1:
2: \documentclass{JHEP3}
3: \usepackage{amsmath}
4:
5: \title{Calculating the gauge-invariant two-point correlation function of gluonic field strengths}
6:
7: \author{Dmitri Antonov
8: \thanks{Permanent address:
9: ITEP, B. Cheremushkinskaya 25, RU-117 218 Moscow, Russia.}\\
10: Institute of Physics, Humboldt University of Berlin,\\
11: Newtonstr. 15, 12489 Berlin, Germany \\
12: E-mail: \email{antonov@physik.hu-berlin.de}}
13:
14:
15:
16:
17:
18: \abstract{The two-point gauge-invariant correlation function of gluonic field strengths,
19: which is the main input in the stochastic vacuum model, is derived by using its relation to the
20: Green functions of one- and two-gluon gluelumps. These Green functions are found by evaluating the
21: path integrals of one and two gluons confined in the field of an infinitely heavy adjoint source. This
22: allows one to obtain also the
23: terms in the exponential asymptotics of the two-point correlation function, which are produced by fluctuations
24: of the gluons around their mean separation from the source.
25: The pre-exponents are derived as well, demonstrating some novel dependences on the distance.}
26:
27:
28:
29: \keywords{Nonperturbative Effects; Confinement; Phenomenological Models; QCD; Lattice Gauge Field Theories}
30:
31: \preprint{HU-EP-05/70}
32:
33: \dedicated{}
34:
35: \begin{document}
36:
37:
38: \section{Introduction}
39: Despite a remarkable impact of the stochastic vacuum model~\cite{cdm, svm} on QCD phenomenology (see e.g.~\cite{book} for a recent
40: review), a field-theoretic establishment of this model is still pending. The main problem is an analytic calculation
41: of the nonperturbative content of the two-point gauge-invariant correlation function of gluonic field strengths, which can be
42: parametrized as~\cite{svm}
43: $$\frac{g^2}{N_c}{\rm tr}{\,}\left<F_{\mu\nu}(x)\Phi_{xx'}F_{\lambda\rho}(x')\Phi_{x'x}\right>_{\rm YM}=
44: (\delta_{\mu\lambda}\delta_{\nu\rho}-\delta_{\mu\rho}\delta_{\nu\lambda})D(x-x')+$$
45: \begin{equation}
46: \label{cf}
47: +\frac12\left[\partial_\mu^x\left((x-x')_\lambda\delta_{\nu\rho}-(x-x')_\rho\delta_{\nu\lambda}\right)+
48: \partial_\nu^x\left((x-x')_\rho\delta_{\mu\lambda}-(x-x')_\lambda\delta_{\mu\rho}\right)\right]D_1(x-x').
49: \end{equation}
50: Here, $\Phi_{xx'}\equiv {\cal P}{\,}\exp\left(ig\int_{x'}^{x}dz_\mu A_\mu^a(z)T_{\rm fund}^a\right)$ is a phase
51: factor along the straight line, which goes through $x'$ and $x$; $T_{\rm fund}^a$'s
52: stand for the generators of the group SU($N_c$) in the fundamental representation; the average $\left<\ldots\right>_{\rm YM}$
53: is taken with respect to the Euclidean Yang-Mills action, $\frac14\int d^4x(F_{\mu\nu}^a)^2$;
54: $F_{\mu\nu}^a=\partial_\mu A_\nu^a-\partial_\nu A_\mu^a+gf^{abc}A_\mu^b A_\nu^c$, $a=1, \ldots, N_c^2-1$.
55:
56: Numerous lattice measurements of the correlation function~(\ref{cf}) were performing over the years (see~\cite{cdm, d} for the calculations
57: with the cooling method and ref.~\cite{nga}, where this correlation function was reconstructed from the lattice data on the
58: static quark-antiquark potential). As for
59: the papers attempting to analytically evaluate this quantity, some of those were dealing with the
60: Abelian theories, where confinement was produced by monopoles~\cite{early, later}. At the same time, some progress was achieving in the
61: non-Abelian case as well~\cite{dm, na}. In particular, a relation between the correlation length of the vacuum (that is the
62: length at which the functions $D$ and $D_1$ experience an exponential fall-off)
63: and the gluonic condensate was derived in~\cite{dm}. Further analysis of the correlation lengths of $D$ and $D_1$
64: was done in~\cite{na}.~\footnote{It was supporting the conjecture, tested on the lattice in~\cite{nga}, that these lengths might be different.}
65: Recently, a complete derivation of these functions was presented in~\cite{S}. Using initially
66: the idea~\cite{dm} on a relation between the correlation function~(\ref{cf}) and the Green functions
67: of one and two (valence) gluons confined in the field of an infinitely heavy adjoint source (the so-called one- and two-gluon gluelumps),
68: the paper~\cite{S} further accommodated the Hamiltonian formalism for the analysis
69: of these Green functions. This eventually yielded certain nonperturbative expressions for the functions $D$ and $D_1$.
70:
71: In the present paper, the Green functions (as well as the functions $D$ and $D_1$) will be calculated upon an evaluation
72: of the corresponding integrals over trajectories of the gluons in the gluelumps. It turns out that, when the minimal-area-law
73: Ansatz is applied to the Wilson loops of these (confined) gluons, further calculation of the path integrals
74: becomes possible by making use of the Feynman variational method. By means of this method, we manage to find
75: some novel terms in both the exponents and pre-exponents of the functions $D$ and $D_1$, which stem from
76: the vibrations of the gluons around their mean separations from the heavy source.
77:
78: The paper is organized as follows. In the next section, we demonstrate our method at the example of a one-gluon gluelump and
79: calculate in this way the function $D_1$. In section~3, we proceed with the two-gluon gluelump and the function $D$.
80: In section~4, a brief summary of the obtained results is presented.
81:
82:
83:
84:
85:
86: \section{Green function of a one-gluon gluelump and the function $D_1$}
87: In the approximation when the gluon spin term is disregarded~\cite{S},
88: the one-gluon gluelump Green function has the form $\delta_{\mu\nu}G(T,0)$, where
89: $$G(T,0)=\int\limits_{0}^{\infty}ds\int\limits_{x_4(0)=0,{\,}x_4(s)=T}^{}{\cal D}x_\mu
90: \exp\left(-\int_{0}^{s}d\tau\frac{\dot x_\mu^2}{4}\right)\left<W(C)\right>_{\rm YM}.$$
91: Here, in the Wilson loop $\left<W(C)\right>_{\rm YM}={\rm tr}\left<{\cal P}{\,}\exp\left(ig\oint_{C}^{}dx_\mu A_\mu^aT_{\rm adj}^a\right)\right>_{\rm YM}$,
92: $T_{\rm adj}^a$'s stand for the SU($N_c$)-generators in the adjoint representation, $(T_{\rm adj}^a)^{bc}=-if^{abc}$; the closed contour $C$
93: is formed by the piece of the 4-th axis from the point $({\bf 0}, T)$ to the origin and the trajectory $x_\mu(\tau)$ of a gluon, which
94: moves back from the origin to that point. Adopting for $\left<W(C)\right>_{\rm YM}$ the minimal-area-law Ansatz, we have
95: $$
96: G(T,0)=N_c(N_c^2-1)\int\limits_{0}^{\infty}ds\int\limits_{x_4(0)=0,{\,}x_4(s)=T}^{}{\cal D}x_4\int{\cal D}{\bf x}\times$$
97: $$\times\exp\left[-\int_{0}^{s}d\tau\left(\frac{\dot{\bf x}^2}{4}+\frac{\dot x_4^2}{4}+\sigma_{\rm adj}|{\bf x}|\dot x_4
98: \right)\right].$$
99: It will further be assumed that, in the directions transverse to $x_4$, gluon, along most of its path, only slightly fluctuates
100: around some mean distance, $L$, from the 4-th axis.~\footnote{This distance will be found numerically at the end of the calculation.}
101: This means the following parametrization ${\bf x}(\tau)=
102: {\bf L}+{\bf l}(\tau)$, where ${\bf l}(0)={\bf l}(s)=0$, ${\bf L}=L{\bf n}$. Then, up to the term
103: linear in ${\bf l}$, $|{\bf x}|\simeq L+{\bf n}{\bf l}$. In this approximation, we have
104: $$
105: G(T,0)\simeq N_c(N_c^2-1){\rm e}^{-\sigma_{\rm adj}LT}
106: \int\limits_{0}^{\infty}ds\int\limits_{x_4(0)=0,{\,}x_4(s)=T}^{}{\cal D}x_4\int{\cal D}{\bf l}{\cal D}\mu\times$$
107: $$\times\exp\left[-\int_{0}^{s}d\tau\left(\frac{\dot{\bf l}^2}{4}+\frac{\dot x_4^2}{4}+\frac{\mu}{2}+
108: \frac{\sigma_{\rm adj}^2\dot x_4^2({\bf n}{\bf l})^2}{2\mu}\right)\right],$$
109: where we have introduced an integration over the ``einbein'' field $\mu$ in order to make the potential term
110: quadratic in ${\bf l}$. Next,
111: the unit vector ${\bf n}$ may, without loss of generality, be chosen along the 1-st axis, ${\bf n}=(1,0,0)$.
112: The integrations over $l_2(\tau)$ and $l_3(\tau)$ then become those of a free particle. As for the integration over $l_1(\tau)$,
113: since ${\bf l}$, by definition,
114: only weakly varies along the trajectory, one may replace $l_1^2(\tau)$ by its value averaged along the trajectory,
115: $l_1^2\to\frac{1}{s}\int_{0}^{s}d\tau l_1^2$. Altogether we have
116: $$
117: G(T,0)\simeq N_c(N_c^2-1){\rm e}^{-\sigma_{\rm adj}LT}
118: \int\limits_{0}^{\infty}\frac{ds}{4\pi s}
119: \int\limits_{x_4(0)=0,{\,}x_4(s)=T}^{}{\cal D}x_4\int{\cal D}l_1{\cal D}\mu\times$$
120: \begin{equation}
121: \label{G}
122: \times\exp\left[-\int_{0}^{s}d\tau\left(\frac{\dot l_1^2}{4}+\frac{\dot x_4^2}{4}+\frac{\mu}{2}+
123: U(l_1)\right)\right],~~ U(l_1)=\left(\frac{\sigma_{\rm adj}^2}{2s}\int_{0}^{s}d\tau\frac{\dot x_4^2}{\mu}\right)
124: l_1^2.
125: \end{equation}
126: Although the partition function of a harmonic oscillator, resulting now from the integration over $l_1(\tau)$, is known exactly,
127: its substitution would be of little use for a practical calculation. (Merely since the frequency of this oscillator
128: is a functional of $\dot x_4/\mu$, and the subsequent integrals over $x_4$ and $\mu$ can hardly be handled.)
129: Rather, it is much more useful to apply
130: Feynman variational method~\cite{2}, that is fully legitimate, since $l_1$ only slightly fluctuates
131: around the origin. This method yields the leading quantum correction to the classical expression
132: \begin{equation}
133: \label{clasint}
134: \frac{1}{\sqrt{4\pi s}}{\rm e}^{-sU(0)}
135: \end{equation}
136: for the integral $\int\limits_{x(0)=x(s)=0}^{}{\cal D}x(\tau)\exp\left[-\int_{0}^{s}d\tau
137: \left(\frac{\dot x^2}{4}+U(x)\right)\right]$. The correction originates just from the fluctuations
138: of the trajectory around the origin and leads to the following expression for the above-mentioned integral (see e.g.~\cite{sb} for the details of derivation):
139: $\frac{1}{\sqrt{4\pi s}}\exp\left[-\sqrt{\frac{3s}{\pi}}\int_{-\infty}^{+\infty} dxU(x){\rm e}^{-\frac{3x^2}{s}}\right]$.
140: The classical approximation~(\ref{clasint}) is recoverable from this formula in the limit $s\to 0$.
141:
142: Substituting for $U$ the expression from eq.~(\ref{G}), we then get for
143: the path integral over $l_1(\tau)$: $\frac{1}{\sqrt{4\pi s}}\exp\left(-\frac{\sigma_{\rm adj}^2s}{12}\int_{0}^{s}
144: d\tau\frac{\dot x_4^2}{\mu}\right)$. This yields
145: $$G(T,0)= N_c(N_c^2-1){\rm e}^{-\sigma_{\rm adj}LT}
146: \int\limits_{0}^{\infty}\frac{ds}{(4\pi s)^{3/2}}
147: \int\limits_{x_4(0)=0,{\,}x_4(s)=T}^{}{\cal D}x_4{\cal D}\mu\times$$
148: \begin{equation}
149: \label{intl}
150: \times\exp\left[-\int_{0}^{s}d\tau\left(\frac{\dot x_4^2}{4}+\frac{\mu}{2}+
151: \frac{\sigma_{\rm adj}^2s}{12}\frac{\dot x_4^2}{\mu}\right)\right].
152: \end{equation}
153: Further integration over $\mu$ replaces the sum $\frac{\mu}{2}+\frac{\sigma_{\rm adj}^2s}{12}\frac{\dot x_4^2}{\mu}$
154: by a single term $\sigma_{\rm adj}\sqrt{s/6}|\dot x_4|$. This is a world-line Lagrangian of a free scalar particle, which moves along the 4-th axis.
155: (The mass of this particle $m\equiv\sigma_{\rm adj}\sqrt{s/6}$ is, however, a function of $s$.)
156: Such a Lagrangian is known to be equivalent to the one usually used for the world-line calculations, $\frac{\dot x_4^2}{4}+m^2$,
157: so that the total world-line action reads $\int_{0}^{s}d\tau\frac{\dot x_4^2}{2}+m^2s$. To restore the standard coefficient
158: 1/4 at the kinetic term, one can perform the rescaling $\tau=2\tau_{\rm new}$, $s=2s_{\rm new}$. The integral over $x_4$
159: then becomes standard, and we obtain
160: $$
161: G(T,0)=N_c(N_c^2-1)\frac{{\rm e}^{-\sigma_{\rm adj}LT}}{16\sqrt{2}\pi^2}\int\limits_{0}^{\infty}\frac{ds}{s^2}
162: \exp\left(-\frac{T^2}{4s}-\frac{2\sigma_{\rm adj}^2}{3}s^2\right).
163: $$
164: To calculate the integral over $s$, let us change the integration variable to $\xi=(\sigma s)^{-1}$, that yields for this integral
165: $\sigma\int_{0}^{\infty}d\xi\exp\left(-\frac{\sigma_{\rm adj}T^2}{4}\xi-\frac{2}{3\xi^2}\right)$. The saddle-point
166: approximation to the integral $\int_{0}^{\infty}d\xi {\rm e}^{-f(\xi)}$
167: works well when the width of the corresponding Gaussian distribution,
168: which is $\frac{1}{\sqrt{f''(\xi_0)}}$ (where $\xi_0$ is the saddle point),
169: is much smaller than 1. In our case, $\frac{1}{\sqrt{f''(\xi_0)}}=
170: (32/9)^{1/3}(\sigma_{\rm adj}T^2)^{-2/3}$,
171: therefore the saddle-point approximation is legitimate when the distances under consideration are large in the sense of the inequality
172: $\sigma_{\rm adj}T^2\gg 1$. Calculation of the integral is then straightforward, and we have
173: $$
174: G(T,0)=N_c(N_c^2-1)\frac{(\sigma_{\rm adj}/T^4)^{1/3}}{3^{2/3}\cdot 2^{7/3}\cdot \pi^{3/2}}
175: \exp\left[-\sigma_{\rm adj}LT-(9/32)^{1/3}(\sigma_{\rm adj}T^2)^{2/3}\right].$$
176: The function $D_1(x)$ is related to this expression as~\cite{S} $D_1(x)=-\frac{2g^2}{N_c^2}\frac{dG(x,0)}{dx^2}$.
177: Performing the differentiation we obtain
178: $$
179: D_1(x)=\frac{C_2\alpha_s}{\sqrt{\pi}}(32/9)^{1/3}\frac{\sigma_{\rm adj}}{x^2}\left[\frac{1}{12^{1/3}}+\frac{L}{2}
180: \left(\frac{\sigma_{\rm adj}}{|x|}\right)^{1/3}+\frac{2}{3(\sigma_{\rm adj}x^2)^{2/3}}\right]\times$$
181: \begin{equation}
182: \label{D1}
183: \times\exp\left[-\sigma_{\rm adj}L|x|-(9/32)^{1/3}(\sigma_{\rm adj}x^2)^{2/3}\right]~ {\rm at}~ \sigma_{\rm adj}x^2\gg 1,
184: \end{equation}
185: where $C_2=\frac{N_c^2-1}{2N_c}$ is the quadratic Casimir operator of the fundamental representation.
186: Note that a gluon is confined only at large $N_c$, where $\sigma_{\rm adj}=2\sigma_{\rm fund}$ (while,
187: at $N_c\sim 1$, it is screened by other gluons). In the large-$N_c$ limit, $C_2\alpha_s\to\frac{\lambda}{8\pi}$, where
188: $\lambda=g^2N_c$ is the so-called 't~Hooft coupling, which stays finite in this limit.
189:
190: Equation~(\ref{D1}) can further be compared to the expression for $D_1(x)$
191: found in ref.~\cite{S} by the Hamiltonian approach, which reads
192: \begin{equation}
193: \label{D1S}
194: D_1(x)=\frac{2C_2\alpha_sM_0\sigma_{\rm adj}}{|x|}{\rm e}^{-M_0|x|}~ {\rm valid}~ {\rm at}~ M_0|x|\gg 1,
195: \end{equation}
196: where $M_0\simeq 1.5{\,}{\rm GeV}$ at $\sigma_{\rm fund}=0.18{\,}{\rm GeV^2}$.
197: One can see that the term $(9/32)^{1/3}(\sigma_{\rm adj}x^2)^{2/3}$ in eq.~(\ref{D1}), although parametrically dominating the exponent
198: in the limit $\sigma_{\rm adj}x^2\gg 1$, is numerically smaller than the leading term $\sigma_{\rm adj}L|x|$.
199: This is not surprising, since the term $(9/32)^{1/3}(\sigma_{\rm adj}x^2)^{2/3}$, originating
200: from small fluctuations of the valence gluon around its mean separation from the heavy source, can only be a correction
201: to the leading result. Indeed, equating $\sigma_{\rm adj}L$ to $M_0$, we have
202: $\frac{(9/32)^{1/3}(\sigma_{\rm adj}x^2)^{2/3}}{\sigma_{\rm adj}L|x|}=
203: \frac{\left(9\sigma_{\rm fund}^2|x|\right)^{1/3}}{2M_0}$.
204: Next, in the pure Yang-Mills theory which we consider here,
205: the stochastic vacuum model is only applicable up to the distances where the glueball production starts. This happens
206: at $|x|_{\rm max}=\frac{M_{\rm glueball}}{\sigma_{\rm fund}}$. Setting for an estimate $M_{\rm glueball}=1{\,}{\rm GeV}$,
207: we get $|x|_{\rm max}=5.56{\,}{\rm GeV}^{-1}$. (This yields $\sigma_{\rm adj}|x|_{\rm max}^2=11.11$, that still leaves
208: lot of space for the inequality $\sigma_{\rm adj}x^2\gg 1$ to hold.) Accordingly, $\frac{\left(9\sigma_{\rm fund}^2|x|\right)^{1/3}}{2M_0}\le 0.39$,
209: that demonstrates the numerical smallness of the term
210: $(9/32)^{1/3}(\sigma_{\rm adj}x^2)^{2/3}$ with respect to the term $\sigma_{\rm adj}L|x|$.
211:
212: Let us finally consider the pre-exponent of eq.~(\ref{D1}). First, the relation $\sigma_{\rm adj}L=M_0$ yields
213: $L=0.92{\,}{\rm fm}$. This value for the mean distance, by which a valence gluon
214: is separated from an infinitely heavy adjoint source, well exceeds the
215: corresponding vacuum correlation length $M_0^{-1}\simeq0.15{\,}{\rm fm}$.
216: The string in a gluelump is therefore a clearly shaped object. One can further see that, at
217: this value of $L$, the term $\frac{1}{12^{1/3}}$ in the brackets is always larger than the term
218: $\frac{L}{2}\left(\frac{\sigma_{\rm adj}}{|x|}\right)^{1/3}$.
219: Indeed, to compare the latter to the former means to compare
220: $\frac{\sigma_{\rm adj}L^3}{|x|}=\frac{26.08{\,}{\rm GeV}^{-1}}{|x|}$,
221: to 2/3 (at $|x|\gg\frac{1}{\sqrt{\sigma_{\rm adj}}}=1.67{\,}{\rm GeV}^{-1}$).
222: Clearly, $\frac{26.08{\,}{\rm GeV}^{-1}}{|x|}<\frac23$
223: up to $|x|=39.12{\,}{\rm GeV}^{-1}$, that is the distance much larger than $|x|_{\rm max}$. This proves our statement that
224: $\frac{1}{12^{1/3}}>\frac{L}{2}\left(\frac{\sigma_{\rm adj}}{|x|}\right)^{1/3}$.
225: As for the term $\frac{2}{3(\sigma_{\rm adj}x^2)^{2/3}}$, it exceeds $\frac{1}{12^{1/3}}$ only in the
226: limiting case $\sigma_{\rm adj}x^2<1.89$, that is on the border of applicability of the inequality $\sigma_{\rm adj}x^2\gg 1$.
227: Thus, we conclude that, in general, the term $\frac{1}{12^{1/3}}$ is the defining one for the pre-exponential behavior of the
228: function $D_1(x)$. This behavior is, therefore, $\sim x^{-2}$, that differs from the behavior $\sim |x|^{-1}$ of
229: eq.~(\ref{D1S}).
230:
231:
232: \section{Green function of a two-gluon gluelump and the function $D$}
233: This gluelump is formed by two gluons, which are attached by fundamental strings to an infinitely heavy adjoint source and joined
234: to each other by a third fundamental string. Since the gluons are identical, the vectors of their spatial coordinates,
235: ${\bf x}(\tau)$ and ${\bf y}(\tau)$, can be chosen in the form
236: ${\bf x}(\tau)=L'{\bf n}_1+{\bf l}_1(\tau)$, ${\bf y}(\tau)=L'{\bf n}_2+{\bf l}_2(\tau)$. Here, ${\bf n}_1=(\cos\theta, -\sin\theta, 0)$,
237: ${\bf n}_2=(\cos\theta, \sin\theta, 0)$, and $\theta$ is some constant angle, whose possible values as a function of $L'$
238: will be discussed at the end of the calculation.
239: The mean distance $L'$ from the gluons to the heavy source is, in general, different from $L$.
240: Within such notations, the area swept out by the string
241: joining the gluons is $2L'T\sin\theta$. After the extraction of the mean-area dependent factor,
242: the scalar function $G^{(2{\rm gl})}(T,0)$, through which the
243: Green function of the gluelump is expressed~\cite{S}, factorizes into the product of integrals over trajectories of each of the gluons:
244: $$
245: G^{(2{\rm gl})}(T,0)={\rm e}^{-2\sigma_{\rm fund}L'T\sin\theta}
246: \int\limits_{0}^{\infty}ds\int\limits_{0}^{\infty}ds'
247: \int\limits_{x_4(0)=0,{\,}x_4(s)=T}^{}{\cal D}x_4\int\limits_{y_4(0)=0,{\,}y_4(s)=T}^{}{\cal D}y_4
248: \int{\cal D}{\bf x}{\cal D}{\bf y}\times$$
249: \begin{equation}
250: \label{2g}
251: \times\exp\left[-\int_{0}^{s}d\tau\left(\frac{\dot{\bf x}^2}{4}+\frac{\dot x_4^2}{4}+\sigma_{\rm fund}|{\bf x}|\dot x_4\right)
252: -\int_{0}^{s'}d\tau\left(\frac{\dot{\bf y}^2}{4}+\frac{\dot y_4^2}{4}+\sigma_{\rm fund}|{\bf y}|\dot y_4\right)\right].
253: \end{equation}
254: Let us consider, for example, the integral over ${\bf x}(\tau)$. Substituting $|{\bf x}|\simeq L'+{\bf n}_1{\bf l}_1$, we have for it
255: $$
256: {\rm e}^{-\sigma_{\rm fund}L'T}
257: \int{\cal D}{\bf l}_1{\cal D}\mu
258: \exp\left[-\int_{0}^{s}d\tau\left(\frac{\dot{\bf l}_1^2}{4}+\frac{\mu}{2}+
259: \frac{\sigma_{\rm fund}^2\dot x_4^2({\bf n}_1{\bf l}_1)^2}{2\mu}\right)\right].$$
260: We can further use the same
261: approximation as in the previous section, namely substitute $({\bf n}_1{\bf l}_1)^2\to\frac{1}{s}\int_{0}^{s}
262: d\tau({\bf n}_1{\bf l}_1)^2$. Together with the (free) integration over $l_1^3(\tau)$, this yields
263: $$
264: \frac{{\rm e}^{-\sigma_{\rm fund}L'T}}{\sqrt{4\pi s}}\int{\cal D}{\cal R}{\cal D}\mu\exp\left[-\int_{0}^{s}d\tau\left(
265: \frac{\dot{\cal R}^2}{4}+\frac{\mu}{2}+U({\cal R})\right)\right],$$
266: where ${\cal R}\equiv(l_1^1,l_1^2)$ and
267: $$
268: U({\cal R})=
269: \left(\frac{\sigma_{\rm fund}^2}{2s}\int_{0}^{s}d\tau\frac{\dot x_4^2}{\mu}\right)(l_1^1\cos\theta-l_1^2\sin\theta)^2.$$
270: Using the variational method in the 2-d case (see~\cite{sb} for details), we further obtain
271: $$
272: \int{\cal D}{\cal R}\exp\left[-\int_{0}^{s}d\tau\left(
273: \frac{\dot{\cal R}^2}{4}+U({\cal R})\right)\right]\simeq
274: \frac{1}{4\pi s}\exp\left[-\frac{3}{\pi}\int d^2{\cal R}U({\cal R}){\rm e}^{-\frac{3{\cal R}^2}{s}}\right]=$$
275: $$=\frac{1}{4\pi s}\exp\left(-\frac{\sigma_{\rm fund}^2s}{12}\int_{0}^{s}d\tau\frac{\dot x_4^2}{\mu}\right).$$
276: The resulting integral over $x_4$ and $\mu$ then coincides (modulo the
277: replacement $\sigma_{\rm adj}\to\sigma_{\rm fund}$) with that of
278: eq.~(\ref{intl}), so that the total integral over $ds{\cal D}x_4{\cal D}{\bf x}$ in eq.~(\ref{2g}) reads
279: $$
280: \frac{(\sigma_{\rm fund}/T^4)^{1/3}}{3^{2/3}\cdot 2^{7/3}\cdot \pi^{3/2}}
281: \exp\left[-\sigma_{\rm fund}L'T-(9/32)^{1/3}(\sigma_{\rm fund}T^2)^{2/3}\right].$$
282: The integral over $ds'{\cal D}y_4{\cal D}{\bf y}$ apparently produces the same result.
283: Taking into account the relation~\cite{S}
284: $D(x)=\frac{g^4(N_c^2-1)}{2}G^{(2{\rm gl})}(x,0)$, we finally obtain that, at $\sigma_{\rm fund}x^2\gg 1$,
285: \begin{equation}
286: \label{Dfunc}
287: D(x)=\frac{g^4(N_c^2-1)}{3^{4/3}\cdot 2^{17/3}\cdot \pi^3}\left(\frac{\sigma_{\rm fund}}{|x|^4}\right)^{2/3}
288: \exp\left[-2\sigma_{\rm fund}L'(1+\sin\theta)|x|-
289: (9/4)^{1/3}(\sigma_{\rm fund}x^2)^{2/3}\right].
290: \end{equation}
291: Again, as in the case of the function $D_1(x)$, the term $2\sigma_{\rm fund}L'(1+\sin\theta)|x|$ in the exponent
292: is numerically larger than $(9/4)^{1/3}(\sigma_{\rm fund}x^2)^{2/3}$. Indeed, identifying
293: $2\sigma_{\rm fund}L'(1+\sin\theta)$ with $M_0^{(2{\rm gl})}=2.56{\,}{\rm GeV}$ from the formula~\cite{S}
294: \begin{equation}
295: \label{DD}
296: D(x)\simeq\frac{g^4(N_c^2-1)}{2}0.108\sigma_{\rm fund}^2{\rm e}^{-M_0^{(2{\rm gl})}|x|}~ {\rm valid}~ {\rm at}~ M_0^{(2{\rm gl})}|x|\gg 1,
297: \end{equation}
298: we have $\frac{(9/4)^{1/3}(\sigma_{\rm fund}x^2)^{2/3}}{2\sigma_{\rm fund}L'(1+\sin\theta)|x|}\le
299: \frac{\left(9\sigma_{\rm fund}^2|x|_{\rm max}/4\right)^{1/3}}{M_0^{(2{\rm gl})}}=0.12$.
300: Furthermore, the mean distance between the two gluons, as well as the mean distance
301: between these gluons and the heavy source may not exceed $|x|_{\rm max}$, i.e. the inequalities $2L'\sin\theta\le|x|_{\rm max}$, $L'\le|x|_{\rm max}$
302: should be respected together with the equation $2\sigma_{\rm fund}L'(1+\sin\theta)=M_0^{(2{\rm gl})}$. One can readily
303: see that the unphysical case $\theta=0$
304: cannot be realized also numerically. Indeed, one should then have $\frac{M_0^{(2{\rm gl})}}{\sigma_{\rm fund}}=2L'\le2|x|_{\rm max}$,
305: that is not feasible, since the l.h.s. equals $14.22{\,}{\rm GeV}^{-1}$,
306: whereas the r.h.s. equals $11.12{\,}{\rm GeV}^{-1}$. For all other cases, we have: $\theta\le\arcsin\frac{|x|_{\rm max}}{2L'}$
307: at $\frac{|x|_{\rm max}}{2}\le L'\le|x|_{\rm max}$, while no restrictions on the values of $\theta$
308: exist at $L'<\frac{|x|_{\rm max}}{2}$.~\footnote{Note that $L'$ is bounded from below only by the inequality $L'>\frac{1}{M_0^{(2{\rm gl})}}$.}
309: Finally, we note that the pre-exponent in the function~(\ref{Dfunc}) falls off
310: as $|x|^{-8/3}$, whereas the pre-exponent of the function~(\ref{DD}) is a constant.
311:
312:
313:
314: \section{Summary}
315: In the present paper, we have evaluated the functions $D(x)$ and $D_1(x)$, which parametrize the two-point gauge-invariant
316: correlation function of gluonic field strengths in the stochastic vacuum model of QCD. To this end, we have applied the Feynman
317: variational method to the calculation of the Green functions of one- and two-gluon gluelumps, through which the functions
318: $D$ and $D_1$ can be expressed. At the end of the calculation, we have compared our results with those found by the
319: Hamiltonian analysis in ref.~\cite{S}. In case of the function $D_1$, this comparison yielded the mean separation of a gluon from the heavy source
320: in the one-gluon gluelump. In case of the function $D$, it yielded some constraints on the values of the angle between the
321: planar world sheets of the gluons in the two-gluon gluelump. We have further seen that, although fluctuations of the gluons
322: (which are accounted for by the variational method) around their mean separations from the source produce the terms in the
323: exponential asymptotics of the functions $D$ and $D_1$, which are parametrically larger than the leading ones,
324: numerically these terms are nevertheless subleading, as they should be. Furthermore, we have also found the distance dependences
325: of the pre-exponential factors in the both functions $D$ and $D_1$, which turn out to be different from those of ref.~\cite{S}.
326: These differences are, however, of a rather academic interest, since numerically a distance dependence in the pre-exponent
327: plays an unimportant role at large distances considered both here and in ref.~\cite{S}.
328:
329: In conclusion, the obtained results might be important for a field-theoretic establishment of the stochastic vacuum model. Besides that,
330: the performed calculation demonstrates how one can account for the fact that the virtual particles in QCD diagrams are confined rather than free.
331: Last but not least, the novel expressions for $D$ and $D_1$ obtained here might be useful for applications in high-energy scattering.
332:
333: \section*{Acknowledgments}
334: This research was initially motivated by the discussions and collaboration with A.~Di~Giacomo, D.~Ebert,
335: E.~Meggiolaro, and Yu.A.~Simonov. Recently, I was encouraged by the discussions with N.~Brambilla and A.~Vairo, whom I also
336: thank for informing me about some of the references.
337: I would further like to thank the staff of the Physics Department of the Humboldt University of Berlin for the cordial hospitality.
338: The work was supported by the Alexander~von~Humboldt foundation.
339:
340:
341:
342:
343:
344: \begin{thebibliography}{99}
345:
346: \bibitem{cdm}
347: M.~Campostrini, A.~Di Giacomo and G.~Mussardo,
348: Z.\ Phys.\ C {\bf 25} (1984) 173.
349:
350: \bibitem{svm}
351: H.~G.~Dosch,
352: Phys.\ Lett.\ B {\bf 190} (1987) 177;
353: Yu.~A.~Simonov,
354: Nucl.\ Phys.\ B {\bf 307} (1988) 512;
355: H.~G.~Dosch and Yu.~A.~Simonov, Phys.\ Lett.\ B {\bf 205} (1988) 339.
356: %%CITATION = PHLTA,B205,339;%%
357: %%CITATION = PHLTA,B190,177;%%
358: %%CITATION = NUPHA,B307,512;%%
359: %%CITATION = ZEPYA,C25,173;%%
360:
361:
362:
363: \bibitem{book}
364: S.~Donnachie, G.~Dosch, O.~Nachtmann and P.~Landshoff,
365: {\it Pomeron physics and QCD},
366: Camb.\ Monogr.\ Part.\ Phys.\ Nucl.\ Phys.\ Cosmol.\ {\bf 19} (2002) 1.
367: %%CITATION = CMPCE,19,1;%%
368:
369: \bibitem{d}
370: A.~Di Giacomo and H.~Panagopoulos,
371: Phys.\ Lett.\ B {\bf 285} (1992) 133;
372: L.~Del Debbio, A.~Di Giacomo and Yu.~A.~Simonov,
373: Phys.\ Lett.\ B {\bf 332} (1994) 111
374: [hep-lat/9403016];
375: M.~D'Elia, A.~Di Giacomo and E.~Meggiolaro,
376: Phys.\ Lett.\ B {\bf 408} (1997) 315
377: [hep-lat/9705032];
378: A.~Di Giacomo, E.~Meggiolaro and H.~Panagopoulos,
379: Nucl.\ Phys.\ B {\bf 483} (1997) 371
380: [hep-lat/9603018];
381: A.~Di Giacomo and E.~Meggiolaro,
382: Phys.\ Lett.\ B {\bf 537} (2002) 173
383: [hep-lat/0203012];
384: M.~D'Elia, A.~Di Giacomo and E.~Meggiolaro,
385: Phys.\ Rev.\ D {\bf 67} (2003) 114504
386: [hep-lat/0205018];
387: for reviews see:
388: E.~Meggiolaro,
389: Phys.\ Lett.\ B {\bf 451} (1999) 414 [hep-ph/9807567];
390: A.~Di Giacomo,
391: {\it Non perturbative QCD}, hep-lat/9912016;
392: Czech.\ J.\ Phys.\ {\bf 51} (2001) B9
393: [hep-lat/0012013];
394: {\it QCD vacuum and confinement},
395: hep-lat/0204001.
396: %%CITATION = HEP-PH 9807567;%%
397: %%CITATION = HEP-LAT 0204001;%%
398: %%CITATION = HEP-LAT 0012013;%%
399: %%CITATION = HEP-LAT 9912016;%%
400: %%CITATION = HEP-LAT 0205018;%%
401: %%CITATION = HEP-LAT 0203012;%%
402: %%CITATION = HEP-LAT 9603018;%%
403: %%CITATION = HEP-LAT 9705032;%%
404: %%CITATION = HEP-LAT 9403016;%%
405: %%CITATION = PHLTA,B285,133;%%
406:
407:
408: \bibitem{nga}
409: G.~S.~Bali, N.~Brambilla and A.~Vairo,
410: Phys.\ Lett.\ B {\bf 421} (1998) 265
411: [hep-lat/9709079].
412: %%CITATION = HEP-LAT 9709079;%%
413:
414:
415: \bibitem{early}
416: M.~Baker, N.~Brambilla, H.~G.~Dosch and A.~Vairo,
417: Phys.\ Rev.\ D {\bf 58} (1998) 034010
418: [hep-ph/9802273];
419: D.~Antonov and D.~Ebert,
420: Eur.\ Phys.\ J.\ C {\bf 8} (1999) 343
421: [hep-th/9806153];
422: Eur.\ Phys.\ J.\ C {\bf 12} (2000) 349
423: [hep-th/9812112];
424: Phys.\ Lett.\ B {\bf 444} (1998) 208
425: [hep-th/9809018];
426: U.~Ellwanger,
427: Eur.\ Phys.\ J.\ C {\bf 7} (1999) 673
428: [hep-ph/9807380].
429: %%CITATION = HEP-TH 9809018;%%
430: %%CITATION = HEP-TH 9812112;%%
431: %%CITATION = HEP-PH 9802273;%%
432: %%CITATION = HEP-TH 9806153;%%
433: %%CITATION = HEP-PH 9807380;%%
434:
435: \bibitem{later}
436: D.~Antonov,
437: JHEP {\bf 07} (2000) 055
438: [hep-ph/0006156];
439: Phys.\ Lett.\ B {\bf 506} (2001) 303
440: [hep-ph/0011066];
441: for a recent review see:
442: D.~Antonov and D.~Ebert,
443: Phys.\ Atom.\ Nucl.\ {\bf 68} (2005) 558
444: [hep-th/0404258].
445: %%CITATION = HEP-TH 0404258;%%
446: %%CITATION = HEP-PH 0011066;%%
447: %%CITATION = HEP-PH 0006156;%%
448:
449:
450: \bibitem{dm}
451: H.~G.~Dosch, M.~Eidem\"uller and M.~Jamin,
452: Phys.\ Lett.\ B {\bf 452} (1999) 379
453: [hep-ph/9812417].
454: %%CITATION = HEP-PH 9812417;%%
455:
456: \bibitem{na}
457: N.~Brambilla, A.~Pineda, J.~Soto and A.~Vairo,
458: Nucl.\ Phys.\ B {\bf 566} (2000) 275
459: [hep-ph/9907240]; for reviews see:
460: N.~Brambilla and A.~Vairo,
461: {\it The role of the QCD vacuum in the heavy-quark bound state dynamics},
462: hep-ph/0004192;
463: N.~Brambilla, A.~Pineda, J.~Soto and A.~Vairo,
464: {\it Effective field theories for heavy quarkonium},
465: hep-ph/0410047.
466: %%CITATION = HEP-PH 0410047;%%
467: %%CITATION = HEP-PH 0310130;%%
468: %%CITATION = HEP-PH 9907240;%%
469: %%CITATION = HEP-PH 0004192;%%
470:
471: \bibitem{S}
472: Yu.~A.~Simonov,
473: {\it Analytic calculation of field-strength correlators},
474: hep-ph/0501182.
475: %%CITATION = HEP-PH 0501182;%%
476:
477: \bibitem{2}
478: R.~P.~Feynman, {\it Statistical mechanics. A set of lectures} (Addison-Wesley, Reading, MA, 1972).
479:
480: \bibitem{sb}
481: D.~Antonov and A.~Di Giacomo,
482: JHEP {\bf 03} (2005) 017
483: [hep-th/0501065].
484: %%CITATION = HEP-TH 0501065;%%
485:
486:
487:
488:
489: \end{thebibliography}
490: \end{document}
491:
492:
493:
494:
495:
496:
497:
498:
499:
500:
501:
502:
503:
504:
505:
506:
507:
508:
509:
510:
511:
512:
513:
514:
515:
516:
517:
518:
519:
520:
521:
522:
523:
524:
525:
526:
527: