{\vrh}}{4} E_q^\nu(\vrh)=
1162: \frac{c_\gamma|\vk|^{2\gamma-4}}{\pi^2}\int d^2w\ e^{i Re(w)}|w|^{2-2\gamma}
1163: \int d^2v |v|^{-2\gamma} |1\!-\!v|^{-2\gamma}|1\!-\!v\bar q/\bar
1164: k|^{2\gamma-4}\,,
1165: \]
1166: Then, using
1167: \[
1168: \int d^2w\ e^{i Re(w)}|w|^{2-2\gamma}=
1169: \pi 2^{4-2\gamma}\frac{\Gamma(2-\gamma)}{\Gamma(\gamma-1)}\ .
1170: \]
1171: and the following result~\cite{gernav}
1172: \begin{eqnarray*}
1173: \int d^2v\ |v|^{2a_1-2} |1-v|^{2b_1-2a_1-2}|1-vx|^{-2a_0}=\pi
1174: \left[\frac{\Gamma(a_1)\Gamma(b_1\!-\!a_1)\Gamma(1\!-\!b_1)}
1175: {\Gamma(1\!-\!a_1)\Gamma(1\!-\!b_1\!+\!a_1)\Gamma(b_1)}
1176: {}_2 F_1(a_0,a_1;b_1;x){}_2 F_1(a_0,a_1;b_1;\bar x)+\right.\\
1177: \left.+\frac{\Gamma(b_1\!-\!1)\Gamma(1\!-\!a_0)\Gamma(1\!-\!b_1\!+\!a_0)}
1178: {\Gamma(2\!-\!b_1)\Gamma(a_0)\Gamma(b_1\!-\!a_0)}
1179: |x|^{2\!-\!2b_1}{}_2 F_1(a_0\!-\!b_1\!+\!1,a_1\!-\!b_1\!+\!1;2\!-\!b_1;x)
1180: {}_2 F_1(a_0\!-\!b_1\!+\!1,a_1\!-\!b_1\!+\!1;2\!-\!b_1;\bar x)\right],
1181: \end{eqnarray*}
1182: one easily recovers % the result \eqref{eq:fkkpq}.
1183: \begin{eqnarray*}
1184: \lefteqn{\int d^2\rho\, e^{i(\vk-\vq/2).\vrh} \frac{\abs{\vrh}}{4}
1185: E_q^\nu(\vrh)=
1186: \frac{1}{4}\abs{q}^{2i\nu}2^{3-6i\nu}\Gamma^2(1-i\nu)\cos(i\nu\pi)\frac{1}
1187: {\abs{k}^3}}\\
1188: &&\left\lbrack\frac{\Gamma^2\left(\frac{3}{2}+i\nu\right)}{\Gamma^2(1+i\nu)}
1189: \abs{\frac{q}{4k}}^{2i\nu}\hyper\left(\frac{3}{2}+i\nu,\frac{1}{2}+i\nu;1+2i\nu;
1190: \frac{q}{k}\right)\hyper\left(\frac{3}{2}+i\nu,\frac
1191: {1}{2}+i\nu;1+2i\nu;\frac{\bar{q}}{\bar{k}}\right)-(\nu\to-\nu)\right\rbrack.
1192: \end{eqnarray*}
1193: The same integral appears in the calculation of matrix elements of
1194: $SL(2,\mathbb{C})$ \cite{Janik:1999fk} and more generally in
1195: conformal field theory \cite{gernav}.
1196:
1197:
1198: \section{Asymptotic expansion at large $q$}\label{ap:asymp}
1199:
1200: We want to emphasise the main steps yielding the asymptotic behaviour
1201: \eqref{eq:2f1exp}. We shall start by the asymptotic expansion
1202: of the hypergeometric function at infinity ($z = q/k \gg 1$)
1203: \begin{eqnarray*}
1204: \hyper(\gamma,\gamma+1;2\gamma;z) & = &
1205: \frac{2^{2\gamma-1}}{\sqrt{\pi}}\frac{\Gamma\left(\frac{1}{2}+\gamma\right)}
1206: {\Gamma(1+\gamma)}(-z)^{-\gamma}
1207: \left\{
1208: 1-\gamma(1-\gamma)\frac{\log(-z)}{z}
1209: \hyper\left(\gamma+1,2-\gamma;2;\frac{1}{z}\right)\right.\\
1210: &&\left.-\sum_{k=0}^\infty
1211: \frac{(\gamma)_{k+1}(1-\gamma)_{k+1}}{k!\,(k+1)!}z^{-k-1}
1212: \left[
1213: \psi(k+1)+\psi(k+2)-\psi(\gamma+k+1)-\psi(\gamma-k-1)
1214: \right]
1215: \right\}.
1216: \end{eqnarray*}
1217: When combining the $z$ and $\bar{z}$ parts, most of the pre-factors cancel:
1218: \[
1219: \frac{\Gamma^2(1+\gamma)}{\Gamma^2\left(\frac{1}{2}+\gamma\right)}
1220: \abs{\frac{z}{4}}^{2\gamma-1}\hyper(\gamma,\gamma+1;2\gamma;z)\hyper(\gamma,
1221: \gamma+1;2\gamma;\bar{z}) =
1222: \frac{1}{\pi\abs{z}}\left[S_\gamma(z)+A_\gamma(z)\right]\left[S_\gamma(\bar
1223: z)+A_\gamma(\bar z)\right],
1224: \]
1225: where
1226: \begin{eqnarray*}
1227: S_\gamma(z) & = &
1228: 1-\gamma(1-\gamma)\frac{\log(-z)+\psi(1)+\psi(2)}{z}+{\cal{O}}(1/z^2), \\
1229: A_\gamma(z) & = &
1230: \frac{1}{z}\left\{\gamma(1-\gamma)\left[\psi(\gamma+1)+\psi(\gamma-1)\right]+
1231: (\gamma+1)\gamma(\gamma-1)(\gamma-2)\frac{\psi(\gamma+2)
1232: +\psi(\gamma-2)}{z}+{\cal{O}}(1/z^2)\right\}.
1233: \end{eqnarray*}
1234: In this splitting we have put in $S_\gamma$ the terms which are invariant under
1235: the replacement $\gamma\to 1-\gamma$ and the
1236: remaining ones in $A_\gamma$. Therefore, if we consider the full $z$-dependent
1237: term, we easily get
1238: \begin{eqnarray*}
1239: \lefteqn{\frac{\Gamma^2(1+\gamma)}{\Gamma^2\left(\frac{1}{2}+\gamma\right)}
1240: \abs{\frac{z}{4}}^{2\gamma-1}\hyper(\gamma,\gamma+1;2\gamma;z)\hyper(\gamma,
1241: \gamma+1;2\gamma;\bar{z}) - (\gamma\to 1-\gamma)}\\
1242: & = & \frac{1}{\pi\abs{z}} \left\{
1243: S_\gamma(z)\left\lbrack A_\gamma(\bar z)-A_{1-\gamma}(\bar z)\right\rbrack
1244: + S_\gamma(\bar z)\left\lbrack A_\gamma(z)-A_{1-\gamma}(z)\right\rbrack
1245: + \abs{A_\gamma(z)}^2 - \abs{A_{1-\gamma}(z)}^2\right\}.
1246: \end{eqnarray*}
1247: Obviously, the leading term in that development is proportional to
1248: $z^{-1}+\bar{z}^{-1}$, {\em i.e.} to $\frac{k}{q}+\frac{\bar
1249: k}{\bar q}$, which vanishes if we average over the angle between the two
1250: vectors. With less restrictions, one can also remark that
1251: this term is antisymmetric under the replacement $k\to -k$.
1252:
1253: We thus go to the next order which finally gives a leading contribution of the
1254: form
1255: \[
1256: \frac{1}{\pi}\gamma^2(1-\gamma)^2\left[\psi(2-\gamma)+\psi(-\gamma)-\psi(\gamma+
1257: 1)-\psi(\gamma-1)\right]\frac{\log\left(\abs{z}^2\right)}{\abs{z}^3} =
1258: 2\gamma^2(1-\gamma)^2\frac{\cos(\gamma\pi)}{\sin(\gamma\pi)}\frac{\log\left(
1259: \abs{z}^2\right)}{\abs{z}^3}.
1260: \]
1261:
1262: \end{appendix}
1263:
1264: \begin{thebibliography}{99}
1265:
1266: \bibitem{Stasto:2000er}
1267: A.~M. Sta\'sto, K.~Golec-Biernat, and J.~Kwiecinski,
1268: \newblock Phys. Rev. Lett. {\bf 86}, 596 (2001) [arXiv:hep-ph/0007192].
1269: %%CITATION = HEP-PH 0007192;%%
1270:
1271: \bibitem{endnote17}{For a review on saturation and references, see A.H.~Mueller,
1272: ``Parton saturation: An overview'', arXiv:hep-ph/0111244.}
1273:
1274: \bibitem{Nikolaev:1991ja}
1275: N.~N. Nikolaev and B.~G. Zakharov,
1276: \newblock Z. Phys. {\bf C49}, 607 (1991);
1277: %%CITATION = ZEPYA,C49,607;%%
1278: \newblock A.~H. Mueller,
1279: \newblock Nucl. Phys. {\bf B415}, 373 (1994).
1280: %%CITATION = NUPHA,B415,373;%%
1281:
1282: \bibitem{Balitsky}
1283: I.~I.~Balitsky,
1284: %``Operator expansion for high-energy scattering,''
1285: Nucl.\ Phys.\ {\bf B463}, 99 (1996)
1286: [arXiv:hep-ph/9509348];
1287: %%CITATION = HEP-PH 9509348;%%
1288:
1289: \bibitem{Kovchegov}
1290: Y.~V. Kovchegov,
1291: \newblock Phys. Rev. {\bf D60}, 034008 (1999) [arXiv:hep-ph/9901281];
1292: %%CITATION = HEP-PH 9901281;%%
1293: Y.~V. Kovchegov,
1294: \newblock Phys. Rev. {\bf D61}, 074018 (2000), [arXiv:hep-ph/9905214].
1295: %%CITATION = HEP-PH 9905214;%%
1296:
1297: %\cite{Munier}
1298: \bibitem{Munier1}
1299: S.~Munier and R.~Peschanski,
1300: %``Geometric scaling as traveling waves,''
1301: Phys.\ Rev.\ Lett.\ {\bf 91}, 232001 (2003)
1302: [arXiv:hep-ph/0309177];
1303: %%CITATION = HEP-PH 0309177;%%
1304: %``Traveling wave fronts and the transition to saturation,''
1305: Phys.\ Rev.\ {\bf D69}, 034008 (2004)
1306: [arXiv:hep-ph/0310357];
1307: %%CITATION = HEP-PH 0309177;%%
1308: %``UNIVERSALITY AND TREE STRUCTURE OF HIGH-ENERGY QCD,''
1309: Phys.\ Rev.\ {\bf D70}, 077503 (2004)
1310: [arXiv:hep-ph/0310357].
1311:
1312: %\cite{Munier:2001nr}
1313: \bibitem{Munier:2001nr}
1314: S.~Munier, A.~M.~Stasto and A.~H.~Mueller,
1315: % ``Impact parameter dependent S-matrix for dipole proton scattering from
1316: %diffractive meson electroproduction,''
1317: Nucl.\ Phys.\ {\bf B603}, 427 (2001)
1318: [arXiv:hep-ph/0102291].
1319: %%CITATION = HEP-PH 0102291;%%
1320: S.~Munier and S.~Wallon,
1321: %``Geometric scaling in exclusive processes,''
1322: Eur.\ Phys.\ J.\ {\bf C30} (2003) 359
1323: [arXiv:hep-ph/0303211].
1324: %%CITATION = HEP-PH 0303211;%%
1325:
1326: %\cite{Kowalski:2003hm}
1327: \bibitem{Kowalski:2003hm}
1328: H.~Kowalski and D.~Teaney,
1329: %``An impact parameter dipole saturation model,''
1330: Phys.\ Rev.\ {\bf D68}, 114005 (2003)
1331: [arXiv:hep-ph/0304189].
1332: %%CITATION = HEP-PH 0304189;%%
1333:
1334: %\cite{Bondarenko:2003ym}
1335: \bibitem{Bondarenko:2003ym}
1336: S.~Bondarenko, M.~Kozlov and E.~Levin,
1337: %``QCD saturation in the semi-classical approach,''
1338: Nucl.\ Phys.\ {\bf A727}, 139 (2003)
1339: [arXiv:hep-ph/0305150].
1340: %%CITATION = HEP-PH 0305150;%%
1341:
1342: %\cite{Golec-Biernat:2003ym}
1343: \bibitem{Golec-Biernat:2003ym}
1344: K.~Golec-Biernat and A.~M.~Stasto,
1345: %``On solutions of the Balitsky-Kovchegov equation with impact parameter,''
1346: Nucl.\ Phys.\ {\bf B668}, 345 (2003)
1347: [arXiv:hep-ph/0306279].
1348: %%CITATION = HEP-PH 0306279;%%
1349:
1350: %\cite{Gotsman:2004ra}
1351: \bibitem{Gotsman:2004ra}
1352: E.~Gotsman, M.~Kozlov, E.~Levin, U.~Maor and E.~Naftali,
1353: % ``Towards a new global QCD analysis: Solution to the non-linear equation at
1354: %arbitrary impact parameter,''
1355: Nucl.\ Phys.\ {\bf A742}, 55 (2004)
1356: [arXiv:hep-ph/0401021].
1357: %%CITATION = HEP-PH 0401021;%%
1358:
1359: %\cite{Ikeda:2004zp}
1360: \bibitem{Ikeda:2004zp}
1361: T.~Ikeda and L.~McLerran,
1362: %``Impact parameter dependence in the Balitsky-Kovchegov equation,''
1363: arXiv:hep-ph/0410345.
1364: %%CITATION = HEP-PH 0410345;%%
1365:
1366: %\cite{Ferreiro:2002kv}
1367: \bibitem{Ferreiro:2002kv}
1368: E.~Ferreiro, E.~Iancu, K.~Itakura and L.~McLerran,
1369: %``Froissart bound from gluon saturation,''
1370: Nucl.\ Phys.\ {\bf 1710}, 373 (2002)
1371: [arXiv:hep-ph/0206241].
1372: %%CITATION = HEP-PH 0206241;%%
1373:
1374: %\cite{Kovner:2001bh}
1375: \bibitem{Kovner:2001bh}
1376: A.~Kovner and U.~A.~Wiedemann,
1377: %``Nonlinear QCD evolution: Saturation without unitarization,''
1378: Phys.\ Rev.\ {\bf D66}, 051502 (2002)
1379: [arXiv:hep-ph/0112140];
1380: Phys.\ Rev.\ {\bf D66}, 034031 (2002)
1381: [arXiv:hep-ph/0204277];
1382: Phys.\ Lett.\ {\bf B551}, 311 (2003)
1383: [arXiv:hep-ph/0207335].
1384: %%CITATION = HEP-PH 0207335;%%
1385:
1386: \bibitem{KPP}
1387: R.~A. Fisher,
1388: \newblock Ann. Eugenics {\bf 7}, 355 (1937);
1389: \newblock A.~Kolmogorov, I.~Petrovsky, and N.~Piscounov,
1390: \newblock Moscou Univ. Bull. Math. {\bf A1}, 1 (1937).
1391:
1392: \bibitem{Bramson}
1393: M.~Bramson,
1394: \newblock Memoirs of the American Mathematical Society {\bf 285} (1983).
1395:
1396: \bibitem{ebert}
1397: U. Ebert, W. van Saarloos,
1398: \newblock Physica {\bf D146}, 1 (2000) [arXiv:cond-mat/0003181];
1399: For a review, see W. van Saarloos,
1400: \newblock Phys. Rep. {\bf 386}, 29 (2003).
1401:
1402: \bibitem{brunet}
1403: E.~Brunet and B.~Derrida,
1404: \newblock Phys. Rev. {\bf E56}, 2597 (1997). See in particular the appendix for
1405: a derivation of the traveling wave solutions.
1406:
1407: \bibitem{Lipatov86}
1408: L.~N.~Lipatov,
1409: %``The Bare Pomeron In Quantum Chromodynamics,''
1410: Sov.\ Phys.\ JETP {\bf 63}, 904 (1986)
1411: [Zh.\ Eksp.\ Teor.\ Fiz.\ {\bf 90}, 1536 (1986)].
1412: %%CITATION = SPHJA,63,904;%%
1413:
1414: \bibitem{NP} H.~Navelet and R.~Peschanski,
1415: %``Conformal invariance and the exact solution of BFKL equations,''
1416: Nucl.\ Phys.\ {\bf B507}, 353 (1997)
1417: [arXiv:hep-ph/9703238].
1418: %%CITATION = HEP-PH 9703238;%%
1419:
1420: %\cite{Navelet:1997tx}
1421: \bibitem{Navelet:1997tx}
1422: H.~Navelet and S.~Wallon,
1423: %``Onium onium scattering at fixed impact parameter: Exact equivalence between
1424: %the color dipole model and the BFKL pomeron,''
1425: Nucl.\ Phys.\ {\bf B522}, 237 (1998)
1426: [arXiv:hep-ph/9705296].
1427: %%CITATION = HEP-PH 9705296;%%
1428:
1429: \bibitem{LipatovReview}
1430: L.~N.~Lipatov,
1431: Phys.\ Rept.\ {\bf 286}, 131 (1997)
1432: [arXiv:hep-ph/9610276].
1433: %%CITATION = HEP-PH 9610276;%%
1434:
1435: \bibitem{GLR} L.V. Gribov, E.M. Levin and M.G. Ryskin, {Phys. Rep.} {\bf
1436: 100}, 1 (1983).
1437:
1438: \bibitem{mueltri}
1439: A.~H.~Mueller and D.~N.~Triantafyllopoulos,
1440: %``The energy dependence of the saturation momentum,''
1441: Nucl.\ Phys.\ B {\bf 640} (2002) 331
1442: [arXiv:hep-ph/0205167].
1443: %%CITATION = HEP-PH 0205167;%%
1444:
1445: \bibitem{math} I.~S.~Gradshtein and I.~M.~Ryzhik, {\em Tables of integrals,
1446: series and products}, 1980, Academic Press, New-York.
1447:
1448: %\cite{Janik:1999fk}
1449: \bibitem{Janik:1999fk}
1450: R.~A.~Janik and R.~Peschanski,
1451: %``Conformal invariance and {QCD} pomeron vertices in the 1/N(c) limit,''
1452: Nucl.\ Phys.\ {\bf B549}, 280 (1999)
1453: [arXiv:hep-ph/9901426].
1454: %%CITATION = HEP-PH 9901426;%%
1455:
1456: \bibitem{gernav}
1457: Vl.S. Dotsenko and V.A. Fateev, {\it Nucl. Phys.} {\bf B240}, 312 (1984);
1458: J. Geronimo and H. Navelet, {\it J. Math. Phys.} {\bf 44}, 2293 (2003)
1459: [arXiv:math-ph/0003019.
1460:
1461: \end{thebibliography}
1462: \end{document}
1463: