1: \documentclass[fleqn,twoside]{article}

2: \usepackage{espcrc2}

3: \usepackage{graphicx}

4: 

5: 

6: % put your own definitions here:

7: \newcommand{\lsim}{\raisebox{-4pt}{$

8: \,\stackrel{\textstyle <}{\sim}\,$}}

9: \newcommand{\gsim}{\raisebox{-4pt}{$

10: \,\stackrel{\textstyle >}{\sim}\,$}}

11: 

12: \newcommand{\gev}{\mbox{~GeV}}

13: \newcommand{\mev}{\mbox{~MeV}}

14: 

15: \newcommand{\half}{{\textstyle\frac{1}{2}}}

16: 

17: % transverse vectors

18: \newcommand{\tvec}[1]{\mbox{\boldmath{$#1$}}}

19: \newcommand{\svec}[1]{\mbox{\boldmath{$\scriptstyle #1$}}}

20: 

21: % small space

22: \newcommand{\slim}{\mskip 1.5mu}

23: \newcommand{\abst}{|\slim t\slim|}

24: 

25: 

26: % add words to TeX's hyphenation exception list

27: \hyphenation{author another created paper re-commend-ed Post-Script

28:   Ra-dyush-kin}

29: 

30: 

31: % declarations for front matter

32: % for hep ph version:

33: \title{Generalized parton distributions from form factors%

34: \thanks{To appear in: Procs.\ of the Workshop on Light-Cone QCD and

35: Nonperturbative Hadron Physics 2005 (LC 2005), Cairns, Australia, 2005}

36: }

37: 

38: % for printed version:

39: %\title{Generalized parton distributions from form factors}

40: 

41: \author{M. Diehl\,\address{Deutsches Elektronen-Synchroton DESY, 22603

42:     Hamburg, Germany}}

43: 

44: 

45: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

46: 

47: \begin{document}

48: 

49: \begin{abstract}

50: The electromagnetic nucleon form factors provide constraints on

51: generalized quark distributions.  Key results of the study presented

52: here are a strong dependence of the average impact parameter of quarks

53: on their longitudinal momentum fraction, a striking difference in the

54: $t$ dependence of $u$ and $d$ quark contributions to elastic form

55: factors, and an estimate of the orbital angular momentum carried by

56: valence quarks in the nucleon.

57: \vspace{1pc}

58: \end{abstract}

59: 

60: % typeset front matter (including abstract)

61: \maketitle

62: 

63: 

64: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

65: 

66: \section{INTRODUCTION}

67: \label{sec:intro}

68: 

69: Much of what we know about hadron structure comes from measurements of

70: parton densities, which quantify the distribution of longitudinal

71: momentum and helicity of partons in a fast-moving hadron.  Generalized

72: parton distributions (GPDs) complement this essentially

73: one-dimensional picture with information in the plane perpendicular to

74: the direction of movement.  These distributions parameterize matrix

75: elements of non-local quark or gluon operators.  We focus here on the

76: unpolarized quark sector, where the distribution $H^q(x,\xi,t)$ is

77: diagonal in proton helicity and $E^q(x,\xi,t)$ describes proton

78: helicity flip.  The variables $x$ and $\xi$ parameterize longitudinal

79: quark momentum fractions relative to the average proton momentum

80: $\half (p+p')$ as shown in Fig.~\ref{fig:gpd}, whereas the invariant

81: $t=(p-p')^2$ depends on both longitudinal and transverse components of

82: the momentum transferred to the proton.  In the forward limit $p=p'$

83: one recovers the usual quark and antiquark densities as $q(x) =

84: H^q(x,0,0)$ and $\bar{q}(x) = -H^q(-x,0,0)$ with $x>0$.  For the

85: precise definitions of GPDs and further information we refer to the

86: recent reviews~\cite{Diehl:2003ny}.

87: 

88: \begin{figure}[htb]

89: \begin{center}

90: \includegraphics[width=0.3\textwidth]{gpd.eps}

91: \vspace{-3.5em}

92: \end{center}

93: %

94: \caption{\label{fig:gpd} Relevant variables in a GPD.}

95: \end{figure}

96: 

97: According to factorization theorems, GPDs appear in the scattering

98: amplitudes of suitable hard exclusive processes such as deeply virtual

99: Compton scattering, $\gamma^* p\to \gamma\slim p$, and exclusive meson

100: production, e.g.\ $\gamma^* p\to \rho\slim p$.  In these processes,

101: the longitudinal momentum transfer $\xi$ is fixed by the kinematics,

102: whereas $x$ is a loop variable.  To disentangle the $x$ and $\xi$

103: dependence of GPDs from measured process amplitudes remains an

104: outstanding task, but generically the dominating values of $x$ in the

105: loop integrals will be of order~$\xi$.

106: 

107: 

108: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

109: 

110: \section{IMPACT PARAMETER DENSITIES}

111: \label{sec:impact}

112: 

113: To represent GPDs in transverse position space, we form wave packets

114: \begin{equation}

115:   \label{impact-state}

116: |p^+, \tvec{b}\rangle = \int\frac{d^2\tvec{p}}{(2\pi)^2}\,

117:   e^{-i\svec{b} \svec{p}}\, |p^+, \tvec{p} \rangle

118: \end{equation}

119: from momentum eigenstates $|p^+, \tvec{p} \rangle$, where we write

120: $v^{\pm} = (v^0 \pm v^3) /\sqrt{2}$ for the light-cone components and

121: $\tvec{v} = (v^1, v^2)$ for the transverse part of a four-vector $v$.

122: The state $|p^+, \tvec{b}\rangle$ is localized at $\tvec{b}$ in the

123: transverse plane (often called impact parameter plane).  Formally it

124: is an eigenstate of a suitably defined transverse position operator

125: \cite{Soper:1972xc}.  It is thus possible to localize a relativistic

126: state exactly in \emph{two} dimensions, whereas localization in all

127: \emph{three} dimensions involves ambiguities at the level of the

128: Compton wavelength.  For a parton interpretation it is natural to

129: consider states $|p^+, \tvec{b}\rangle$ with large $p^+$, which

130: describe a fast-moving proton.  Further analysis reveals that

131: $\tvec{b}$ is the ``center of momentum'' of the partons in the proton,

132: given as $\tvec{b} = \sum_i p_i^+ \tvec{b}_i^{\phantom{+}} /\sum_i

133: p_i^+$ in terms of their plus-momenta and transverse positions.  The

134: center of momentum is related by Noether's theorem to transverse

135: boosts, in analogy to the relation between the center of mass and

136: Galilean transformations in nonrelativistic mechanics.

137: 

138: Taking matrix elements between impact parameter states

139: (\ref{impact-state}) of the quark or gluon operators defining general

140: parton distributions in momentum space, one obtains Fourier transforms

141: of these distributions.  For vanishing skewness parameter $\xi$ one

142: finds that

143: \begin{equation}

144: q(x,\tvec{b}) = \int \frac{d^2\tvec{\Delta}}{(2\pi)^2}\,

145:   e^{-i \svec{b} \svec{\Delta}}\,

146:   H^q(x,0,-\tvec{\Delta}^2)

147: \end{equation}

148: is the density of quarks with longitudinal momentum fraction $x$ and

149: transverse distance $\tvec{b}$ from the center of momentum of the

150: proton \cite{Burkardt:2000za}.  For nonzero $\xi$ one no longer has a

151: probability interpretation because the two quark momentum fractions in

152: Fig.~\ref{fig:gpd} are not the same, but $\tvec{b}$ still describes

153: the distribution of the struck quark in the transverse plane

154: \cite{Diehl:2002he}.  According to the discussion at the end of the

155: introduction, the combined $\xi$ and $t$ dependence of hard exclusive

156: scattering processes thus yields information about the impact

157: parameter distribution of partons with longitudinal momentum fraction

158: of order $\xi$.  Note that the connection between GPDs and spatial

159: distributions in the transverse plane discussed here differs from the

160: well-known representation of form factors in terms of spatial

161: distributions in \emph{three} dimensions, which has been extended to

162: GPDs in~\cite{Belitsky:2003nz}.

163: 

164: Just as the ordinary parton densities, the distributions

165: $q(x,\tvec{b})$ depend on the scale $\mu$ at which the partons are

166: resolved.  The scale evolution is local in $\tvec{b}$ and described by

167: the usual DGLAP equations.  For the valence quark distributions

168: $q_v(x,\tvec{b}) = q(x,\tvec{b}) - \bar{q}(x,\tvec{b})$ we have

169: \begin{equation}

170: \mu^2 \frac{d}{d\mu^2}\, q_v(x,\tvec{b}) 

171: = \int_x^1 \frac{dz}{z}\,

172:   \Big[ P\Big(\frac{x}{z}\Big) \Big]_+\, q_v(z,\tvec{b}) \,,

173: \end{equation}

174: where $P(z)$ denotes the quark splitting function.  As a consequence

175: the $\tvec{b}$ dependence of the distributions at given $x$ changes

176: with $\mu$.  In particular, the average squared impact parameter

177: $\langle \tvec{b}^2 \rangle_x$ for valence quark distributions evolves

178: as \cite{Diehl:2004cx}

179: \begin{eqnarray}

180:   \label{dglap-impact}

181: \lefteqn{

182: \mu^2 \frac{d}{d\mu^2}\, \langle \tvec{b}^2 \rangle_x 

183: }

184: \\

185: &=& \hspace{-0.8em} {}-\frac{1}{q_v(x)} \int_x^1 \frac{d z}{z}\,

186:   P\Big(\frac{x}{z}\Big)\, q_v(z) 

187:   \Big[ \langle \tvec{b}^2 \rangle_x 

188:       - \langle \tvec{b}^2 \rangle_z \Big] \,,

189: \nonumber

190: \end{eqnarray}

191: where we defined

192: \begin{eqnarray}

193:   \label{av-b}

194: \langle \tvec{b}^2 \rangle_x

195: &=& \frac{\int d^2\tvec{b}\; \tvec{b}^2\,

196:   q_v(x,\tvec{b})}{\int d^2 \tvec{b}\; q_v(x,\tvec{b})} 

197: \nonumber \\

198: &=& 4\, \frac{\partial}{\partial t} \log H_v^q(x,t) \Big|_{t=0}

199: \end{eqnarray}

200: with $H_v^q(x,t) = H^q(x,0,t) + H^q(-x,0,t)$ being the momentum space

201: counterpart of $q_v(x,\tvec{b})$.  The evolution equations for singlet

202: distributions mix quarks and gluons as usual.  We will argue below

203: that $\langle \tvec{b}^2 \rangle_x$ is a decreasing function of $x$.

204: Since $P(z) > 0$, the average impact parameter at given $x$ then

205: decreases with $\mu$ according to (\ref{dglap-impact}).  This is

206: readily understood: at fixed $\tvec{b}$ evolution to higher scale

207: $\mu$ decreases the longitudinal momentum of quarks because they

208: radiate gluons.  As sketched in Fig.~\ref{fig:evol}, this implies a

209: smaller typical $\tvec{b}$ at given $x$ as $\mu$ increases.

210: 

211: \begin{figure}

212: \begin{center}

213: \includegraphics[width=0.25\textwidth]{evolution.eps}

214: \vspace{-3em}

215: \end{center}

216: %

217: \caption{\label{fig:evol} Typical pattern of scale evolution in $x$

218: and $\tvec{b}^2$.}

219: \end{figure}

220: 

221: \begin{figure}[tb]

222: \begin{center}

223: \includegraphics[width=0.38\textwidth]{geometry.eps}

224: \vspace{-3em}

225: \end{center}

226: %

227: \caption{\label{fig:geometry} Three-quark configuration with one fast

228:   quark in a proton.  The thick line denotes the center of momentum of

229:   the proton and the dashed line the center of momentum of the two

230:   spectator quarks.}

231: \end{figure}

232: 

233: At large $x$, the struck quark takes most of the proton momentum, so

234: that its impact parameter tends to coincide with the center of

235: momentum of the entire proton.  In the limit $x\to 1$ one thus expects

236: a narrow distribution in $\tvec{b}$, or equivalently a flat $t$

237: dependence of GPDs in momentum space.  An estimate for the overall

238: transverse size of the proton in that limit is provided by the

239: transverse distance $\tvec{b} /(1-x)$ between the struck quark and the

240: center of momentum of the \emph{spectator} partons, as shown in

241: Fig.~\ref{fig:geometry}.  It is plausible to assume that this distance

242: remains finite due to confinement \cite{Burkardt:2004bv}, so that the

243: average squared impact parameter of partons vanishes like $\langle

244: \tvec{b}^2 \rangle_x \sim (1-x)^2$ for $x\to 1$.

245: 

246: 

247: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

248: 

249: \section{THE DIRAC FORM FACTORS}

250: \label{sec:dirac}

251: 

252: Information on the interplay between $x$ and $t$ in the valence

253: distributions $H_v^q(x,t)$ can be obtained from the Dirac form factors

254: of proton and neutron via the sum rules

255: \begin{eqnarray}

256:   \label{p-sum}

257: F_1^{p}(t) &=& \textstyle \int_0^1 dx\, \Big[

258:    \frac{2}{3} H_v^{u}(x,t) 

259:  - \frac{1}{3} H_v^{d}(x,t) \Big] ,

260: \\

261:   \label{n-sum}

262: F_1^{n}(t) &=& \textstyle \int_0^1 dx\, \Big[

263:    \frac{2}{3} H_v^{d}(x,t) 

264:  - \frac{1}{3} H_v^{u}(x,t) \Big] ,

265: \end{eqnarray}

266: where we have neglected the contribution from strange quarks.  Flavor

267: labels in $H_v^q$ refer to quarks in the proton.  Notice that the

268: dependence of GPDs on the resolution scale $\mu$ cancels in these

269: integrals, because the electromagnetic form factors belong to a

270: conserved current.  In this way, form factors measured at low $t$ can

271: constrain the distributions of partons resolved at much higher

272: resolution scales $\mu^2$.

273: 

274: The information from elastic form factors is very complementary to

275: what can be learned from hard exclusive scattering processes.

276: Experimental coverage in $t$ and the precision of measurements and

277: their quantitative interpretation is typically greater for form

278: factors than for more complex exclusive processes.  Electromagnetic

279: form factors are sensitive to the difference of quark and antiquark

280: distributions and thus insensitive to sea quarks and gluons, which can

281: be accessed in processes like deeply virtual Compton scattering or

282: vector meson production.  Finally, parton momentum fractions only

283: appear under an integral in elastic form factors, whereas a combined

284: measurement of the $\xi$ and $t$ dependence in exclusive processes

285: gives a more direct correlation of longitudinal and transverse

286: variables as discussed in Sect.~\ref{sec:impact}.

287: 

288: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

289: 

290: \subsection{An ansatz for the distribution $H_v^q$}

291: \label{sec:fit}

292: 

293: In the following we present results of the analysis of electromagnetic

294: form factors performed in Ref.~\cite{Diehl:2004cx}, to which we refer

295: for details.  A study along similar lines can be found in

296: \cite{Guidal:2004nd}.  Our ansatz for $H_v^q$ is of the form

297: \begin{equation}

298:   \label{H-ansatz}

299: H_v^q(x,t) = q_v(x) \exp[\, t f_q(x) \,] ,

300: \end{equation}

301: where for $q_v(x)$ we have taken the CTEQ6M parameterization

302: \cite{Pumplin:2002vw}.  The results of our analysis are stable within

303: the CTEQ error estimates on the parton densities.  All distributions

304: in the following refer to the scale $\mu=2 \gev$.  The interplay

305: between $x$ and $t$ dependence in $H_v^q$ is controlled by the profile

306: function $f_q(x)$, which according to (\ref{av-b}) is readily

307: identified as $\frac{1}{4} \langle \tvec{b}^2 \rangle_x$.

308: 

309: \begin{figure*}

310: \begin{center}

311: \includegraphics[width=.4\textwidth,

312:   bb=120 235 500 590]{fig-F1p.ps}

313: \hspace{2em}

314: \includegraphics[width=.4\textwidth,

315:   bb=110 337 490 690]{fig-F1n.ps}

316: \vspace{-3em}

317: \end{center}

318: %

319: \caption{\label{fig:F1fit} Results of fitting the ansatz given by

320:   (\protect\ref{H-ansatz}) and (\protect\ref{profile-ansatz}) to the

321:   proton and neutron Dirac form factors using the sum rules

322:   (\protect\ref{p-sum}) and (\protect\ref{n-sum}).  Shaded bands

323:   reflect the 1--$\sigma$ uncertainties on the fitted parameters.  The

324:   data for $F_1^p$ are described within 5\% except for the point at

325:   the highest $\abst$.}

326: \end{figure*}

327: 

328: 

329: For small $x$, Regge phenomenology of soft hadronic interactions

330: suggests an ansatz \cite{Goeke:2001tz}

331: \begin{equation}

332:   \label{regge-ansatz}

333: H_v^q(x,t) \;\sim\;  x^{-(\alpha+\alpha' t)}

334:   \;=\;  x^{-\alpha} \, e^{\,t \alpha' \log(1/x)}

335: \end{equation}

336: for the $x$ dependence, with $\alpha\approx 0.4 \mbox{~to~} 0.5$ and

337: $\alpha'\approx 0.9 \gev^{-2}$ corresponding to the leading meson

338: exchange trajectories.  This is known to work rather well for the

339: usual valence quark densities, i.e.\ in the forward limit $t=0$.  We

340: note that in the singlet sector the situation is more complicated: the

341: powers $\alpha$ parameterizing sea quark and gluon densities at scales

342: of a few GeV are significantly larger than the respective values for

343: meson and pomeron exchange in soft hadronic reactions.  Furthermore,

344: the value of $\alpha'$ measured in exclusive $J/\Psi$ production,

345: which involves the generalized gluon distribution, is smaller than the

346: corresponding value for pomeron exchange in soft scattering processes

347: \cite{Chekanov:2004mw}.  It remains an outstanding task to constrain

348: the shrinkage parameter $\alpha'$ in generalized sea quark

349: distributions and to understand its interplay with $\alpha'$ for

350: gluons through evolution in $\mu$.

351: 

352: For the limit $x\to 1$ we impose $f_q(x) \sim (1-x)^2$, corresponding

353: to a finite transverse size of the proton as discussed in

354: Sect.~\ref{sec:impact}.  We investigated several forms of $f_q$ that

355: interpolate between the limiting behavior for $x\to 0$ and $x\to 1$

356: just discussed, and found good results with

357: \begin{eqnarray}

358:   \label{profile-ansatz}

359: f_q(x) &=& \alpha' (1-x)^3 \log(1/x) 

360: \nonumber \\

361:  && {}+ B_q (1-x)^3 + A_q x (1-x)^2 .

362: \end{eqnarray}

363: The high power of $(1-x)$ multiplying the term with $\log(1/x)$

364: ensures that the parameter $\alpha'$ controls the behavior of the

365: distribution at small but not at moderate or large $x$ -- we expect

366: that the physics in these $x$ regions is very different and not

367: naturally described by the same parameters.

368: 

369: 

370: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

371: 

372: \subsection{Results and lessons of the fit}

373: 

374: With the ansatz just described we obtain a good fit to the data for

375: the Dirac form factors, as seen in Fig.~\ref{fig:F1fit}.  To reduce

376: the number of free parameters we set $\alpha' =0.9 \gev^{-2}$; leaving

377: it free we find $\alpha' =0.97\pm 0.04 \gev^{-2}$ well in the region

378: suggested by Regge phenomenology.  We also imposed $B_u=B_d\slim$;

379: relaxing this constraint improves the fit only slightly.  The main

380: result of our fit is a strong $x$ dependence of the average impact

381: parameter of valence quarks over the entire $x$ range.  This is

382: illustrated for $u$ quarks in Fig.~\ref{fig:profile}, where we plot

383: the average distance $d_u$ between struck quark and spectators.

384: 

385: We have also fitted the form factors to the alternative ansatz

386: \begin{eqnarray}

387:   \label{alt-ansatz}

388: f_q(x) &=& \alpha' (1-x)^2 \log(1/x) 

389: \nonumber \\

390:  && {}+ B_q (1-x)^2 + A_q x (1-x)

391: \end{eqnarray}

392: for the profile function.  This leads to a similarly good description

393: of the data as with the form (\ref{profile-ansatz}).  The reason can

394: be seen in Fig.~\ref{fig:profile}.  Up to $x \sim 0.8$ the fitted

395: results for $d_u(x)$ are barely distinguishable; only for larger $x$

396: does the fit with (\ref{alt-ansatz}) produce a rise of $d_u(x)$ to

397: values that appear unphysically large.  In the $t$ range where there

398: is data, the form factor $F_1^p$ is however barely sensitive to

399: $x>0.8$.  This is shown in Fig.~\ref{fig:minmax}, where we also plot

400: the average value

401: \begin{equation}

402:   \label{xavg}

403: \langle x \rangle_t = \frac{\sum_q e_q \int_0^1 dx\, x

404:   H_v^q(x,t)}{\sum_q e_q \int_0^1 dx\, H_v^q(x,t)}

405: \end{equation}

406: of $x$ in the integral (\ref{p-sum}).  The more scarce data on $F_1^n$

407: do not constrain the region $x>0.8$ either.  We see from this exercise

408: that with observables sensitive to $x \lsim 0.8$ one cannot

409: unambiguously determine a power-law behavior in $(1-x)$ for the limit

410: $x\to 1$.

411: 

412: The average impact parameter for $d$ quarks is less well constrained

413: by our fit since $F_1^p$, for which data is abundant, is dominated by

414: $u$ quarks.  We found however that a good description of the $F_1^n$

415: data requires a larger impact parameter of $d$ quarks compared with

416: $u$ quarks at moderate to large $x$.  It will be interesting to see

417: whether future data on $F_1^n$ confirm this trend.  This would be an

418: analog to the very different distribution in $x$ of $u$ and $d$

419: quarks, which may for instance hint at a quark-diquark structure of

420: nucleon configurations at large $x$ \cite{Feynman:1972}.

421: 

422: \begin{figure}

423: \begin{center}

424: \includegraphics[width=0.45\textwidth]{avb.eps}

425: \vspace{-3em}

426: \end{center}

427: %

428: \caption{\label{fig:profile} Average distance between struck quark and

429:   spectators in the valence distribution for $u$ quarks, given by $d_u

430:   = (1-x)^{-1}\, [\langle b^2 \rangle_x ]^{1/2}$ according to

431:   Fig.~\protect\ref{fig:geometry}.  The lower curve is for our fit

432:   (\protect\ref{profile-ansatz}) and the upper one for the alternative

433:   fit (\protect\ref{alt-ansatz}).}

434: \end{figure}

435: 

436: \begin{figure}[t]

437: \begin{center}

438: \includegraphics[width=0.4\textwidth,%

439:   bb=60 320 410 565]{fig-minmax-p20.ps}

440: \vspace{-3.4em}

441: \end{center}

442: %

443: \caption{\label{fig:minmax} Region of $x$ (white region) which

444: accounts for $90\%$ of $F_1^p(t)$ in the integral

445: (\protect\ref{p-sum}) for our fit of $H_v^q(x,t)$.  The upper and

446: lower shaded $x$-regions each account for $5\%$ of $F_1^p(t)$.  The

447: thick line shows the average $\langle x\rangle_t$ as defined in

448: (\protect\ref{xavg}).}

449: %\end{figure}

450: %

451: \vspace{0.7em}

452: %

453: %\begin{figure}

454: \begin{center}

455: \includegraphics[width=0.45\textwidth]{Lambdaeff.eps}

456: \vspace{-3em}

457: \end{center}

458: %

459: \caption{\label{fig:leff} The scale parameter $\Lambda_{\rm eff}$ for

460:   the Feynman mechanism, as explained in the text.}

461: \end{figure}

462: 

463: 

464: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

465: 

466: \subsection{Large $t\,$: Feynman and Drell-Yan}

467: 

468: If one assumes that elastic form factors are dominated by

469: configurations where partons not struck by the photon have

470: virtualities of the order of a strong interaction scale $\Lambda^2$,

471: then at large $t$ the momentum fraction of the struck quark must

472: become large -- this is the mechanism originally proposed by Feynman

473: \cite{Feynman:1972}.  The integrals (\ref{p-sum}) and (\ref{n-sum})

474: are then dominated by the region where $1-x \sim \Lambda /\sqrt{-t}$.

475: The large-$t$ asymptotics of our ansatz (\ref{H-ansatz}) and

476: (\ref{profile-ansatz}) indeed follows this behavior, as is readily

477: seen from the saddle point approximation of the relevant integrals.

478: To quantify this we have evaluated $\Lambda_{\rm eff}(t) = \sqrt{-t}\;

479: \langle 1 - x \rangle_t$ with $\langle 1 - x \rangle_t = 1 - \langle x

480: \rangle_t$ from (\ref{xavg}).  The result is displayed in

481: Fig.~\ref{fig:leff} and shows that the expected asymptotic behavior

482: slowly sets in for $\abst$ around $10 \gev^2$.

483: 

484: To explore the dependence of our conclusions on the assumed form of

485: $H_v^q(x,t)$, we have performed fits with the exponential $t$

486: dependence in (\ref{H-ansatz}) replaced by a power law

487: \begin{equation}

488: H_v^q(x,t) = q_v(x) \,\Big( 1 - \frac{t f_q(x)}{p} \,\Big)^{-p}

489: \end{equation}

490: with $f_q(x)$ of the form (\ref{profile-ansatz}).  We obtain a good

491: description of the form factors in a wide range of $p$, from $p

492: \approx 2.5$ up to the limit $p\to \infty$, where we recover the

493: exponential (\ref{H-ansatz}).  Asymptotically, our modified ansatz

494: still satisfies $\langle 1 - x \rangle_t \sim \Lambda /\sqrt{-t}$.

495: For small $p$ this behavior is however not reached at values of $t$

496: where there is data, and the form factor sum rules are not dominated

497: by large $x$.  (This underlines the need to check asymptotic

498: considerations against numerical estimates when describing baryon form

499: factors.)  The available data for $F_1^p$ does hence \emph{not prove}

500: that the Feynman mechanism is at work in the high-$t$ region, but it

501: is \emph{consistent} with this assumption, given the success of our

502: fits with large $p$.

503: 

504: \begin{figure*}[t]

505: \begin{center}

506: \includegraphics[width=0.46\textwidth]{hu.eps} \hfill

507: \includegraphics[width=0.46\textwidth]{hd.eps} \\

508: \vspace{1em}

509: \includegraphics[width=0.46\textwidth]{eu.eps} \hfill

510: \includegraphics[width=0.46\textwidth]{ed.eps} \\

511: \vspace{-3em}

512: \end{center}

513: %

514: \caption{\label{fig:moments} Scaled $x$ moments

515:   (\protect\ref{h-moments}) and (\protect\ref{e-moments}) of

516:   generalized $u$ and $d$ valence distributions at $\mu=2\gev$,

517:   obtained from our fits described in Sects.~\protect\ref{sec:fit} and

518:   \protect\ref{sec:pauli}.}

519: \end{figure*}

520: 

521: Dominance of the Feynman mechanism implies the Drell-Yan relation

522: \begin{equation}

523:   \label{dy}

524: F_1^q(t) \sim \abst^{- (1+\beta_q)/2}

525: ~~\mbox{for}~~ q_v(x) \sim (1-x)^{\,\beta_q}

526: \end{equation}

527: between the form factors $F_1^q(t) = \int dx\, H_v^q(x,t)$ at large

528: $t$ and parton distributions at large $x$.  We can understand

529: $\beta_q$ as an \emph{effective} power describing the behavior of

530: $q_v(x)$ at large $x$ (rather than in the experimentally unexplored

531: limit $x\to 1$).  With the CTEQ6M distributions at $\mu=2\gev$ we find

532: $u_v(x) \sim (1-x)^{3.4}$ and $d_v(x) \sim (1-x)^{5.0}$ for $0.5\le x

533: \le 0.9$.  The Drell-Yan relation then implies a drastically different

534: $t$ dependence of the contributions from $u$ and $d$ quarks to the

535: nucleon form factors.  $F_1^u = 2F_1^p + F_1^n$ should approximately

536: scale like $\abst^{-2}$ at large $t$, whereas $F_1^d = F_1^p + 2F_1^n$

537: should fall off like $\abst^{-3}$.  This difference in the large-$t$

538: behavior is already felt at lower values of $t$, as can be seen in the

539: two upper panels of Fig.~\ref{fig:moments}, where we plot the $x$

540: moments

541: \begin{equation}

542:   \label{h-moments}

543: h^q_{i}(t) = \textstyle \int_{0}^1 dx \, x^{i-1} \, H_v^q(x,t)

544: \end{equation}

545: with $i=1,2,3$ for the GPDs obtained in our fit.  The lowest moments

546: $h^u_1$ and $h^d_1$ can be extracted from experimental data on the

547: electromagnetic proton and neutron form factors, and the higher

548: moments are accessible to calculation in lattice QCD

549: \cite{Gockeler:2003jf,Hagler:2003jd,Schierholz:2005}.  Both types of

550: studies are challenging for $\abst$ above $3 \gev^2$, where our fit

551: predicts the most striking differences between $u$ and $d$ quarks, but

552: will hopefully be feasible in the future.

553: 

554: 

555: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

556: 

557: \section{THE PAULI FORM FACTORS}

558: \label{sec:pauli}

559: 

560: The proton helicity flip distributions $E^q$ admit a density

561: interpretation at $\xi=0$, similar to the distributions $H^q$

562: discussed so far.  To see this one changes basis from proton helicity

563: states $|\!\!\uparrow \rangle$, $|\!\!\downarrow \rangle$ to states

564: $|X\pm \rangle = (\, |\!\!\uparrow \rangle \pm |\!\!\downarrow \rangle

565: \,) /\sqrt{2}$ polarized along the positive or negative $x$ axis.  In

566: impact parameter space one then obtains the density

567: \begin{equation}

568:   \label{trans-density}

569: q_X(x,\mathbf{b}) = q(x,\tvec{b}) 

570:   - \frac{b^y}{m} \frac{\partial}{\partial \tvec{b}{}^2}\,

571:     e^q(x,\tvec{b})

572: \end{equation}

573: of unpolarized quarks in a proton polarized in the positive $x$

574: direction, where $q(x,\tvec{b})$ and

575: \begin{equation}

576: e^q(x,\tvec{b}) = \int \frac{d^2\tvec{\Delta}}{(2\pi)^2}\,

577:   e^{-i \svec{b} \svec{\Delta}}\,

578:   E^q(x,0,-\tvec{\Delta}^2)

579: \end{equation}

580: depend on $\tvec{b}$ only through $\tvec{b}^2$ due to rotation

581: invariance.  The impact parameter distribution of quarks in a

582: transversely polarized proton is thus shifted in the direction

583: perpendicular to the polarization.  The interpretation of

584: (\ref{trans-density}) as a density implies positivity bounds for

585: $e^q(x,\tvec{b})$, which in momentum space can be written as

586: \cite{Burkardt:2003ck}

587: \begin{eqnarray}

588:   \label{e-bound}

589: \lefteqn{ \Big[ E^q(x,0,t=0) \Big]^2 }

590: \\

591: &\leq &  m^2 

592: \Big[ q(x) + \Delta q(x) \Big]\, \Big[ q(x) - \Delta q(x) \Big]\,

593: \nonumber \\

594:  && {}\times 4\, \frac{\partial}{\partial t} 

595:     \log \Big[H^q(x,0,t)\pm \tilde H^q(x,0,t)\Big]_{t=0} \;.

596: \nonumber

597: \end{eqnarray}

598: Depending on the sign, the term in the third line is the average

599: squared impact parameter of quarks with positive or negative helicity,

600: which according to our previous discussion tends to zero for large

601: $x$.  In addition, the densities $d + \Delta d$ and $u - \Delta u$ of

602: right-handed $d$ and left-handed $u$ quarks are phenomenologically

603: known to decrease strongly with $x$, so that the right-hand side of

604: (\ref{e-bound}) restricts $|E^q(x,0,0)|$ quite severely for larger

605: values of $x$.  The distribution $E^q$ involves one unit of orbital

606: angular momentum since in the associated matrix elements the proton

607: helicity is flipped but the quark helicity conserved, see

608: Fig.~\ref{fig:gpdE}.  The bound (\ref{e-bound}) thus limits the amount

609: of orbital angular momentum that can be carried by quarks with large

610: $x$.

611: 

612: \begin{figure}

613: \begin{center}

614: \includegraphics[width=0.25\textwidth]{gpdE.eps}

615: \vspace{-3em}

616: \end{center}

617: %

618: \caption{\label{fig:gpdE} $E^q$ describes transitions

619:    where the proton helicity is flipped but the quark helicity

620:    conserved.  The helicity mismatch is compensated by one unit of

621:    orbital angular momentum, according to angular momentum

622:    conservation.}

623: \end{figure}

624: 

625: \begin{figure*}[t]

626: \begin{center}

627: \includegraphics[width=0.4\textwidth,%

628:   bb=50 50 398 291]{Lu.eps}

629: \hspace{2.5em}

630: \includegraphics[width=0.4\textwidth,%

631:   bb=50 50 398 291]{Ld.eps}

632: \vspace{-3em}

633: \end{center}

634: %

635: \caption{\label{fig:L} The orbital angular momentum carried by valence

636:   quarks at scale $\mu=2 \gev$, obtained in the range of parameters

637:   for which we obtain a good fit to the proton and neutron Pauli form

638:   factors. The corresponding values for the total angular momentum

639:   $J_v^q = L_v^q + \half \int dx\, \Delta q_v(x) $ are $J_u = L_u +

640:   \half\, 0.93$ and $J_v^d = L_v^d - \half\, 0.34$, so that $J_v^d$

641:   comes out close to zero in our estimate.}

642: \end{figure*}

643: 

644: Sum rules analogous to (\ref{p-sum}) and (\ref{n-sum}) relate the

645: Pauli form factors of proton and neutron with the valence combinations

646: $E_v^q(x,t) = E^q(x,0,t) + E^q(-x,0,t)$ of proton helicity flip

647: distributions.  These distributions cannot be measured at $t=0$ so

648: that in contrast to $H_v^q(x,t)$ their forward limit is unknown.  We

649: have made an ansatz

650: \begin{equation}

651:   \label{E-ansatz}

652: E_v^q(x,t) = e_v^q(x) \exp[\, t g_q(x) \,]

653: \end{equation}

654: with $g_q(x)$ of the same functional form as $f_q(x)$ in

655: (\ref{profile-ansatz}).  For the forward limit of $E_v^q$ we assumed a

656: form

657: \begin{equation}

658:   \label{eq-ansatz}

659: e_v^q(x) = \mathcal{N}_q\, x^{-\alpha} (1-x)^{\beta_q} \,,

660: \end{equation}

661: which is known to work quite well for the ordinary valence quark

662: distributions $q_v(x)$.  The normalization factors $\mathcal{N}_q$ are

663: fixed by the requirement $\int dx\, e_v^q(x) = \kappa_q$ with

664: $\kappa_u\approx 1.67$ and $\kappa_d\approx -2.03$ obtained from the

665: magnetic moments of proton and neutron.  The overall normalization of

666: $E_v^q$ is hence quite large, which according to our discussion

667: implies significant spin-orbit effects of $u$ and $d$ quarks in the

668: proton.

669: 

670: Using the ansatz just described we obtain a good fit to the data for

671: the Pauli form factors of proton and neutron, with $\alpha=0.55$ and

672: $\alpha'= 0.9\gev^{-2}$ in agreement with expectations from Regge

673: phenomenology.  The quality of the fit is similar to the one we

674: achieved for the Dirac form factors.  We find a very large range of

675: allowed fit parameters, which is hardly surprising since we have to

676: determine both functions $e_v^q(x)$ and $g_q(x)$ in (\ref{E-ansatz}).

677: An important reduction of the allowed parameter space is due to the

678: positivity bound (\ref{e-bound}) and its analog in impact parameter

679: space at large $x$ (where it is reasonable to neglect the contribution

680: from antiquarks, which is subtracted in the valence distributions and

681: invalidates positivity conditions if it is large).  We find in

682: particular that $\beta_u \ge 3.5$ and $\beta_d \ge 5$ is required with

683: our ansatz, which quantifies our above statement that $E_v^q$ must

684: rather strongly decrease with~$x$.

685: 

686: We can now evaluate the orbital angular momentum carried by valence

687: quarks in the proton, which according to Ji's sum rule

688: \cite{Ji:1996ek} is

689: \begin{equation}

690: L_v^q =

691: \half \textstyle\int_0^1 dx\, 

692:   \Big[ x e_v^q(x) + x q_v(x) - \Delta q_v(x) \Big] .

693: \end{equation}

694: With our simple ansatz (\ref{eq-ansatz}) one readily obtains

695: \begin{equation}

696: {\textstyle\int_0^1} dx\, x e_v^q(x) = \kappa_q\,

697:   (1-\alpha) /(2-\alpha+\beta_q) .

698: \end{equation}

699: Remarkably, the results for $L_v^u$ and $L_v^d$ show only little

700: variation in the range of parameters $\alpha$, $\beta_u$ and $\beta_d$

701: for which we achieve a good fit to the Pauli form factors, as shown in

702: Fig.~\ref{fig:L}.  For the isovector combination we obtain a rather

703: well determined value $L_v^u - L_v^d = - \half\, ( 0.77 ~\mbox{to}~

704: 0.92)$.  The isoscalar combination $L_v^u + L_v^d = - \half\, (0.11

705: ~\mbox{to}~ 0.22)$ has a large uncertainty and is rather small, due to

706: partial cancellation between the two quark flavors.  Lattice

707: calculations by the QCDSF Collaboration obtain $L^u - L^d = - \half\,

708: (0.90\pm 0.12)$ and $L^u + L^d$ compatible with zero within errors

709: \cite{Schierholz:2005}.  We find the agreement with our estimates

710: encouraging, especially for the isovector combination, where

711: contributions from sea quarks should be small.

712: 

713: As in the case of $H_v^q$, we find that the different behavior of

714: $E_v^u$ and $E_v^d$ at large $x$ is reflected in a different $t$

715: dependence of their moments, as is characteristic for the Feynman

716: mechanism.  This is seen in the lower panels of

717: Fig.~\ref{fig:moments}, where we plot

718: \begin{equation}

719:   \label{e-moments}

720: e^q_{i}(t) = \textstyle \int_{0}^1 dx \, x^{i-1} \, E_v^q(x,t)

721: \end{equation}

722: for $i=1,2,3$ obtained with our best fit, where the large-$x$ powers

723: in (\ref{eq-ansatz}) are $\beta_u\approx 4$ and $\beta_d\approx 5.6$.

724: It will be very interesting to see whether such a behavior can be

725: established in form factor measurements or lattice calculations.

726: 

727: 

728: \begin{figure*}

729: \begin{center}

730: \includegraphics[width=0.49\textwidth]{F1F2.eps}

731: \hfill

732: \includegraphics[width=0.49\textwidth]{F2.eps}

733: \vspace{-3em}

734: \end{center}

735: %

736: \caption{\label{fig:F1F2} Data for the proton form factors, scaled by

737:   $t^2$ (left) or $\abst^3$ (right).  The plateau of $t^2 F_2^p(t)$

738:   for $\abst$ between $2$ and $6\gev^2$ illustrates that, in a limited

739:   kinematic region, observables may exhibit an approximate power law

740:   which is only transient and very different from their asymptotic

741:   behavior.}

742: \end{figure*}

743: 

744: Let us finally comment on the behavior of the ratio $F_2^p /F_1^p$ for

745: $\abst$ up to about $6\gev^2$, where there is data from the

746: polarization transfer method \cite{Gayou:2001qd}.  This data is rather

747: well described by a behavior $F_2^p /F_1^p \sim \abst^{-1}

748: \log^2(\abst/\Lambda^2)$ with $\Lambda\approx 300 \mev$, suggested by

749: the recent study \cite{Belitsky:2002kj} of $F_2^p$ in the

750: leading-twist hard-scattering mechanism \cite{Lepage:1980fj}.  For the

751: individual form factors, this study obtains an approximate behavior

752: \begin{eqnarray}

753:   \label{hsp}

754: F_1(t) &\sim & \alpha_s^{2 + 32/(9\beta)}\, \abst^{-2} ,

755: \nonumber \\

756: F_2(t) &\sim & \alpha_s^{2 + 8/(3\beta)}\,

757:                \log^2(\abst/\Lambda^2)\; \abst^{-3} ,

758: \end{eqnarray}

759: where the squared logarithm in $F_2$ arises from cutting off endpoint

760: singularities in the integrations over quark momentum fractions.  The

761: terms with $\beta=11-2n_f /3$ in the exponents are due to the

762: evolution of the proton distribution amplitude and numerically small,

763: with $32/(9\beta) \approx 0.4$ and $8/(3\beta) \approx 0.3$.  If one

764: takes $\alpha_s$ at scale $t$, then the logarithms in $F_2$

765: approximately cancel and $\abst^3 F_2(t)$ should be nearly $t$

766: independent.  If in contrast one assumes that the relevant scale in

767: $\alpha_s$ is so low that the running coupling is effectively frozen,

768: then $\abst^2 F_1(t)$ should be flat.  Figure~\ref{fig:F1F2} shows

769: that neither behavior is realized for $\abst$ below $6 \gev^2$, where

770: both $\abst^3 F_2^p(t)$ and $\abst^2 F_1^p(t)$ increase.  Thus, a

771: calculation of $F_1^p$ and $F_2^p$ at leading twist and leading order

772: in $\alpha_s$ not only underestimates the normalization of the form

773: factors (as remarked in \cite{Belitsky:2002kj}) but also fails to

774: describe their $t$ dependence in the region under discussion.  Whether

775: the fact that the ratio $F_2^p/F_1^p$ can be described with

776: (\ref{hsp}) is fortuitous or a sign of precious scaling behavior

777: remains a matter of debate.  An important source of power corrections

778: is the effect of transverse quark momentum in the hard-scattering

779: subprocess \cite{Bolz:1994hb}.  This effect decreases the form factors

780: more strongly at lower than at higher $\abst$, which may improve the

781: description of the $t$ dependence but makes the discrepancy for their

782: normalization worse.  The contribution of the Feynman mechanism to

783: $F_1^p$ and $F_2^p$ originates from different kinematic configurations

784: than the hard-scattering mechanism.  As a correction to the

785: leading-twist result it is hence additive rather than multiplicative

786: and will therefore not obviously cancel in the form factor ratio.

787: 

788: 

789: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

790: 

791: \section*{Acknowledgments}

792: 

793: It is a pleasure to thank D. Leinweber, L. von Smekal and T. Williams

794: for organizing an excellent workshop.  I am grateful to D. Br\"ommel

795: for a careful reading of the manuscript.  This work is supported by

796: the Helmholtz Association, contract number VH-NG-004.

797: 

798: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

799: 

800: \begin{thebibliography}{99}

801: 

802: \bibitem{Diehl:2003ny} M.~Diehl,

803: %``Generalized parton distributions,''

804: Phys.\ Rept.\  {\bf 388}, 41 (2003)

805: [hep-ph/0307382];

806: %%CITATION = HEP-PH 0307382;%%

807: %

808: %\bibitem{Belitsky:2005qn}

809: A.~V.~Belitsky and A.~V.~Radyushkin,

810: %``Unraveling hadron structure with generalized parton

811: %distributions,'' 

812: hep-ph/0504030.

813: %%CITATION = HEP-PH 0504030;%%

814: 

815: \bibitem{Soper:1972xc}

816: D.~E.~Soper,

817: %``Infinite-Momentum Helicity States,''

818: Phys.\ Rev.\ D {\bf 5}, 1956 (1972);\\

819: %%CITATION = PHRVA,D5,1956;%%

820: %

821: %\bibitem{Soper:1976jc}

822: %D.~E.~Soper,

823: %``The Parton Model And The Bethe-Salpeter Wave Function,''

824: Phys.\ Rev.\ D {\bf 15}, 1141 (1977).

825: %%CITATION = PHRVA,D15,1141;%%

826: 

827: \bibitem{Burkardt:2000za}

828: M.~Burkardt,

829: %``Impact parameter dependent parton distributions and off-forward

830: %parton distributions for zeta $\to$ 0,''

831: Phys.\ Rev.\ D {\bf 62}, 071503 (2000),

832: Erratum ibid.\ D {\bf 66}, 119903 (2002)

833: [hep-ph/0005108].

834: %%CITATION = HEP-PH 0005108;%%

835: 

836: \bibitem{Diehl:2002he}

837: M.~Diehl,

838: %``Generalized parton distributions in impact parameter space,''

839: Eur.\ Phys.\ J.\ C {\bf 25}, 223 (2002),

840: Erratum ibid.\ C {\bf 31}, 277 (2003)

841: [hep-ph/0205208].

842: %%CITATION = HEP-PH 0205208;%%

843: 

844: \bibitem{Belitsky:2003nz}

845: A.~V.~Belitsky, X.~D.~Ji and F.~Yuan,

846: %``Quark imaging in the proton via quantum phase-space

847: %distributions,'' 

848: Phys.\ Rev.\ D {\bf 69}, 074014 (2004)

849: [hep-ph/0307383].

850: %%CITATION = HEP-PH 0307383;%%

851: 

852: \bibitem{Diehl:2004cx}

853: M.~Diehl, T.~Feldmann, R.~Jakob and P.~Kroll,

854: %``Generalized parton distributions from nucleon form factor data,''

855: Eur.\ Phys.\ J.\ C {\bf 39}, 1 (2005)

856: [hep-ph/0408173].

857: %%CITATION = HEP-PH 0408173;%%

858: 

859: \bibitem{Burkardt:2004bv}

860: M.~Burkardt,

861: %``Generalized parton distributions for large x,''

862: Phys.\ Lett.\ B {\bf 595}, 245 (2004)

863: [hep-ph/0401159].

864: %%CITATION = HEP-PH 0401159;%%

865: 

866: \bibitem{Guidal:2004nd}

867: M.~Guidal, M.~V.~Polyakov, A.~V.~Radyushkin and M.~Vanderhaeghen,

868: %``Nucleon form factors from generalized parton distributions,''

869: Phys.\ Rev.\ D {\bf 72}, 054013 (2005).

870: %[hep-ph/0410251].

871: %%CITATION = HEP-PH 0410251;%%

872: 

873: \bibitem{Pumplin:2002vw}

874: J.~Pumplin {\it et al.},

875: %, D.~R.~Stump, J.~Huston, H.~L.~Lai, P.~Nadolsky and W.~K.~Tung,

876: %``New generation of parton distributions with uncertainties from

877: %global QCD analysis,''

878: JHEP {\bf 0207}, 012 (2002)

879: [hep-ph/0201195].

880: %%CITATION = HEP-PH 0201195;%%

881: 

882: \bibitem{Goeke:2001tz}

883: K.~Goeke, M.~V.~Polyakov and M.~Vanderhaeghen,

884: %``Hard exclusive reactions and the structure of hadrons,''

885: Prog.\ Part.\ Nucl.\ Phys.\  {\bf 47}, 401 (2001)

886: [hep-ph/0106012].

887: %%CITATION = HEP-PH 0106012;%%

888: 

889: \bibitem{Chekanov:2004mw}

890: S.~Chekanov {\it et al.}\ (ZEUS Collab.),

891: %``Exclusive electroproduction of J/psi mesons at HERA,''

892: {\it Nucl.\ Phys.}\ {\bf B 695}, 3 (2004)

893: [hep-ex/0404008];\\

894: %%CITATION = HEP-EX 0404008;%%

895: %

896: A.~Aktas {\it et al.}\ (H1 Collab.),

897: %``Elastic J/psi production at HERA,''

898: hep-ex/0510016.

899: %%CITATION = HEP-EX 0510016;%%

900: 

901: \bibitem{Feynman:1972}

902: R. P. Feynman,

903: \emph{Photon-Hadron Interactions},

904: Benjamin, New York, 1972.

905: 

906: \bibitem{Gockeler:2003jf}

907: M.~G\"ockeler {\it et al.}\

908: % R.~Horsley, D.~Pleiter, P.~E.~L.~Rakow, A.~Schafer, G.~Schierholz

909: % and W.~Schroers 

910: (QCDSF Collab.),

911: %``Generalized parton distributions from lattice QCD,''

912: Phys.\ Rev.\ Lett.\  {\bf 92}, 042002 (2004)

913: [hep-ph/0304249].

914: %%CITATION = HEP-PH 0304249;%%

915: 

916: \bibitem{Hagler:2003jd}

917: P.~H\"agler {\it et al.}\

918: % J.~W.~Negele, D.~B.~Renner, W.~Schroers, T.~Lippert and K.~Schilling

919: (LHPC Collab.),

920: %``Moments of nucleon generalized parton distributions in lattice

921: %QCD,'' 

922: Phys.\ Rev.\ D {\bf 68}, 034505 (2003)

923: [hep-lat/0304018];

924: %%CITATION = HEP-LAT 0304018;%%

925: %

926: %\bibitem{Hagler:2003is}

927: %``Transverse structure of nucleon parton distributions from lattice

928: %QCD,'' 

929: Phys.\ Rev.\ Lett.\  {\bf 93}, 112001 (2004)

930: [hep-lat/0312014].

931: %%CITATION = HEP-LAT 0312014;%%

932: 

933: \bibitem{Schierholz:2005}

934: G. Schierholz, these proceedings.

935: 

936: \bibitem{Burkardt:2003ck}

937: M.~Burkardt,

938: %``Some inequalities for the generalized parton distribution

939: %E(x,0,t),'' 

940: Phys.\ Lett.\ B {\bf 582}, 151 (2004)

941: [hep-ph/0309116].

942: %%CITATION = HEP-PH 0309116;%%

943: 

944: \bibitem{Ji:1996ek}

945: X.~D.~Ji,

946: %``Gauge invariant decomposition of nucleon spin,''

947: Phys.\ Rev.\ Lett.\  {\bf 78}, 610 (1997)

948: [hep-ph/9603249].

949: %%CITATION = HEP-PH 9603249;%%

950: 

951: \bibitem{Gayou:2001qd}

952: O.~Gayou {\it et al.}\ (Jlab Hall A Collab.),

953: %``Measurement of G(E(p))/G(M(p)) in e(pol.) p $\to$ e p(pol.) to Q**2

954: %= 5.6-GeV**2,''

955: Phys.\ Rev.\ Lett.\  {\bf 88}, 092301 (2002)

956: [nucl-ex/0111010].

957: %%CITATION = NUCL-EX 0111010;%%

958: 

959: \bibitem{Belitsky:2002kj}

960: A.~V.~Belitsky, X.~D.~Ji and F.~Yuan,

961: %``A perturbative QCD analysis of the nucleon's Pauli form factor

962: %F2(Q**2),'' 

963: Phys.\ Rev.\ Lett.\  {\bf 91}, 092003 (2003)

964: [hep-ph/0212351].

965: %%CITATION = HEP-PH 0212351;%%

966: 

967: \bibitem{Lepage:1980fj}

968: G.~P.~Lepage and S.~J.~Brodsky,

969: %``Exclusive Processes In Perturbative Quantum Chromodynamics,''

970: Phys.\ Rev.\ D {\bf 22} (1980) 2157.

971: %%CITATION = PHRVA,D22,2157;%%

972: 

973: \bibitem{Bolz:1994hb}

974: J.~Bolz, R.~Jakob, P.~Kroll, M.~Bergmann and N.~G.~Stefanis,

975: %``A Critical analysis of the proton form-factor with Sudakov

976: %suppression and intrinsic transverse momentum,''

977: Z.\ Phys.\ C {\bf 66}, 267 (1995)

978: [hep-ph/9405340].

979: %%CITATION = HEP-PH 9405340;%%

980: 

981: \end{thebibliography}

982: 

983: \end{document}

984: