1: %
2: % ARTICLE ``SPIN--DEPENDENT TWIST--4 MATRIX ELEMENTS FROM THE INSTANTON
3: % VACUUM: FLAVOR--SINGLET AND NONSINGLET''
4: % AUTHORS NAM-YOUNG LEE, K. GOEKE, C. WEISS
5: %
6: \documentclass[12pt]{article}
7: %
8: % PACKAGES
9: %
10: \usepackage{graphicx}
11: \usepackage[figuresright]{rotating}
12: %
13: \usepackage{hyperref}
14: \newcommand{\wwwspires}{http://www.slac.stanford.edu/spires/find/hep/www}
15: \newcommand{\myhrefstring}{SPIRES}
16: %\newcommand{\myhrefstring}{}
17: %
18: % FORMATING (A4)
19: %
20: \setlength{\textwidth}{16.1cm}
21: \setlength{\textheight}{22cm}
22: \setlength{\oddsidemargin}{-0.1cm}
23: \setlength{\topmargin}{-1.2cm}
24: %
25: % GENERAL DEFINITIONS
26: %
27: \newcommand{\beq}{\begin{equation}}
28: \newcommand{\eeq}{\end{equation}}
29: \newcommand{\be}{\begin{eqnarray}}
30: \newcommand{\ee}{\end{eqnarray}}
31: %
32: \newcommand{\Fdual}{\widetilde{F}}
33: \newcommand{\fourint}[1]{\int\!\frac{d^4 #1}{(2\pi)^4}}
34: %
35: \begin{document}
36: %
37: \rightline{RUB-TPII-06/01}
38: \rightline{hep-ph/0105173}
39: %
40: \vspace{.3cm}
41: %
42: \begin{center}
43: \begin{large}
44: {\bf Spin--dependent twist--4 matrix elements from the instanton \\
45: vacuum: Flavor--singlet and nonsinglet} \\[1cm]
46: \end{large}
47: %
48: \vspace{1.0cm}
49: {\bf Nam--Young Lee}$^{\rm a, 1}$,
50: {\bf K.\ Goeke}$^{\rm a, 2}$
51: {\bf and C. Weiss}$^{\rm b, 3}$
52: \\[1.cm]
53: $^{a}${\em Institut f\"ur Theoretische Physik II,
54: Ruhr--Universit\"at Bochum, \\ D--44780 Bochum, Germany}
55: \\
56: $^{b}${\em Institut f\"ur Theoretische Physik,
57: Universit\"at Regensburg, \\ D--93053 Regensburg, Germany}
58: \end{center}
59: %
60: \vspace{1.5cm}
61: %
62: \begin{abstract}
63: \noindent
64: We estimate the twist--4 spin--1 nucleon matrix element $f_2$
65: in an instanton--based description of the QCD vacuum.
66: In addition to the flavor--nonsinglet we compute also the flavor--singlet
67: matrix element, which appears in next--to--leading order of the
68: $1/N_c$--expansion. The corresponding twist--3 spin--2 matrix elements $d_2$
69: are suppressed in the packing fraction of the instanton medium,
70: $\bar\rho / \bar R \ll 1$. We use our results to estimate the leading $1/Q^2$
71: power corrections to the first moment of the proton and neutron
72: spin structure functions $G_1$, as well as the intrinsic charm contribution
73: to the nucleon spin.
74: \end{abstract}
75: %
76: \vfill
77: \rule{5cm}{.2mm} \\
78: {\footnotesize $^{\rm 1}$ E-mail: lee@tp2.ruhr-uni-bochum.de} \\
79: {\footnotesize $^{\rm 2}$ E-mail: goeke@tp2.ruhr-uni-bochum.de} \\
80: {\footnotesize $^{\rm 3}$ E-mail: weiss@tp2.ruhr-uni-bochum.de} \\
81: %
82: \newpage
83: %
84: %
85: \section{Introduction \label{introduction}}
86: %
87: The structure of the nucleon as measured in deep--inelastic scattering
88: is described by matrix elements of QCD operators of a certain
89: twist (dimension minus spin). The moments of the structure functions
90: at asymptotically large $Q^2$ are given by matrix elements of operators of
91: twist 2, which have a simple interpretation as number densities of the
92: quarks and antiquarks in the nucleon in the infinite--momentum frame.
93: Power ($1/Q^2$--) corrections to the asymptotic result are determined by
94: matrix elements of operators of twist 3 and 4. The latter describe either
95: the correlation of the quark fields with the non-perturbative gluon
96: field in the target, or quark--quark correlations. In the polarized
97: case the twist--3 and 4 matrix elements of lowest spin are
98: that of the twist--4 spin--1 operator
99: \be
100: \langle p, \lambda | \bar\psi_f \gamma_\alpha \Fdual^{\beta\alpha}
101: \psi_f | p, \lambda \rangle &=& 2 M_N^2 \; f_{2,f} \; s^\beta ,
102: \label{f2_def}
103: \ee
104: and of the twist--3 spin--2 operator
105: \be
106: \lefteqn{ \langle p , \lambda |
107: \bar\psi_f \left( \gamma^\alpha \Fdual^{\beta\gamma}
108: + \gamma^\beta \Fdual^{\alpha\gamma} \right) \psi_f
109: | p, \lambda \rangle - \mbox{traces}} && \nonumber \\
110: &=& 2 \, d_{2, f} \; \left[ 2 p^\alpha p^\beta s^\gamma
111: - p^\gamma p^\beta s^\alpha - p^\alpha p^\gamma s^\beta
112: + (\alpha \leftrightarrow \beta ) - \mbox{traces} \right] .
113: \label{d2_def}
114: \ee
115: Here $\bar\psi_f, \psi_f$ denote the quark fields of flavor $f$,
116: \be
117: \Fdual^{\mu\nu} &\equiv& \frac{1}{2} \epsilon^{\mu\nu\rho\sigma}
118: F_{\rho\sigma}
119: \ee
120: is the dual of the gauge field strength ($\epsilon^{0123} = 1$),
121: $p$ the nucleon four--momentum ($p^2 = M_N^2$), $\lambda$ the
122: helicity, and $s = s(\lambda )$ the polarization
123: vector of the nucleon state, which satisfies
124: \beq
125: s\cdot p \;\; = \;\; 0, \hspace{2cm}s^2 \;\; =\;\; -M_N^2 .
126: \eeq
127: The nucleon states are normalized according to
128: $\langle p, \lambda |p', \lambda' \rangle =
129: 2p^0(2\pi)^3\delta^{(3)}({\bf p} -{\bf p}') \delta_{\lambda\lambda'}$.
130: The matrix elements (\ref{f2_def}) and (\ref{d2_def}) can be taken either
131: in a proton or neutron state.
132: \par
133: The matrix elements
134: (\ref{f2_def}) and
135: (\ref{d2_def}) together determine the leading power corrections to the
136: first moment of the spin structure function $G_1$
137: (the Bjorken and Ellis--Jaffe sum
138: rules) \cite{Shuryak:1982pi,Ji:1994sv,Ehrnsperger:1994hh}. The twist--3
139: matrix element $d_2$ also appears in the QCD expression for the third
140: moment of the second spin structure function $G_2$, where it is not
141: power--suppressed relative to the twist--2 part and can therefore be
142: measured with
143: good accuracy \cite{Bosted:2000tf}. By studying these matrix elements in
144: theoretical models of the nucleon one may hope to get some insight
145: into the transition from the region of asymptotically large $Q^2$
146: ($Q^2 \gg 1 \, {\rm GeV}^2$),
147: where the $Q^2$--dependence of the structure functions is described
148: by perturbative QCD, to the resonance region
149: ($Q^2 \approx 1 \, {\rm GeV}^2$) \cite{Edelmann:2000yp}. Eventually, this may
150: help to understand also the discrepancy of the moments of the structure
151: functions
152: measured at $Q^2 \ge 1 \, {\rm GeV}^2$ with the GDH sum rule
153: for photoabsorption ($Q^2 = 0$) \cite{Drechsel:1995az}.
154: \par
155: The twist--4 matrix element $f_2$, (\ref{f2_def}), also plays
156: a role in the so-called ``intrinsic'' charm contribution
157: to the nucleon spin \cite{Polyakov:1999rb}.
158: Making use of an expansion in inverse powers of the charm quark
159: mass, $1/m_c$, Franz {\it et al.}\ \cite{Franz:1999hw}, correcting
160: an earlier results \cite{Halperin:1997jq,Araki:1998yk}, related the charm
161: quark contribution to the axial current to the matrix element of
162: the flavor--singlet twist--4 operator, (\ref{f2_def}).
163: Model estimates of flavor--singlet $f_2$ can thus be used to
164: estimate the intrinsic charm contribution to the nucleon spin.
165: Knowledge of this contribution is
166: a prerequisite for attempts to measure the gluon polarization
167: through open charm production \cite{Baum:1996yv}.
168: \par
169: Here we report about an estimate of the spin--dependent
170: twist--3 and 4 nucleon matrix elements, (\ref{f2_def}) and
171: (\ref{d2_def}), within an instanton--based description of the QCD vacuum.
172: In this approach the QCD ground state is approximately described as a
173: ``medium'' of topological vacuum fluctuations --- instantons and
174: antiinstantons, with perturbative fluctuations about
175: them \cite{Shuryak:1982ff,Diakonov:1984hh,Schafer:1998wv}.
176: This picture explains the dynamical breaking
177: of chiral symmetry \cite{Diakonov:1986eg},
178: which happens due to the fermionic zero modes associated with
179: the individual (anti--) instantons, and manifests itself in the appearance of
180: a dynamical mass of the quarks, accompanied by a coupling to Goldstone
181: bosons --- the pions. Building on that, this picture explains
182: a host of phenomenological data on ``vacuum structure''
183: and hadronic correlation functions \cite{Schafer:1998wv}.
184: It also gives rise to a successful description of the nucleon
185: as a chiral soliton, in the formal limit of a large number of colors,
186: $N_c \rightarrow \infty$ \cite{Diakonov:1988ty,Christov:1996vm}.
187: \par
188: A crucial element in this approach is the fact that the instanton medium is
189: dilute, {\it i.e.}\ the average size of the (anti--) instantons,
190: $\bar\rho \approx (600\, {\rm MeV})^{-1}$, is small
191: compared to their average distance, $\bar R \approx (200 \, {\rm MeV})^{-1}$,
192: with $\bar\rho / \bar R \approx 1/3$. The existence of this small parameter
193: makes possible a systematic analysis of non-perturbative effects generated
194: by the instantons.
195: Matrix elements of quark--gluon operators of twist $> 2$ have been studied in
196: this approach in Ref.\cite{Balla:1998hf}. In particular, it was shown
197: there that the matrix element of the twist--3 operator, $d_2$, is
198: suppressed relative to the twist--4 one, $f_2$, by a factor of
199: $(\bar\rho / \bar R )^4$, and thus
200: \beq
201: \begin{array}{lcr}
202: d_2 & \ll & f_2 . \\[1ex]
203: \mbox{twist-3} & & \mbox{twist-4}
204: \end{array}
205: \eeq
206: In Ref.\cite{Balla:1998hf} the value of $d_2$ was estimated to be of the
207: order of $10^{-3}$; this prediction is confirmed by the recent results
208: of the E155x experiment \cite{Bosted:2000tf}. Thus, in the instanton vacuum
209: the parametrically leading higher--twist effects are due to the
210: twist--4 matrix element $f_2$. In this letter we present a
211: quantitative estimate of $f_2$ in the instanton vacuum.
212: In addition to the flavor--nonsinglet
213: nucleon matrix element, which appears in leading order of the $1/N_c$
214: expansion and has been calculated in Ref.\cite{Balla:1998hf}, we
215: calculate here also the flavor--singlet one. This allows us to
216: make predictions for the power corrections to both the proton and
217: neutron spin structure functions $G_1$, as well as to estimate the intrinsic
218: charm contribution to the nucleon spin. For a detailed description of the
219: foundations of the approach employed in the present investigation
220: here we refer to Refs.\cite{Balla:1998hf,Diakonov:1996qy}.
221: %
222: \section{Twist--4 matrix elements from the instanton vacuum}
223: %
224: {\it Chiral symmetry breaking by instantons.}
225: Following Diakonov and Petrov \cite{Diakonov:1986eg,Diakonov:1986aj},
226: the dynamical breaking of
227: chiral symmetry in the instanton vacuum can be studied in the
228: large--$N_c$ limit, where it manifests itself in the appearance
229: of a dynamical quark mass, $M \neq 0$, accompanied by a
230: coupling of the quarks to a pion (Goldstone boson) field, $\pi^a (x)$,
231: $(a = 1, \ldots N_f^2 - 1)$ which arises
232: from the ``bosonization'' of the 't Hooft many--fermionic interaction
233: induced by the instantons. The resulting low--energy dynamics is
234: summarized by the effective Lagrangian
235: \beq
236: L_{\rm eff} \;\; = \;\; \bar\psi (x) \left[ i\!{\not\partial}
237: - M F(\partial^2) \, U^{\gamma_5}(x)
238: F(\partial^2 ) \right] \psi(x),
239: \label{L_eff}
240: \eeq
241: where
242: \beq
243: U^{\gamma_5}(x) \;\; = \;\; \frac{1 + \gamma_5}{2} U(x)
244: + \frac{1 - \gamma_5}{2} U^\dagger (x),
245: \hspace{4em} U(x) \;\; = \;\; e^{i \pi^a (x) \tau^a} ,
246: \eeq
247: is a unitary matrix variable containing the pion field,
248: and $F(p^2 )$ are form factors, proportional to the
249: Fourier transform of the zero mode of the (anti--) instantons,
250: which drop to zero for spacelike momenta of the order
251: of $-p^2 \sim \bar\rho^{-2}$. It is crucial that the
252: dynamical quark mass, $M$, is parametrically small compared to the
253: ultraviolet cutoff, $\bar\rho^{-1}$:
254: \beq
255: M\bar\rho\; \sim \;\left(\frac{\bar\rho}{\bar R}\right)^2 .
256: \eeq
257: In particular, this implies that for spacelike quark momenta of the
258: order $M \ll \bar\rho^{-1} \approx 600\, {\rm MeV}$
259: one may neglect the form factors
260: and consider the quark--pion coupling as effectively pointlike.
261: \par
262: {\it Nucleon as chiral soliton.}
263: The effective Lagrangian (\ref{L_eff}) serves as a basis for the
264: calculation of correlation functions in the instanton vacuum
265: within the $1/N_c$--expansion \cite{Diakonov:1986eg,Diakonov:1986aj}.
266: In particular, it gives rise to a picture of the nucleon as a chiral
267: soliton \cite{Diakonov:1988ty}. Baryon correlation functions
268: in the large--$N_c$ limit are characterized by a classical pion field,
269: which in the baryon rest frame is of ``hedgehog'' form
270: \beq
271: U_c ({\bf x}) \;\;=\;\; e^{i n^a \tau^a P(r)} ,
272: \hspace{4em} r \; \equiv \; |{\bf x}|,
273: \hspace{4em} {\bf n} \; = \; \frac{\bf x}{r} .
274: \label{hedge}
275: \eeq
276: Baryon states of definite spin/isospin and momentum quantum numbers emerge
277: from quantizing the collective rotations and translations of the
278: classical soliton
279: \beq
280: U({\bf x}, t) \;\; = \;\;
281: R (t) U_c ({\bf x}-{\bf X}(t)) R^\dagger (t) .
282: \label{U_time_dependent}
283: \eeq
284: Since the soliton moment of inertia and mass are of order $N_c$,
285: the angular velocity of the collective rotation,
286: \beq
287: \Omega\;\; = \;\; -iR^\dag\dot{R} ,
288: \label{angular_vel}
289: \eeq
290: as well as the linear velocity of the translation, are of order
291: $1/N_c$, the quantization of the collective motion can be performed
292: within a $1/N_c$ expansion \cite{Diakonov:1988ty}.
293: The $\Omega^2$--contribution to the baryon mass gives rise to the
294: $N$--$\Delta$ mass splitting, which is of order $1/N_c$.
295: This approach allows also to compute matrix elements of
296: operators between baryon states. Technically, this is done by
297: representing the matrix element as a functional integral over
298: the collective translations/rotations, which is then evaluated
299: by expanding the integrand in $\Omega$. For details we refer to
300: Refs.\cite{Diakonov:1988ty,Christov:1996vm}.
301: %
302: % FIGURE
303: %
304: \begin{figure}[t]
305: %
306: \begin{center}
307: \includegraphics[width=12.0cm,height=1.86cm]{effop.eps}
308: \end{center}
309: %
310: \caption[]{Schematic illustration of the contributions to the matrix element
311: of the quark--gluon operator, (\ref{f2_def}), in the instanton vacuum.
312: The cross denotes the quark--gluon operator, the solid line the quark
313: propagator, the dashed line the gauge field of the (anti--) instanton,
314: and the circle the chirality--flipping 't Hooft interaction of the
315: instanton with the quark mediated by the zero modes. The picture
316: shows the situation for one light quark flavor; for $N_f$ flavors the
317: 't Hooft interaction is a $2 N_f$--fermionic vertex, which
318: at large $N_c$ is equivalent to a coupling of the quark to the pion field
319: (``bosonization''). Using the equations of motion of the effective low--energy
320: theory, the instanton--induced vertex shown here can be converted to
321: the chirally even effective operator (\ref{f2_equiv}).
322: For details, see Refs.\cite{Balla:1998hf,Dressler:1999zi}.}
323: \label{fig_effop}
324: \end{figure}
325: %
326: \par
327: {\it Twist--4 quark--gluon operator in the instanton vacuum.}
328: The instanton vacuum, being
329: a microscopic model of the non-perturbative fluctuations of the gluon
330: field, allows to estimate hadronic matrix elements of quark--gluon operators,
331: such as the twist--4 operator of Eq.(\ref{f2_def})
332: \cite{Diakonov:1996qy,Balla:1998hf}. It is understood that
333: the QCD operators are normalized at the scale defined by the
334: inverse average instanton size, $\mu = \bar\rho^{-1}$.
335: The main idea of this approach is that because of the diluteness of the
336: instanton medium the gauge field operator in (\ref{f2_def}) can
337: effectively be replaced by the field of a {\it single}
338: (anti--) instanton, which then interacts with the quark fields through
339: its zero modes (this is schematically illustrated in Fig.~\ref{fig_effop}).
340: As a result, the original QCD quark--gluon operator is replaced by a
341: chirality--flipping ``effective operator'', which is to be evaluated in
342: the effective theory where chiral symmetry is spontaneously
343: broken. As has been discussed in Refs.\cite{Balla:1998hf,Dressler:1999zi},
344: this effective operator can be brought to a chirality--conserving
345: form by making use of the equations of motion of the effective theory.
346: For the QCD operator of interest,
347: $\bar\psi_f \gamma_\alpha \Fdual_{\beta\alpha} \psi_f$,
348: the resulting operator is of the form \cite{Balla:1998hf}
349: \beq
350: - f_2^{\rm quark} \;
351: \bar\psi_f \gamma_\beta \gamma_5 \partial^2 \psi_f .
352: \label{f2_equiv}
353: \eeq
354: It is understood here that the momenta of the quark fields are
355: of the order $p \sim M \ll \bar\rho^{-1}$. Here $f_2^{\rm quark}$
356: denotes the matrix element of the
357: original chirality--flipping effective operator in a quark state,
358: given by
359: \beq
360: f_2^{\rm quark} \;\; = \;\; I_1 (-p^2 ) |_{p^2 = M^2} ,
361: \label{f2_from_I}
362: \eeq
363: where $I_1 (-p^2)$ denotes the quark loop integral defined
364: by ($\bar p$ and $\bar k$ are Euclidean momenta, with $\bar p^2 = - p^2$)
365: \beq
366: I_1 (\bar p^2 ) \;\; = \;\; \bar\rho^2 \fourint{\bar k}
367: \frac{F(\bar p - \bar k) {\cal G} (\bar k)}{(\bar p - \bar k)^2}
368: \left[ -\frac{3}{4} \frac{|\bar k|}{|\bar p|} C_1^{(1)} (\cos\theta )
369: + \frac{1}{2} C_2^{(1)} (\cos\theta ) \right] ,
370: \label{f2_integral}
371: \eeq
372: in which $C_n^{(1)}$ are the Gegenbauer polynomials (4--dimensional
373: spherical harmonics) of argument
374: $\cos\theta = (\bar k\cdot \bar p)/(|\bar k||\bar p|)$, and the function
375: ${\cal G} (\bar k)$
376: describes the Fourier transform of the (dual) field strength of
377: a single (anti--) instanton in singular gauge (see Ref.\cite{Balla:1998hf}
378: for details):
379: \be
380: {\cal G} (\bar k)
381: &=& 32 \pi^2 \left[ \left(\frac{1}{2} + \frac{4}{t^2} \right)
382: K_0 (t) + \left( \frac{2}{t} + \frac{8}{t^3} \right) K_1 (t)
383: - \frac{8}{t^4} \right], \hspace{1cm} t \; = \; |\bar k|\rho ,
384: \label{G}
385: \ee
386: with $K_n (t)$ the modified Bessel functions of the second kind. The
387: numerical value of $f_2^{\rm quark}$ in the limit $M\bar\rho \rightarrow 0$
388: is \cite{Balla:1998hf}
389: \beq
390: f_2^{\rm quark} \;\; = \;\; \frac{3}{5} \;\; = \;\; 0.6 .
391: \label{f2_quark}
392: \eeq
393: \par
394: The instanton--induced effective operator (\ref{f2_equiv}) has the
395: form of the axial current operator, but with an additional contracted
396: derivative acting on the quark fields. Note that the coefficient,
397: $f_2^{\rm quark}$, is of order unity in the packing
398: fraction of the instanton medium, $\bar\rho / \bar R$.
399: Technically speaking, this
400: happens because the quark loop integral $I_1$, (\ref{f2_integral}),
401: contains a would--be quadratic divergence, which is regularized
402: by the instanton form factors and cancels the overall factor
403: $\bar\rho^2$ in front of the integral. This implies that the
404: hadronic matrix elements of this operator will also be parametrically
405: large in the instanton packing fraction.
406: We remark that for the twist--3 operator in the matrix element,
407: $d_2$, (\ref{d2_def}), the effective operator
408: would be proportional to $(M\bar\rho )^2 \sim (\bar\rho / \bar R)^4$ and
409: thus parametrically suppressed.
410: \par
411: {\it Nucleon matrix elements: $1/N_c$ expansion.}
412: The nucleon matrix elements of the effective operator, (\ref{f2_equiv}),
413: can be computed using the same techniques as have been used in
414: the calculation of nucleon matrix elements of vector and axial vector
415: current operators (for a review, see \cite{Christov:1996vm}).
416: In particular, we shall exploit the close relation of the
417: effective operator, (\ref{f2_equiv}), with the axial current operator.
418: It is known that within the chiral soliton picture of the nucleon
419: the isovector and isoscalar component of the nucleon axial coupling
420: constant appear in different orders in the expansion in powers of the
421: angular velocity of the collective rotation of the soliton,
422: $\Omega \sim 1/N_c$, (\ref{angular_vel}). We therefore need to
423: consider separately the isovector and isoscalar components of the
424: twist--4 nucleon matrix element $f_2$, (\ref{f2_def}),
425: \beq
426: f_2^{(3)} \;\; = \;\; f_{2, u} - f_{2, d},
427: \hspace{2cm}
428: f_2^{(0)} \;\; = \;\; f_{2, u} + f_{2, d}
429: \label{f2_vec_scal}
430: \eeq
431: (the matrix elements on the R.H.S.\ refer to the proton). We restrict
432: ourselves for the moment to the $SU(2)$ flavor group ($u, d$), the
433: extension to $SU(3)$ will be discussed below.
434: For the isospin components defined by (\ref{f2_vec_scal}) the $N_c$--counting
435: can immediately be inferred from that of the corresponding
436: axial coupling constants of the nucleon:
437: \beq
438: M_N^2 f_2^{(3)} \;\; \sim \;\; N_c,
439: \hspace{2cm}
440: M_N^2 f_2^{(0)} \;\; \sim \;\; 1.
441: \eeq
442: The isovector component of the spin--dependent twist--4 matrix element
443: is leading in the $1/N_c$ expansion.
444: \par
445: {\it Isovector matrix element} $f_2^{(3)}$.
446: We now make a quantitative estimate of the isovector twist--4 nucleon matrix
447: element, $f_2^{(3)}$. This can be done by considering the
448: hypothetical limit of large soliton size, in which one can perform an
449: expansion of the matrix element in gradients of the classical pion
450: field, (\ref{hedge}). The result for $f_2^{(3)}$ in this approximation
451: can immediately be written down by analogy with the well--known result
452: for the isovector axial coupling, $g_A^{(3)}$ \cite{Diakonov:1988ty},
453: \be
454: g_A^{(3)} &=& \frac{4 N_c M^2 J}{9} \int d^3 x \;
455: {\rm tr}\,\left[ -i \tau^a U_c^\dagger ({\bf x})
456: \partial_a U_c ({\bf x}) \right] ,
457: \label{ga_grad}
458: \ee
459: where $J$ denotes the (Euclidean) quark loop integral
460: \beq
461: J \;\; = \;\; \fourint{\bar k} \;
462: \frac{ F^2 (\bar k) \left[ F^2 (\bar k) - \bar k^2 F(\bar k) F' (\bar k)
463: \right]}
464: {\left[\bar k^2 + M^2 F^4 (\bar k) \right]^2} .
465: \label{J}
466: \eeq
467: (Note that the PCAC relation relates this integral to the pion decay
468: constant, $4 N_c M^2 J = F_\pi^2$.) The corresponding result for the
469: isovector matrix element of the operator (\ref{f2_equiv}) is
470: \be
471: M_N^2 f_2^{(3)} &=& \frac{4 N_c M^2 f_2^{\rm quark} J_1}{9} \int d^3 x \;
472: {\rm tr}\,\left[ -i \tau^a U_c^\dagger ({\bf x})
473: \partial_a U_c ({\bf x}) \right] ,
474: \label{f2_grad}
475: \ee
476: where
477: \be
478: J_1 &=& \fourint{\bar k} \;
479: \frac{ \bar k^2 F^2 (\bar k) \left[ F^2 (\bar k) - \bar k^2
480: F (\bar k) F' (\bar k) \right]}
481: {\left[\bar k^2 + M^2 F^4 (\bar k) \right]^2} ,
482: \label{I_f_2}
483: \ee
484: which differs from $J$, (\ref{J}), by an additional power of the
485: quark virtuality, $\bar k^2$, in the integrand.
486: For the ratio of $f_2^{(3)}$ to $g_A^{(3)}$ we thus obtain
487: \beq
488: \frac{M_N^2 f_2^{(3)}}{g_A^{(3)}} \;\; = \;\;
489: -\frac{f_2^{\rm quark} J_1}{J} ,
490: \label{f2_grad_as_ratio}
491: \eeq
492: With the standard parameters of the instanton vacuum,
493: $\bar\rho = (600\, {\rm MeV})^{-1}$ and $M = 350\, {\rm MeV}$,
494: we find
495: \beq
496: \frac{f_2^{\rm quark} J_1}{J} \;\; =\;\; 0.49 \, \bar\rho^{-2} .
497: \eeq
498: With the experimental values $M_N = 940\, {\rm MeV}$ and
499: $g_A^{I=1} = 1.25$, Eq.(\ref{f2_grad_as_ratio}) thus
500: gives\footnote{In the numerical estimate quoted in Ref.\cite{Balla:1998hf},
501: $f_2^{(3)} = -0.11$, some contributions to the gradient expansion of the
502: matrix element were missed. Hence the differences in the numerical values.}
503: \beq
504: f_2^{(3)} \;\; = \;\; -0.25 \; .
505: \label{f2_3_num}
506: \eeq
507: \par
508: The integral $J_1$ in the gradient expansion of $f_2^{(3)}$, (\ref{I_f_2}),
509: contains a would--be quadratic divergence, {\it i.e.}, it is parametrically
510: of the order $J_1 \sim \bar\rho^{-2}$, contrary to the integral $J$
511: in $g_A^{(3)}$, which is depends only logarithmically
512: on $\bar\rho^{-1}$. Thus, in $J_1$ the dominant contribution comes
513: from quark momenta of the order of the UV cutoff, $\bar k \sim \bar\rho^{-1}$.
514: In principle the representation of the instanton--induced effective
515: operator by the local operator, (\ref{f2_equiv}), is accurate only
516: for quark momenta of the order $\bar k \sim M \ll \bar\rho^{-1}$;
517: for momenta of the order $\bar\rho^{-1}$ one should take into account the
518: momentum--dependence (non-locality) of the instanton--induced
519: effective operator. One should keep in mind that, anyway, the precise form
520: of the UV cutoff of the effective low--energy theory depends on
521: the details of the approximations made in the description of the instanton
522: medium; only the gross features can be assumed to be generic. Thus,
523: results for quantities given by ``quadratically divergent'' integrals, such
524: as $f_2^{(3)}$, should generally be regarded as rough estimates ($\pm 50 \%$).
525: Whether or not to include the form factors in the effective operator is just
526: part of this larger uncertainty. We have verified that using instead of
527: (\ref{f2_equiv})
528: the ``exact'' non-local operator $-\bar\psi \gamma_\beta \gamma_5
529: \left[ I_1 (\partial^2 ) / F(\partial^2 ) \right] \partial^2 \psi$
530: changes the numerical result for $f_2^{(3)}$ by less than 10\%, so
531: using the local approximation to the operator seems completely justified.
532: \par
533: {\it Isoscalar matrix element} $f_2^{(0)}$.
534: The calculation of the $1/N_c$--suppressed isoscalar component of the
535: twist--4 matrix element, $f_2^{(0)}$, is somewhat more involved, since
536: it requires to take into account the time dependence of the
537: saddle--point pion field, (\ref{U_time_dependent}), to first order in the
538: angular velocity of the soliton, $\Omega$, (\ref{angular_vel}).
539: An important question is the degree of the would--be ultraviolet
540: divergence of the isoscalar nucleon matrix element, $f_2^{(0)}$. This question
541: can be investigated by gradient expansion. Expanding the average over
542: quark fields of the flavor--singlet version of the operator
543: (\ref{f2_equiv}),
544: \beq
545: (-i) f_2^{\rm quark}
546: N_c \, {\rm Sp} \left[ \, 1_{\rm Flavor}\; \gamma_\beta \gamma_5 \partial^2
547: \frac{1}{i\!{\not\partial} - M F(\partial^2 ) U^{\gamma_5} F(\partial^2 )}
548: \right] ,
549: \label{singlet_trace}
550: \eeq
551: in gradients of the (space-- and time--dependent) background pion field,
552: one finds that the functional trace is at most {\it logarithmically}
553: divergent. This implies that the matrix element, which is obtained
554: by substituting in (\ref{singlet_trace}) the slowly rotating pion
555: field (\ref{U_time_dependent}), and expanding to first order in the angular
556: velocity, $\Omega$, can also be at most logarithmically divergent.
557: This circumstance is very fortunate, as it allows us to neglect the
558: instanton--induced form factors in the quark--pion coupling
559: in the Lagrangian, (\ref{L_eff}), and simply apply an external UV
560: regularization in the form of a Pauli--Villars subtraction.
561: (If the matrix element were quadratically divergent, as $f_2^{(3)}$ is,
562: the result would have been strongly dependent on the specific form of
563: ultraviolet cutoff applied.) With this simplification the calculation of
564: the matrix element of the instanton--induced effective operator
565: (\ref{f2_equiv}) becomes completely analogous to that of the isovector
566: axial coupling constant, $g_A^{(0)}$ \cite{Christov:1994ny,Blotz:1996wi},
567: and can be performed using the same techniques.
568: \par
569: A new feature in the calculation of the matrix element
570: of the instanton--induced operator (\ref{f2_equiv}) compared to
571: that of the axial current, $\bar\psi\gamma_\beta\gamma_5\psi$,
572: is that with (\ref{f2_equiv}) contributions of order $\Omega^1$ can
573: appear from the action of the derivatives contained in the operator
574: on the time--dependent isospin rotation matrices, $R(t)$. Such contributions
575: were discussed in Ref.\cite{Pobylitsa:1999tk} in the context of the
576: calculation of the isovector unpolarized quark distribution in the
577: nucleon, and we refer to this article (in particular Section 3.1) for a
578: detailed description of the $\Omega$-- ($1/N_c$--) expansion.
579: After performing the $\Omega$--expansion one expresses the
580: remaining Green functions in the static background pion field,
581: (\ref{hedge}), in a basis of eigenfunctions of the quark single--particle
582: Hamiltonian in the:
583: \be
584: H &=& -i \gamma^0 \gamma^i \nabla_i + \gamma^0 M U^{\gamma_5} ,
585: \\
586: H |n\rangle &=& E_n |n\rangle .
587: \ee
588: The result for the isoscalar matrix element then reads
589: \be
590: f_2^{(0)} &=& \frac{N_c \, f_2^{\rm quark}}{M_N^2}
591: \left\{ -\frac{1}{4I}
592: \sum_{\scriptstyle n \atop \scriptstyle {\rm non-occup.}}
593: \sum_{\scriptstyle m \atop \scriptstyle {\rm occup.}}
594: \frac{1}{E_n-E_m}
595: \right.
596: \nonumber \\
597: && \phantom{\frac{N_c \, f_2^{\rm quark}}{M_N^2}}
598: \;\;\;\;\; \times \left[
599: \langle n|\sigma_3(E_m^2- {\bf p}^2 )|m\rangle\langle m|\tau_3 |n \rangle
600: +\langle n|\tau_3|m\rangle \langle m|\sigma_3(E_n^2- {\bf p}^2 )|n\rangle
601: \right]
602: \nonumber \\
603: && \phantom{\frac{N_c \, f_2^{\rm quark}}{M_N^2}}
604: \left.
605: -\frac{1}{2 I}
606: \sum_{\scriptstyle n \atop \scriptstyle {\rm occup.}}
607: E_n \langle n|\sigma_3\tau_3|n\rangle \right\} .
608: \label{sum}
609: \ee
610: Here $I$ is the moment of inertia of the classical soliton,
611: see Ref.\cite{Diakonov:1988ty}.
612: The first term in the braces is a double sum over quark levels; it
613: coincides with the expression for $g_A^{(0)}$ up to the insertions
614: of the operator $E^2 - {\bf p}^2$ in the single--particle matrix elements,
615: which come from the contracted derivative in the operator (\ref{f2_equiv}).
616: The second term represents the abovementioned ``new'' contribution
617: which results from the action of the derivative in the operator on the
618: rotational matrices in the collective quantization. Up to the factor
619: $E_n$ inside the single sum over quark levels this contribution is
620: identical to the expression for the {\it isovector} axial coupling
621: constant, $g_A^{(3)}$.
622: \par
623: In Eq.(\ref{sum}) the first term (the double sum over quark levels)
624: turns out to be UV--finite and does not require regularization.
625: The second term (the simple sum over levels, related to the isovector
626: axial coupling constant), as it stands, contains a linear divergence
627: when regularized {\it e.g.}\ with an energy cutoff. At the same time
628: one observes an anomaly--type phenomenon in the sense that the sum over
629: occupied quark levels is not equal to minus the sum over non-occupied ones,
630: as it should be on grounds of the analyticity properties of the
631: single--particle Green functions. However, both the linear divergence
632: and the ``anomaly'' disappear when performing a Pauli--Villars (PV)
633: subtraction, leaving behind a finite sum with usual behavior
634: when summing over non-occupied instead of occupied states, in agreement
635: with the above statement that the matrix element is at most logarithmically
636: divergent. A similar phenomenon was encountered in the calculation of
637: matrix elements of twist--2 operators of spin $>2$ in the chiral
638: quark--soliton model in Ref.\cite{Diakonov:1997vc}.
639: In that case, too, the superficial
640: power divergences and the ``anomalies'' simultaneously disappeared
641: after PV subtraction. Note that the PV subtraction, contrary to
642: regularization methods based on an energy cutoff, does not spoil
643: the analyticity properties of the regularized sums.
644: \par
645: The remaining PV regularized sums can be performed numerically,
646: using the Kahana--Ripka method of diagonalizing the single--particle
647: Hamiltonian in a basis of free quark states.
648: The value of the PV cutoff
649: is determined by fitting the pion decay constant \cite{Diakonov:1997vc},
650: $M_{PV} = 557\, {\rm MeV}$. For the numerical estimate we use
651: the variational soliton profile of
652: Refs.\cite{Diakonov:1988ty,Diakonov:1997vc}.
653: We obtain a numerical value of the flavor--singlet twist--4
654: matrix element of
655: \beq
656: f_2^{(0)} \;\; = \;\; 0.01 .
657: \label{f2_0_num}
658: \eeq
659: The flavor--singlet matrix element is more than an order of magnitude
660: smaller than the flavor--nonsinglet one, (\ref{f2_3_num}), which reflects
661: the fact that the former is given by a ``logarithmically divergent'', the
662: latter by a ``quadratically divergent'' expression.
663: \par
664: {\it Comparison with results of other approaches.} The instanton result
665: for the isovector twist--4 matrix element, $f_2^{(3)}$, agrees
666: both in sign and in magnitude with the QCD sum
667: rule results of Refs.\cite{Balitsky:1990jb,Stein:1995si}.
668: However, we disagree with these authors on the sign of the isoscalar
669: matrix element (for a critical discussion of the QCD sum rule
670: calculations, see Refs.\cite{Ioffe:1997ey}).
671: We also disagree in the sign of the isovector matrix element
672: with the bag model estimate of Ref.\cite{Ji:1994sv};
673: however, this model can hardly claim to give a realistic description
674: of the quark--gluon correlations giving rise to the twist--4 matrix
675: elements; {\it e.g}, it does not respect the QCD equations of motion.
676: \par
677: {\it Generalization to the $SU(3)$ flavor group.}
678: So far we have calculated the twist--4 matrix elements assuming
679: the $SU(2)$ flavor group ($u, d$--quarks only). The results
680: can easily be generalized to the $SU(3)$ case (including also $s$ quarks).
681: Here we consider the simplest scenario of perfect $SU(3)$ symmetry,
682: {\it i.e.}, zero strange quark mass.
683: In the chiral quark--soliton model the leading--$N_c$ calculation
684: gives a simple relation between the $SU(3)$ triplet and
685: octet couplings, as the differences between them appear only due
686: to the different states of collective rotations of the soliton.
687: The ratio of the triplet and the octet part of the nucelon matrix element
688: $f_2$ is therefore given by the ratio of the matrix elements of
689: the respective Wigner $D$--functions in the rotational state corresponding
690: to the nucleon spin, isospin and the hypercharge,
691: $|\mbox{rot}\rangle = |S=T=1/2, S_3 = 1/2, T_3 = 1/2 \rangle$
692: (for details see \cite{Christov:1996vm}):
693: \begin{eqnarray}
694: \frac{f_2^{(8)}}{f_2^{(3)}}&=&
695: \frac{\langle\, \mbox{rot} \,
696: |\,D^8_{83}\,|\,\mbox{rot}\,\rangle}{\langle
697: \,\mbox{rot}|\,D^0_{33}\,|\,\mbox{rot}\rangle}
698: \;\; = \;\; \frac{3}{7}.
699: \end{eqnarray}
700: From these relations we obtain the following numerical results
701: for the twist--4 matrix elements in the $SU(3)$ symmetric case:
702: \be
703: f_2^{(0)} |_{SU(3)} &=& \phantom{-} 0.01
704: \label{f20_su3} \\
705: f_2^{(3)} |_{SU(3)} &=& -0.25
706: \label{f23_su3} \\
707: f_2^{(8)} |_{SU(3)} &=& -0.11.
708: \label{f28_su3}
709: \ee
710: %
711: \section{Power corrections to proton and neutron spin structure functions}
712: %
713: With the results obtained in the previous section we can now
714: estimate the $1/Q^2$--corrections to the first moments of the
715: proton and neutron spin structure functions, $G_1^{\rm p, n}$.
716: The expressions obtained from the operator product expansion
717: of QCD at tree level to order $1/Q^2$
718: are \cite{Shuryak:1982pi} \footnote{Note that there
719: is a mistake in Ref.\cite{Shuryak:1982pi} concerning the coefficients in
720: this formula, as was noted in \cite{Ji:1994sv}; see also
721: Ref.\cite{Ehrnsperger:1994hh}.}
722: \beq
723: \int_0^1 dx \, G_1^{p(n)} (x, Q^2 )
724: =\frac{1}{2}a_0^{p(n)}+\frac{M_N^2}{9Q^2}
725: \left(a_2^{p(n)}+4d_2^{p(n)}+4f_2^{p(n)}\right) .
726: \label{g1_pn}
727: \eeq
728: Inclusion of radiative corrections would lead to a logarithmic
729: $Q^2$--dependence of the coefficients. Here $a^{(0)}_n$
730: denote the spin--dependent matrix elements of the twist--2
731: spin--$(n + 1)$ axial vector operator; the proton and neutron
732: matrix elements (which include the quark charges) are
733: given in terms of the $SU(3)$ singlet, triplet and octet
734: components as
735: \begin{equation}
736: a_0^{p(n)} = \pm\frac{1}{6}a^{(3)}+\frac{1}{18}a^{(8)}+\frac{2}{9}a^{(0)},
737: \label{a0_pn}
738: \end{equation}
739: with $+(-)$ for proton (neutron). We quote the expressions
740: for three light quark flavors ($u, d, s$).
741: For $n = 0$ the $a_n$ coincide with the
742: axial coupling constants of the nucleon:
743: \beq
744: a_0^{(3)} \;\; = \;\; g_A^{(3)}, \hspace{4em}
745: a_0^{(8)} \;\; = \;\; g_A^{(8)}, \hspace{4em}
746: a_0^{(0)} \;\; = \;\; g_A^{(0)}.
747: \eeq
748: Of the power corrections the term proportional to $a_2^{p(n)}$
749: are the target mass corrections, while the terms proportional
750: to $d_2^{p(n)}$ and $f_2^{p(n)}$ represent the dynamical
751: higher--twist corrections; the proton and neutron matrix elements
752: are defined in analogy to Eq.(\ref{a0_pn}).
753: \par
754: We now evaluate the dynamical power corrections using the
755: instanton vacuum results.
756: Following the basic philosophy of the instanton vacuum we put the
757: parametrically suppressed twist--3 spin--2 matrix
758: elements to zero (these matrix elements were estimated in
759: Ref.\cite{Balla:1998hf}
760: to be of the order of $\sim 1 \%$ of the twist--4 ones).
761: For the twist--4 matrix elements $f_2^{p(n)}$ we use the
762: results (\ref{f20_su3})--(\ref{f28_su3}), which imply
763: \beq
764: f_2^p \;\; = \;\; -0.046 ,
765: \hspace{4em}
766: f_2^n \;\; = \;\; 0.038 .
767: \eeq
768: To evaluate the twist 2 contribution we use the GRSV 2000
769: parameterization \cite{Gluck:2000dy} (``standard scenario''),
770: which includes the radiative corrections to the coefficient
771: functions in NLO.\footnote{To evaluate the target mass corrections
772: we use instead of the twist--2 matrix element $a_2^{p(n)}$ the third
773: moment of the twist--2 part of $G_1^{p(n)}$ calculated from the GRSV 2000
774: parametrization; the difference between the two is irrelevant in the
775: present context.}
776: The results are shown in Fig.\ref{fig_p_n}. One sees that the dynamical
777: twist--4 contribution
778: has a rather small effect on the structure functions down to
779: $Q^2$ of the order of $1\, {\rm GeV}^2$, in particular in the neutron.
780: Note that the quantitative details may change when $SU(3)$ symmetry
781: breaking effects are included in the twist--4 contribution.
782: %
783: % Figure
784: %
785: \begin{figure}[ht]
786: \begin{center}
787: \begin{tabular}{c}
788: \includegraphics[width=12cm,height=8.4cm]{prot.eps}
789: \\
790: \includegraphics[width=12cm,height=8.4cm]{neut.eps}
791: \end{tabular}
792: \end{center}
793: \caption{The $Q^2$--dependence of the first moment of the polarized structure
794: function $G_1$
795: for the proton (top) and neutron (bottom). {\it Dashed lines:}
796: Twist--2 contribution according to the GRSV 2000 NLO parameterization
797: (standard scenario) \cite{Gluck:2000dy}. {\it Dotted lines:} Sum of twist--2
798: contribution and target mass corrections. {\it Solid lines:}
799: Total result, including also the twist--4 contribution due to
800: the matrix elements $f_2$, as estimated in the instanton vacuum.
801: (In the case of the neutron the target mass corrections to the
802: first moment are very small, so we show only the pure twist--2
803: contribution.)}
804: \label{fig_p_n}
805: \end{figure}
806: %
807: \clearpage
808: %
809: \par
810: In Fig.\ref{fig_bjorken} we show the result for the first moment
811: of the difference of proton and neutron structure functions,
812: $G_1^p - G_1^n$ (the Bjorken sum). This quantity is of particular
813: interest, since the Bjorken sum rule is a rigorous prediction of QCD,
814: and the radiative corrections have been calculated to order $\alpha_s^3$
815: \cite{Larin:1991tj}. We again use the GRSV2000 NLO parametrization
816: to estimate the twist--2 part and target mass corrections.
817: The power corrections now receive contributions only from the
818: flavor--triplet twist--4 matrix element, $f_2^{(3)}$.
819: Also shown in Fig.\ref{fig_bjorken} are the experimental results
820: obtained in the analyses of the SMC \cite{Adeva:1998vv},
821: E143 \cite{Abe:1998wq}, and E155 \cite{Anthony:2000fn} experiments.
822: One observes that the relatively large twist--4 correction obtained
823: from the instanton vacuum improves the agreement of the theoretical
824: prediction with the data; however, the present experimental errors
825: are too large to allow for definite conclusions.
826: %
827: % Figure
828: %
829: \begin{figure}[ht]
830: \begin{center}
831: \includegraphics[width=12cm,height=8.4cm]{nons.eps}
832: \end{center}
833: \caption{Same as Fig.\ref{fig_p_n}, but for the difference of proton
834: and neutron structure functions (Bjorken sum rule). Also shown are the
835: experimental results quoted by the E143 collaboration
836: at $Q^2 = 2, 3$ and $5 \, {\rm GeV}^2$ \cite{Abe:1998wq}, and
837: by the SMC \cite{Adeva:1998vv} and E155 \cite{Anthony:2000fn}
838: collaborations at $Q^2 = 5 \, {\rm GeV}^2$. (The abscissae of the
839: data points at $Q^2 = 5\, {\rm GeV}^2$ have been shifted slightly
840: in order to separate them in the plot.)}
841: \label{fig_bjorken}
842: \end{figure}
843: %
844: \section{Intrinsic charm contribution to the proton spin}
845: The isoscalar component of the twist--4 matrix element, $f_2^{(0)}$,
846: (\ref{f2_def}), plays a role in the nucleon spin structure functions
847: aside from power ($1/Q^2$--) corrections, namely in determining the
848: intrinsic charm contribution to the nucleon spin.
849: Up to power corrections, the first moment of the proton
850: spin structure function for three light flavors ($u, d, s$)
851: is given by [{\it cf.}\ Eqs.(\ref{g1_pn}) and
852: (\ref{a0_pn})]
853: \beq
854: \int_0^1 dx \, G^p_1 (x, Q^2 ) \;\; = \;\; \frac{1}{12}g_A^{(3)}
855: +\frac{1}{36}g_A^{(8)} + \frac{1}{9} \; g_A^{(0)}
856: \; +\; O\left(\frac{1}{Q^2}\right) .
857: \label{proton_spin}
858: \eeq
859: When the charmed quark is included, its contribution can be expressed
860: as a modification of the $SU(3)$ flavor--singlet axial coupling constant
861: in (\ref{proton_spin}):
862: \beq
863: g_A^{(0)} \;\; = \;\; g_{A, {\rm light}}^{(0)} \; + \; 2 g_A^{(c)} ,
864: \eeq
865: where $g_{A, {\rm light}}^{(0)}$ contains the $SU(3)$--singlet
866: contribution from the light flavors only, and $g_A^{(c)}$
867: is defined as
868: \beq
869: \langle p, s |\bar c \gamma_\beta\gamma_5 c |p, s \rangle
870: \;\; = \;\; -2 g^{(c)}_A s_\beta .
871: \eeq
872: In the proton, where the charmed quarks occur in the form of virtual
873: $\bar c c$ pairs, the charm quark axial current operator can be
874: approximated by an operator involving only light quarks ($u, d, s$)
875: and the gluon fields, using the heavy--quark mass ($1/m_c$--) expansion
876: to integrate out the charm degrees of freedom.
877: Refs.\cite{Franz:1999hw,Polyakov:1999rb}, correcting earlier results of
878: Refs.\cite{Halperin:1997jq,Araki:1998yk}, quote the following
879: result for the nucleon matrix element of the charm axial current
880: to order $1/m_c^2$:
881: \beq
882: \langle p, s |\bar c \gamma_\beta \gamma_5 c |p, s \rangle
883: \;\; = \;\; \frac{\alpha_s}{12\pi m_c^2} \; \langle p, s |
884: \sum_{f = u, d, s} \bar{\psi}_f
885: \gamma_\alpha \Fdual_{\beta\alpha} \psi_f |p, s \rangle ,
886: \label{charm}
887: \eeq
888: from which follows that
889: \beq
890: g_A^{(c)} \;\; =\;\;
891: -\frac{\alpha_s}{12\pi}\frac{M_N^2}{m_c^2} f_2^{(0)}.
892: \label{axial_charm}
893: \eeq
894: Using our result for the $SU(3)$ flavor--singlet matrix element,
895: $f_2^{(0)}$, Eq.(\ref{f20_su3}), we can now estimate the
896: charm axial constant, $g_A^{(c)}$. With $\alpha_s (\mu = m_c ) = 0.39$
897: and $m_c = 1.15 \ldots 1.35 \, {\rm GeV}$, we find
898: \beq
899: g_A^{(c)} \simeq -1\times 10^{-4}.
900: \eeq
901: The charm quark contribution to the proton spin is very small,
902: mostly because of the smallness of the flavor--singlet twist--4 matrix
903: element, $f_2^{(0)}$.
904: %
905: \section{Summary}
906: %
907: The instanton vacuum with its inherent small parameter --- the packing fraction
908: of the instantons, $\bar\rho / \bar R$ --- implies a parametrical (and
909: numerical) hierarchy of the spin--dependent twist--3 and 4 matrix elements:
910: $\mbox{twist--3} \ll \mbox{twist--4}$. The unusually small value for the
911: twist--3 matrix element $d^{(2)} \sim 10^{-3}$ obtained from the
912: instanton vacuum \cite{Balla:1998hf} agrees well with the precise
913: measurements of $G_2$ in the E155x \cite{Bosted:2000tf} experiment
914: (and, also, the older E143 \cite{Abe:1998wq} and E155 \cite{Anthony:2000fn}
915: results). Note also that recently revised lattice calculations of
916: $d_2$ \cite{Gockeler:2001ja}, properly accounting for operator mixing
917: effects, confirmed the small value predicted by the instanton vacuum.
918: \par
919: In this paper we have presented numerical estimates of the
920: twist--4 matrix elements $f_2$, including also its flavor--singlet
921: component. Our approach predicts a big numerical difference between the
922: flavor--nonsinglet and singlet matrix elements, $|f_2^{(3)}| \gg |f_2^{(0)}|$,
923: in contrast to QCD sum rules, which tend to give values of the same
924: order of magnitude. Phenomenologically, this would imply much larger power
925: corrections to the Bjorken than to the Ellis--Jaffe sum rule,
926: a prediction which can in principle be tested experimentally.
927: We have also pointed out that the small value of $f_2^{(0)}$ implies
928: a very small intrinsic charm contribution to the nucleon spin.
929: \par
930: The approach laid out here can be extended in many
931: ways; for example, one can compute also the matrix elements determining
932: power corrections to higher moments of the structure functions and
933: restore the the $x$--dependence of the twist--4 contribution.
934: This would allow a more direct comparison of the higher--twist
935: corrections with experimental data.
936: %
937: \begin{thebibliography}{99}
938: %
939: %
940: \bibitem{Shuryak:1982pi}
941: %
942: E.~V.~Shuryak and A.~I.~Vainshtein,
943: Nucl.\ Phys.\ {\bf B201} (1982) 141.
944: \href{\wwwspires?j=NUPHA\%2cB201\%2c141}{\myhrefstring}
945: %
946: %
947: \bibitem{Ji:1994sv}
948: X.~Ji and P.~Unrau,
949: Phys.\ Lett.\ {\bf B333} (1994) 228
950: [hep-ph/9308263].
951: \href{\wwwspires?eprint=HEP-PH/9308263}{\myhrefstring}
952: %
953: %
954: \bibitem{Ehrnsperger:1994hh}
955: B.~Ehrnsperger, A.~Schafer and L.~Mankiewicz,
956: Phys.\ Lett.\ {\bf B323} (1994) 439
957: [hep-ph/9311285].
958: \href{\wwwspires?eprint=HEP-PH/9311285}{\myhrefstring}
959: %
960: %
961: \bibitem{Bosted:2000tf}
962: P.~Bosted [E155x Collaboration], Talk given at the Symposium on the
963: Gerasimov-Drell-Hearn Sum Rule and the Nucleon Spin Structure in the
964: Resonance Region (GDH 2000), Mainz, Germany, 14--17 Jun 2000.
965: %
966: %
967: \bibitem{Edelmann:2000yp}
968: J.~Edelmann, G.~Piller, N.~Kaiser and W.~Weise,
969: Nucl.\ Phys.\ {\bf A665} (2000) 125
970: [hep-ph/9909524].
971: \href{\wwwspires?eprint=HEP-PH/9909524}{\myhrefstring}
972: %
973: %
974: \bibitem{Drechsel:1995az}
975: For a review, see: D.~Drechsel,
976: Prog.\ Part.\ Nucl.\ Phys.\ {\bf 34} (1995) 181
977: [nucl-th/9411034].
978: \href{\wwwspires?eprint=NUCL-TH/9411034}{\myhrefstring}
979: %
980: %
981: \bibitem{Polyakov:1999rb}
982: M.~V.~Polyakov, A.~Schafer and O.~V.~Teryaev,
983: Phys.\ Rev.\ {\bf D60}, 051502 (1999)
984: [hep-ph/9812393].
985: \href{\wwwspires?eprint=HEP-PH/9812393}{\myhrefstring}
986: %
987: %
988: \bibitem{Franz:1999hw}
989: M.~Franz, P.~V.~Pobylitsa, M.~V.~Polyakov and K.~Goeke,
990: Phys.\ Lett.\ {\bf B454} (1999) 335
991: [hep-ph/9810343].
992: \href{\wwwspires?eprint=HEP-PH/9810343}{\myhrefstring}
993: %
994: %
995: \bibitem{Halperin:1997jq}
996: I.~Halperin and A.~Zhitnitsky,
997: hep-ph/9706251.
998: \href{\wwwspires?eprint=HEP-PH/9706251}{\myhrefstring}
999: %
1000: %
1001: \bibitem{Araki:1998yk}
1002: F.~Araki, M.~Musakhanov and H.~Toki,
1003: [hep-ph/9808290].
1004: \href{\wwwspires?eprint=HEP-PH/9808290}{\myhrefstring}
1005: %
1006: %
1007: \bibitem{Baum:1996yv} The COMPASS Collaboration (G. Baum {\em et al.}),
1008: {\it ``COMPASS: A proposal for a common muon and proton
1009: apparatus for structure and spectroscopy''}, CERN-SPSLC-96-14 (1996).
1010: %
1011: %
1012: \bibitem{Shuryak:1982ff}
1013: E.~V.~Shuryak, Nucl.\ Phys.\ {\bf B203} (1982) 93;
1014: \href{\wwwspires?j=NUPHA\%2cB203\%2c93}{\myhrefstring}
1015: {\it ibid.}\ {\bf B203} (1982) 116.
1016: \href{\wwwspires?j=NUPHA\%2cB203\%2c116}{\myhrefstring}
1017: %
1018: %
1019: \bibitem{Diakonov:1984hh}
1020: D.~I.~Diakonov and V.~Y.~Petrov,
1021: Nucl.\ Phys.\ {\bf B245} (1984) 259.
1022: \href{\wwwspires?j=NUPHA\%2cB245\%2c259}{\myhrefstring}
1023: %
1024: %
1025: \bibitem{Schafer:1998wv}
1026: T.~Schafer and E.~V.~Shuryak,
1027: Rev.\ Mod.\ Phys.\ {\bf 70} (1998) 323
1028: [hep-ph/9610451].
1029: \href{\wwwspires?eprint=HEP-PH/9610451}{\myhrefstring}
1030: %
1031: %
1032: \bibitem{Diakonov:1986eg}
1033: D.~I.~Diakonov and V.~Y.~Petrov,
1034: Nucl.\ Phys.\ {\bf B272} (1986) 457.
1035: \href{\wwwspires?j=NUPHA\%2cB272\%2c457}{\myhrefstring}
1036: %
1037: %
1038: \bibitem{Diakonov:1988ty}
1039: D.~Diakonov, V.~Y.~Petrov and P.~V.~Pobylitsa,
1040: Nucl.\ Phys.\ {\bf B306}, 809 (1988).
1041: \href{\wwwspires?j=NUPHA\%2cB306\%2c809}{\myhrefstring}
1042: %
1043: %
1044: \bibitem{Christov:1996vm}
1045: C.~V.~Christov {\it et al.},
1046: Prog.\ Part.\ Nucl.\ Phys.\ {\bf 37}, 91 (1996)
1047: [hep-ph/9604441].
1048: \href{\wwwspires?eprint=HEP-PH/9604441}{\myhrefstring}
1049: %
1050: %
1051: \bibitem{Balla:1998hf}
1052: J.~Balla, M.~V.~Polyakov and C.~Weiss,
1053: Nucl.\ Phys.\ {\bf B510} (1998) 327
1054: [hep-ph/9707515].
1055: \href{\wwwspires?eprint=HEP-PH/9707515}{\myhrefstring}
1056: %
1057: %
1058: \bibitem{Diakonov:1996qy}
1059: D.~I.~Diakonov, M.~V.~Polyakov and C.~Weiss,
1060: Nucl.\ Phys.\ {\bf B461} (1996) 539
1061: [hep-ph/9510232].
1062: \href{\wwwspires?eprint=HEP-PH/9510232}{\myhrefstring}
1063: %
1064: %
1065: \bibitem{Diakonov:1986aj}
1066: D.~I.~Diakonov and V.~Y.~Petrov,
1067: LENINGRAD-86-1153, published (in Russian) in: Hadron Matter under Extreme
1068: Conditions, Eds.\ G.~M.~Zinovev and V.~P.~Shelest, Naukova Dumka,
1069: Kiev (1986), p. 192.
1070: %
1071: %
1072: \bibitem{Dressler:1999zi}
1073: B.~Dressler, M.~Maul and C.~Weiss,
1074: Nucl. Phys. {\bf B 578}, 293 (2000)
1075: [hep-ph/9906444].
1076: \href{\wwwspires?eprint=HEP-PH/9906444}{\myhrefstring}
1077: %
1078: %
1079: \bibitem{Christov:1994ny}
1080: C.~V.~Christov, A.~Blotz, K.~Goeke, P.~Pobylitsa, V.~Petrov,
1081: M.~Wakamatsu and T.~Watabe,
1082: Phys.\ Lett.\ {\bf B325} (1994) 467
1083: [hep-ph/9312279].
1084: \href{\wwwspires?eprint=HEP-PH/9312279}{\myhrefstring}
1085: %
1086: %
1087: \bibitem{Blotz:1996wi}
1088: A.~Blotz, M.~Praszalowicz and K.~Goeke,
1089: Phys.\ Rev.\ {\bf D53} (1996) 485
1090: [hep-ph/9403314].
1091: \href{\wwwspires?eprint=HEP-PH/9403314}{\myhrefstring}
1092: %
1093: %
1094: \bibitem{Pobylitsa:1999tk}
1095: P.~V.~Pobylitsa, M.~V.~Polyakov, K.~Goeke, T.~Watabe and C.~Weiss,
1096: Phys.\ Rev.\ {\bf D59} (1999) 034024
1097: [hep-ph/9804436].
1098: \href{\wwwspires?eprint=HEP-PH/9804436}{\myhrefstring}
1099: %
1100: %
1101: \bibitem{Diakonov:1997vc}
1102: D.~I.~Diakonov, V.~Y.~Petrov, P.~V.~Pobylitsa, M.~V.~Polyakov and C.~Weiss,
1103: Phys.\ Rev.\ {\bf D56} (1997) 4069
1104: [hep-ph/9703420].
1105: \href{\wwwspires?eprint=HEP-PH/9703420}{\myhrefstring}
1106: %
1107: %
1108: \bibitem{Balitsky:1990jb}
1109: I.~I.~Balitsky, V.~M.~Braun and A.~V.~Kolesnichenko,
1110: Phys.\ Lett.\ {\bf B242} (1990) 245
1111: [hep-ph/9310316].
1112: \href{\wwwspires?eprint=HEP-PH/9310316}{\myhrefstring};
1113: Erratum Phys.\ Lett.\ {\bf B318} (1993) 648.
1114: %
1115: %
1116: \bibitem{Stein:1995si}
1117: E.~Stein, P.~Gornicki, L.~Mankiewicz and A.~Schafer,
1118: Phys.\ Lett.\ {\bf B353} (1995) 107
1119: [hep-ph/9502323].
1120: \href{\wwwspires?eprint=HEP-PH/9502323}{\myhrefstring}
1121: %
1122: %
1123: \bibitem{Ioffe:1997ey}
1124: B.~L.~Ioffe,
1125: Phys.\ Atom.\ Nucl.\ {\bf 60} (1997) 1707
1126: [hep-ph/9704295];
1127: \href{\wwwspires?eprint=HEP-PH/9704295}{\myhrefstring}
1128: Lecture at the St.Peters\-burg Winter School on Theoretical Physics, Febr.\
1129: 23--28, 1998, hep-ph/9804238.
1130: \href{\wwwspires?eprint=HEP-PH/9804238}{\myhrefstring}
1131: %
1132: %
1133: \bibitem{Gluck:2000dy}
1134: M.~Gluck, E.~Reya, M.~Stratmann and W.~Vogelsang,
1135: [hep-ph/0011215].
1136: \href{\wwwspires?eprint=HEP-PH/0011215}{\myhrefstring}
1137: %
1138: %
1139: \bibitem{Larin:1991tj}
1140: S.~A.~Larin and J.~A.~Vermaseren,
1141: Phys.\ Lett.\ B {\bf 259} (1991) 345.
1142: %
1143: %
1144: \bibitem{Adeva:1998vv}
1145: B.~Adeva {\it et al.} [Spin Muon Collaboration],
1146: Phys.\ Rev.\ D {\bf 58} (1998) 112001.
1147: %
1148: %
1149: \bibitem{Abe:1998wq}
1150: K.~Abe {\it et al.} [E143 collaboration],
1151: Phys.\ Rev.\ D {\bf 58} (1998) 112003
1152: [hep-ph/9802357].
1153: \href{\wwwspires?eprint=HEP-PH/9802357}{\myhrefstring}
1154: %
1155: %
1156: \cite{Anthony:2000fn}
1157: \bibitem{Anthony:2000fn}
1158: P.~L.~Anthony {\it et al.} [E155 Collaboration],
1159: Phys.\ Lett.\ B {\bf 493} (2000) 19
1160: [hep-ph/0007248].
1161: \href{\wwwspires?eprint=HEP-PH/0007248}{\myhrefstring}
1162: %
1163: %
1164: \bibitem{Gockeler:2001ja}
1165: M.~Gockeler {\it et al.},
1166: Phys.\ Rev.\ D {\bf 63} (2001) 074506
1167: [hep-lat/0011091].
1168: \href{\wwwspires?eprint=HEP-LAT/0011091}{\myhrefstring}
1169: %
1170: \end{thebibliography}
1171: %
1172: %
1173: \end{document}
1174: %
1175: % END DOCUMENT
1176: %
1177:
1178:
1179:
1180:
1181: