1: %\documentclass[draft]{ws-procs9x6}
2: \documentclass{ws-procs9x6}
3: %
4: \usepackage{graphicx}
5: %
6: \begin{document}
7: %
8: %
9: \title{Higher twists in spin structure functions from
10: a ``constituent quark'' point of view}
11: %
12: %
13: \author{A.~V.~Sidorov}
14: \address{Bogoliubov Laboratory of Theoretical Physics,
15: Joint Institute of Nuclear Research,
16: 141980 Dubna, Russian Federation}
17: %
18: %
19: \author{C.~Weiss}
20: \address{Theory Group, Jefferson Lab, 12000 Jefferson Avenue,
21: Newport News, VA~23606, U.S.A.}
22: %
23: \maketitle
24: %
25: \abstracts{We discuss the implications of a ``constituent quark''
26: structure of the nucleon for the leading ($1/Q^2$--) power corrections
27: to the spin structure functions.
28: Our basic assumption is the presence of quark--gluon correlations
29: in the nucleon wave function, whose size, $\rho \sim 0.3\, {\rm fm}$,
30: is small compared to the nucleon radius, $R$ (two--scale picture).
31: We argue that in this picture the isovector twist--4 matrix element
32: in the proton has a sizable negative value,
33: $M_N^2 |f_2^{u - d}| \sim \rho^{-2}$,
34: while the twist--3 matrix elements are small, $M_N^2 d_2 \sim R^{-2}$.
35: These findings are in agreement with the result of a QCD fit to the $g_1$
36: world data, including recent neutron data from HERMES and
37: Jefferson Lab Hall A, which gives
38: $M_N^2 f_2^{u - d} = -0.28 \pm 0.08 \; {\rm GeV}^2$.}
39: %
40: The transition from the scaling regime at large $Q^2$ to the
41: quasi--real regime at small $Q^2$ in the structure
42: functions of inelastic $eN$ scattering represents
43: a major challenge for theory and experiment.
44: Coming from high $Q^2$, the onset of the transition manifests itself
45: in power ($1/Q^2$--) corrections to the $Q^2$ dependence of the structure
46: functions. In QCD, these corrections are related to the interactions of the
47: ``active'' quark/antiquark with the non-perturbative gluon field
48: in the nucleon, described by nucleon matrix elements of certain
49: quark--gluon operators of twist 3 and 4. What are the scale parameters
50: governing the size of these matrix elements?
51:
52: There is ample evidence for a constituent quark structure of the nucleon
53: --- the presence in the nucleon of small--size extended objects ---
54: from hadron spectroscopy and low--energy electromagnetic interactions.
55: The notion of a massive constituent quark of finite size is also
56: intimately related to the spontaneous breaking of chiral symmetry.
57: For instance, the microscopic picture of chiral symmetry breaking
58: based on QCD instantons gives rise to constituent quarks of a
59: ``size'' of $\rho \sim 0.3 \, {\rm fm}$, which is determined by
60: the average size of the instantons in the vacuum.\cite{Diakonov:2002fq}
61: It is natural to ask if this scale could explain the quark--gluon
62: correlations giving rise to power corrections to polarized
63: deep--inelastic scattering.\footnote{Further evidence for
64: constituent quarks of a size $\rho \sim 0.3 \, {\rm fm}$
65: comes from the correlations in the transverse spatial distribution
66: of partons in the nucleon, observed in the production of multiple
67: hard dijets in high--energy $pp$ collisions.\cite{fest}}
68:
69: In this note\cite{paper} we explore the implications of a constituent
70: quark structure of the nucleon for the leading ($1/Q^2$--)
71: power corrections to the nucleon spin structure functions,
72: from a general, model--independent perspective. We think of
73: constituent quarks and antiquarks in the field--theoretical sense,
74: as finite--size correlations in the quark--gluon wave function
75: of the nucleon in QCD, not as elementary objects in the sense
76: of a potential model.
77: We shall try to relate the size of these correlations to the
78: nucleon matrix elements of twist--3 and 4 quark--gluon operators
79: which govern the $1/Q^2$--corrections to the lowest moments
80: of the spin structure functions in QCD.
81:
82: %
83: \begin{figure}[t]
84: %
85: \begin{center}
86: \includegraphics[width=3cm,height=3cm]{twoscale.eps}
87: \end{center}
88: %
89: \caption[*]{}
90: \label{fig_twoscale}
91: \end{figure}
92: %
93: Our basic assumption is that the size of the ``constituent quarks'',
94: $\rho$, be much smaller than the radius of the nucleon, $R$
95: (see Fig.~\ref{fig_twoscale}),
96: \begin{equation}
97: \rho \;\; \ll \;\; R .
98: \label{hierarchy}
99: \end{equation}
100: Various phenomenological considerations point to a constituent quark
101: size of $\sim 0.3\; \mbox{fm}$, which should be compared, say, to the nucleon
102: isoscalar charge radius, $\langle r^2 \rangle^{1/2} = 0.8 \; {\rm fm}$.
103: The precise values of these parameters are not the issue here;
104: what is important is that we have a two--scale picture.
105: We stress that the hierarchy (\ref{hierarchy}) is really a logical
106: necessity --- if the size of the constituent quark were comparable
107: to that of the nucleon, we would not be ``seeing it''
108: as an independent dynamical entity.
109:
110: In QCD, the leading ($1/Q^2$) power corrections to the lowest
111: moments of the spin structure functions $g_1$ and $g_2$
112: are governed by the matrix elements of twist--3 and 4 operators
113: which measure non-perturbative correlations of the quark and gluon fields
114: in the nucleon (for details, see
115: Refs.\cite{Shuryak:1981pi,Ji:1994sv,Balla:1997hf}):
116: \begin{equation}
117: \begin{array}{lccl}
118: d_2 : \;\;\;\; & \bar\psi
119: \gamma_{\left\{\alpha\right.} \widetilde F_{\left.\beta\right\} \gamma}
120: \psi & \phantom{xxxx} & \mbox{Twist--3}
121: \\[.5cm]
122: f_2 : \;\;\;\; & \bar\psi \gamma_\alpha \widetilde F_{\beta\alpha} \psi &
123: & \mbox{Twist--4}
124: \end{array}
125: \label{ope}
126: \end{equation}
127: where $\widetilde F_{\alpha\beta} = (1/2)
128: \epsilon_{\alpha\beta\gamma\delta} F_{\gamma\delta}$ is the dual
129: gluon field strength. With the help of the well--known relation
130: \begin{equation}
131: F_{\alpha\beta} \;\; = \;\;
132: i \left[ \nabla_\alpha , \nabla_\beta \right],
133: \hspace{3em} \nabla_\alpha \equiv \partial_\alpha - i A_\alpha
134: \;\;\; \mbox{covariant derivative},
135: \label{identity}
136: \end{equation}
137: and making use of gamma matrix identities and the
138: equations of motion of the quark fields, the twist--4 operator
139: can equivalently be expressed as
140: \begin{equation}
141: \bar\psi \gamma_\beta \gamma_5 (-\nabla^2 ) \psi .
142: \label{axial_virtuality}
143: \end{equation}
144: In this form, it can be compared with the axial current operator,
145: which measures the quark contribution to the nucleon spin,
146: \begin{equation}
147: g_A : \;\;\;\; \bar\psi \gamma_\beta \gamma_5 \psi .
148: \label{axial}
149: \end{equation}
150: We see that the operator (\ref{axial_virtuality}) measures
151: the correlation of the spin of the quarks with their virtuality
152: (four--momentum squared, $k^2$) in the nucleon.
153:
154: %
155: \begin{figure}[t]
156: %
157: \begin{center}
158: \includegraphics[width=8.0cm,height=4cm]{gadist.eps}
159: \end{center}
160: %
161: \caption[*]{}
162: \label{fig_gadist}
163: \end{figure}
164: %
165: The constituent quark picture implies that,
166: generally speaking, the distribution of virtualities of quarks in
167: the nucleon has two components. The bulk
168: of the distribution is governed by the size of the nucleon,
169: $-k^2 \sim R^{-2}$. In addition, there is a ``tail'' extending up to
170: values of the order of the inverse size of the constituent
171: quark, $-k^2 \sim \rho^{-2}$. This two--component structure, which
172: follows from the Fourier image of the two--scale picture of
173: Fig.~\ref{fig_twoscale}, is the key to our estimates of
174: higher--twist matrix elements.
175:
176: Specifically, let us consider the distribution of quark virtualities
177: in the proton's isovector (flavor--nonsinglet) axial charge, $g_A$,
178: shown schematically in Fig.~\ref{fig_gadist}. In the isovector case one
179: can argue that the large--virtuality tail of the distribution
180: is of positive sign and decays as $1/(-k^2)$,
181: until it is ``cut off'' by the constituent quark size, $\rho$.
182: This follows from the requirement that in the limit of small size of the
183: constituent quark, $\rho \ll R$, the axial charge should exhibit the
184: logarithmic divergence it has in QCD (with $\rho^{-1}$ acting as the
185: ultraviolet cutoff). We note that exactly this behavior is found
186: also in a field--theoretical chiral model in which the constituent
187: quarks/antiquarks couple to a pion field.\footnote{In the isoscalar
188: case the behavior is different, due to the presence of the
189: $U(1)$ anomaly. The following arguments do not apply in this case.}
190: Thus, the isovector axial charge,
191: which is the integral of the distribution shown in Fig.~\ref{fig_gadist},
192: behaves parametrically as
193: \begin{equation}
194: g_A \;\; = \;\; \int_0^\infty d(-k^2) \; g_A (-k^2)
195: \;\; \sim \log \frac{\rho}{R} .
196: \label{g_A_integral}
197: \end{equation}
198: This integral is dominated by virtualities
199: $-k^2 \sim R^{-2} \ll \rho^{-2}$. Consider now the corresponding
200: integral for the isovector (flavor--nonsinglet) twist--4 matrix
201: element, $f_2^{u - d}$. Since the operator
202: (\ref{axial_virtuality}) involves an additional contracted
203: derivative, this quantity is determined by the integral
204: with an additional factor $k^2$, which is parametrically of the
205: order
206: \begin{equation}
207: M_N^2 f_2^{u - d}
208: \;\; = \;\; \int_0^\infty d(-k^2) \; g_A (-k^2) \; k^2
209: \;\; \sim \rho^{-2} .
210: \label{f_2_integral}
211: \end{equation}
212: Thus, the isovector twist--4 matrix element in our constituent
213: quark picture is governed by the size of the constituent quark.
214: Furthermore, since $k^2 < 0$ in the integral (\ref{f_2_integral}),
215: we can say that
216: \begin{equation}
217: f_2^{u - d} \;\; < \;\; 0 .
218: \label{f_2_sign}
219: \end{equation}
220: To summarize, the constituent quark picture suggests a sizable
221: negative value for $M_N^2 f_2^{u - d}$ in the proton,
222: of the order $\rho^{-2}$.
223: It is interesting that the estimates of $f_2^{u - d}$ obtained in
224: various QCD--based approaches are in qualitative agreement with
225: this prediction, see Table~\ref{table}. The QCD sum rule estimates
226: of Refs.\cite{Balitsky:1990jb,Stein:1995si}
227: as well as the instanton vacuum estimate of
228: Ref.\cite{Balla:1997hf} both give negative values of the order of
229: $\sim 0.1 - 0.3 \; {\rm GeV}^2$. These results support
230: the constituent quark interpretation of
231: higher--twist effects.
232: \begin{table}[ht]
233: \begin{tabular}{l|c}
234: & $M_N^2 f_2^{u - d} \, [{\rm GeV}^2]$ \\ \hline
235: QCD sum rules (Balitskii et al.) \cite{Balitsky:1990jb} & -0.18 \\ \hline
236: QCD sum rules (Stein et al.) \cite{Stein:1995si} & -0.06 \\ \hline
237: Instantons \cite{Balla:1997hf} & -0.22 \\ \hline
238: Bag model \cite{Ji:1994sv} & +0.1 \\ \hline
239: \end{tabular}
240: \caption[]{}
241: \label{table}
242: \end{table}
243:
244: The only exception in Table~\ref{table} is the bag model, which
245: gives a positive result for $f_2^{u - d}$. This model, however,
246: does not respect the QCD equations of motion [{\it i.e.}, the two
247: forms of the QCD operator, (\ref{ope}) and (\ref{axial_virtuality}),
248: would give inequivalent results], and therefore cannot claim to give a
249: realistic description of quark--gluon correlations in the nucleon.
250:
251: When applying the same reasoning as above to the twist--3 matrix element,
252: $d_2$, we find that after the substitution (\ref{identity})
253: the quark--gluon operator does not produce a contracted covariant
254: derivative. In this operator, all derivatives are ``kinematical'', {\it i.e.},
255: they are needed to support the spin of the matrix element.
256: This operator does not probe the virtuality of the quarks.
257: Its matrix element is not determined by the size of the constituent
258: quark, but by the size of the nucleon,
259: \begin{equation}
260: M_N^2 d_2 \;\; \sim \;\; R^{-2} .
261: \end{equation}
262: Thus, we find that our two--scale picture implies
263: (see also Ref.~\cite{Wandzura:1977qf})
264: \begin{equation}
265: \begin{array}{ccc}
266: |d_2| & \;\;\; \ll \;\;\; & |f_2| .\\
267: \mbox{Twist--3} & & \mbox{Twist--4}
268: \end{array}
269: \label{d_2_vs_f_2}
270: \end{equation}
271:
272: It is interesting to see to which extent the qualitative predictions
273: of the constituent quark picture are supported by the experimental data.
274: The twist--3 matrix element, $d_2$, can be extracted in a model--independent
275: way from the spin structure function $g_2$ (with the
276: Wandzura--Wilczek contribution subtracted). The SLAC E155X
277: experiment\cite{Anthony:2002hy} and the recent Jefferson Lab Hall A
278: analysis\cite{Zheng:2004ce} report values of $d_2^{p, n}$
279: of few times $10^{-3}$, which are more than an order of magnitude
280: smaller than the theoretical estimates for $f_2^{u - d}$ given
281: in Table~\ref{table} and our estimate for $f_2^{u - d}$ from $g_1$
282: data (see below), in agreement with the parametric ordering implied
283: by the constituent quark picture (\ref{d_2_vs_f_2}).
284:
285: The twist--4 matrix element, $f_2$, can only be extracted from
286: the power corrections to the $Q^2$--dependence of the structure
287: function $g_1$. The QCD expression for the $1/Q^2$
288: corrections to the first moment contains the sum of the
289: matrix elements $d_2$ and $f_2$; however, $d_2$ is known
290: from independent measurements. In fact, the parametric ordering
291: suggested by the constituent quark picture implies that
292: one should ascribe the $1/Q^2$ corrections of the first moment
293: of $g_1$ entirely to the twist--4 matrix element, $f_2$.
294:
295: %
296: \begin{figure}[t]
297: %
298: \includegraphics[width=9.4cm,height=7.5cm]{ht_prot_neutr.eps}
299: %
300: \caption[*]{}
301: \label{fig_fit}
302: \end{figure}
303: %
304: The dynamical higher--twist contribution has been extracted
305: from NLO QCD fits to the world data on the structure functions
306: $g_1^p (x, Q^2)$ and $g_1^n(x, Q^2)$ (with $Q^2 \geq 1\;
307: {\rm GeV}^2$).\cite{Leader:2002ni,Leader:2003vv} They are based on the ansatz
308: \begin{equation}
309: g_1 (x, Q^2) \;\; = \;\; g_1 (x, Q^2)_{\mbox{{\scriptsize LT + TMC}}}
310: \; + \; \frac{h^{g_1} (x)}{Q^2} ,
311: \label{g_1_QCD}
312: \end{equation}
313: where the leading--twist contribution (including target
314: mass corrections) is calculated using
315: the Leader--Stamenov--Sidorov parametrization
316: of polarized parton densities,
317: and $h^{g_1} (x)$ parametrizes the dynamical higher--twist
318: corrections. The results of Ref.~\cite{Leader:2002ni} for $h^{g_1} (x)$
319: for proton and neutron are shown in Fig.~\ref{fig_fit}
320: (filled circles). The second fit\cite{Leader:2003vv} (open circles)
321: includes also the new $g_1^n$ data from Jefferson
322: Lab Hall A\cite{Zheng:2003un}, as well as the preliminary
323: deuteron data from HERMES. One sees that the results for $h^{g_1} (x)$
324: obtained with the two data sets are nicely consistent. The new data
325: allow to significantly reduce the statistical uncertainty in the
326: higher--twist contribution. Integrating the higher--twist
327: contribution over $x$ we get
328: \begin{equation}
329: \int_0^1 dx \; h^{g_1} (x) \;\; = \;\;
330: \left\{\begin{array}{l}
331: 0.007 \pm 0.010 \; {\rm GeV}^2 \;\; \mbox{(proton)}
332: \\[0cm]
333: 0.049 \pm 0.007 \; {\rm GeV}^2 \;\; \mbox{(neutron)}
334: \end{array}
335: \right.
336: \end{equation}
337: Comparing with the QCD expression,
338: neglecting the twist--3 contribution, we obtain
339: \begin{equation}
340: M_N^2 f_2^{u - d} \;\; = \;\; -0.28 \pm 0.08 \; {\rm GeV}^2 .
341: \end{equation}
342: This estimate agrees both in sign and in order--of--magnitude
343: with the predictions of the constituent quark picture,
344: (\ref{f_2_integral}) and (\ref{f_2_sign}).
345:
346: Our result agrees well with that obtained by
347: Deur {\it et al.}\cite{Deur:2004ti} in a recent analysis of
348: power corrections to the Bjorken sum rule
349: (their $f_2^{p - n} \equiv \frac{1}{3} f_2^{u - d}$ in our conventions).
350: It disagrees in sign with the result of Kao {\it et al.}\cite{Kao:2003jd},
351: who use a resonance--based parametrization of the structure function
352: at low $Q^2$. The origin of this discrepancy remains to be understood.
353:
354: The constituent quark picture allows to draw some interesting
355: conclusions about the ``global'' properties of the transition from
356: high to low $Q^2$ in the nucleon spin structure functions
357: ({\it i.e.}, going beyond the leading $1/Q^2$ corrections).
358: Since the characteristic mass scale for the power corrections
359: is set by the size of the constituent quark,
360: one should expect the twist expansion
361: to break down at momenta of the order $Q^2 \sim \rho^{-2}$.
362: For the extraction of the leading ($1/Q^2$--) corrections from
363: QCD fits to the data this implies that one should restrict oneself
364: to the range $Q^2 \gg \rho^{-2}$, where the leading term in the
365: series dominates. Physically speaking, in this region the scattering
366: process takes place ``inside'' the constituent quark.
367:
368: When $Q^2$ is decreased to values of the order $R^{-2}$, the scattering
369: process probes the motion of the constituent quarks in the nucleon.
370: This is the region dominated by nucleon resonances. In the constituent
371: quark picture, these are changes of the state of motion of the
372: constituent quarks at the scale $R$, which do not affect the
373: internal structure of the constituent quark at the scale $\rho$.
374: Thus, our two--scale picture implies a clear distinction between
375: resonance and higher--twist contributions. It is close in spirit
376: to the parametrization of the $Q^2$ dependence of the first moment
377: of $g_1$ by Ioffe and Burkert\cite{Burkert:1992tg}, in which the contribution
378: from the Delta resonance is separated from the continuum, and
379: the leading power corrections are associated with the
380: continuum contribution.\footnote{It is interesting to note that the
381: mass scale governing the power corrections in the Ioffe--Burkert
382: parametrization\cite{Burkert:1992tg}, $M_\rho^2$,
383: is numerically close to value associated
384: with the constituent quark size,
385: $\rho^{-2} \sim (0.3 \, {\rm fm})^{-2} = (600 \, {\rm MeV})^2$.}
386:
387: To summarize, we have shown that the assumption of a two--scale
388: ``constituent quark structure'' of the nucleon implies certain
389: qualitative statements about the twist--3 and 4 matrix elements,
390: which are in agreement with present polarized DIS data.
391: For more quantitative estimates, one eventually has to turn
392: to dynamical models. A consistent realization of the scenario developed
393: here is provided by the instanton model of the QCD vacuum, in which the
394: hierarchy (\ref{hierarchy}) follows from the diluteness of the instanton
395: medium.\cite{Balla:1997hf} In particular, this model
396: incorporates the chiral dynamics at the scale $R$, {\it i.e.}, the binding
397: of the constituent quarks and antiquarks in the
398: nucleon\cite{Diakonov:2002fq}, and thus can serve as the basis of an
399: ``interpolating'' description connecting the scaling region at
400: large $Q^2$ with the photoproduction point.
401:
402: Finally, the constituent quark picture suggested here
403: can be applied also to the power corrections to the
404: unpolarized structure functions; the relation of our approach
405: to that of Petronzio {\it et al.}\cite{Petronzio:2003bw}
406: will be discussed elsewhere\cite{paper}.
407:
408: This work was supported by U.S.\ Department of Energy contract \\
409: DE-AC05-84ER40150, under which the Southeastern Universities Research
410: Association (SURA) operates the Thomas Jefferson National
411: Accelerator Facility. A.~S.\ acknowledges financial support by
412: RFBR (No 02-01-00601, 03-02-16816), INTAS 2000 (No 587),
413: and the Heisenberg--Landau program.
414: The work of C.~W.\ was partly supported by DFG (Heisenberg Fellowship).
415: %
416: \begin{thebibliography}{0}
417: %
418: %
419: \bibitem{Diakonov:2002fq}
420: For a review, see: D.~Diakonov,
421: Prog.\ Part.\ Nucl.\ Phys.\ {\bf 51}, 173 (2003).
422: %%CITATION = HEP-PH 0212026;%%
423: %
424: %
425: \bibitem{fest} For a recent discussion, see: L.~Frankfurt,
426: M.~Strikman, and C.~Weiss, in: Recent Developments in
427: hadron spectroscopy and reactions (Eds.~M.~Polyakov, P.~Schweitzer,
428: and C.~Weiss), to appear in Annalen der Physik (2004).
429: %
430: %
431: \bibitem{paper} Details will be given in:
432: A.~V.~Sidorov and C.~Weiss, in preparation.
433: %
434: %
435: \bibitem{Shuryak:1981pi}
436: E.~V.~Shuryak and A.~I.~Vainshtein,
437: Nucl.\ Phys.\ B {\bf 201}, 141 (1982).
438: %
439: %
440: \bibitem{Ji:1994sv}
441: X.~Ji and P.~Unrau,
442: Phys.\ Lett.\ {\bf B333} (1994) 228.
443: %
444: %
445: \bibitem{Balla:1997hf}
446: J.~Balla, M.~V.~Polyakov and C.~Weiss,
447: Nucl.\ Phys.\ B {\bf 510}, 327 (1998).
448: %
449: %
450: \bibitem{Balitsky:1990jb}
451: I.~I.~Balitsky, V.~M.~Braun and A.~V.~Kolesnichenko,
452: Phys.\ Lett.\ {\bf B242} (1990) 245.
453: Erratum Phys.\ Lett.\ {\bf B318} (1993) 648.
454: %
455: %
456: \bibitem{Stein:1995si}
457: E.~Stein {\it et al.}, Phys.\ Lett.\ {\bf B353} (1995) 107.
458: %
459: %
460: \bibitem{Wandzura:1977qf}
461: S.~Wandzura and F.~Wilczek,
462: Phys.\ Lett.\ B {\bf 72}, 195 (1977).
463: %%CITATION = PHLTA,B72,195;%%
464: %
465: %
466: \bibitem{Anthony:2002hy}
467: P.~L.~Anthony {\it et al.} [E155 Collaboration],
468: Phys.\ Lett.\ B {\bf 553}, 18 (2003).
469: %
470: %
471: \bibitem{Zheng:2004ce}
472: X.~Zheng {\it et al.} [Jefferson Lab Hall A Collaboration],
473: arXiv:nucl-ex/0405006.
474: %%CITATION = NUCL-EX 0405006;%%
475: %
476: %
477: \bibitem{Leader:2002ni}
478: E.~Leader, A.~V.~Sidorov and D.~B.~Stamenov,
479: Phys.\ Rev.\ D {\bf 67}, 074017 (2003).
480: %%CITATION = HEP-PH 0212085;%%
481: %
482: %
483: \bibitem{Leader:2003vv}
484: E.~Leader, A.~V.~Sidorov and D.~B.~Stamenov,
485: arXiv:hep-ph/0309048.
486: %
487: %
488: \bibitem{Zheng:2003un}
489: X.~Zheng {\it et al.} [Jefferson Lab Hall A Collaboration],
490: Phys.\ Rev.\ Lett.\ {\bf 92}, 012004 (2004).
491: %%CITATION = NUCL-EX 0308011;%%
492: %
493: %
494: \bibitem{Deur:2004ti}
495: A.~Deur {\it et al.},
496: arXiv:hep-ex/0407007.
497: %%CITATION = HEP-EX 0407007;%%
498: %
499: %
500: \bibitem{Kao:2003jd}
501: C.~W.~Kao {\it et al.},
502: Phys.\ Rev.\ D {\bf 69}, 056004 (2004).
503: %
504: %
505: \bibitem{Burkert:1992tg}
506: V.~D.~Burkert and B.~L.~Ioffe,
507: Phys.\ Lett.\ B {\bf 296}, 223 (1992).
508: %%CITATION = PHLTA,B296,223;%%
509: %
510: %
511: \bibitem{Petronzio:2003bw}
512: R.~Petronzio, S.~Simula and G.~Ricco,
513: Phys.\ Rev.\ D {\bf 67}, 094004 (2003)
514: [Erratum-ibid.\ D {\bf 68}, 099901 (2003)]
515: %%CITATION = HEP-PH 0301206;%%
516: %
517: %
518: \end{thebibliography}
519: %
520: \end{document}
521: %
522: %
523: