hep-lat0607037/UV.tex
1: %
2: %
3: %
4: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
5: \chapter{The ultraviolet behaviour of  Green functions}
6: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
7: %
8: %
9: %
10: \label{chapter_UV_behaviour}
11: 
12: The Lagrangian of the Pure Yang-Mills theory in a four-dimensional space-time does not contain any dimensional 
13: parameters susceptible to fix an energy scale for dimensionless quantities. However, the spectrum of the 
14: corresponding quantum theory contains massive states (glueballs).  As a matter of fact, the quantum theory 
15: possesses a finite energy scale called $\Lqcd$, which is generated by the quantisation process followed by 
16: the renormalisation. All dimensionful physical quantities are expressed as multiples of powers of this scale, 
17: and thus it should be a renormalisation group invariant:
18: %
19: \begin{align}
20: \mu \frac{d}{d\mu}\Lqcd \left(\mu, g(\mu^2) \right) =0
21: \quad \rightarrow \quad
22: \left[  \mu \frac{\d}{\d \mu} + \beta \left( g(\mu^2) \right) \frac{\d}{\d g} \right] \Lqcd \left(\mu, g(\mu^2) \right) = 0,
23: \end{align}
24: %
25: where $\beta\left( g(\mu^2) \right)$ is the renormalisation group beta function and $\mu$ is the renormalisation scale. 
26: The solution of the above equation reads
27: %
28: \begin{equation}
29: \label{Lqcd_peturbative_integral}
30: \Lqcd \left(\mu, g(\mu^2) \right) = \mu \exp{\left( \displaystyle - \int^{g(\mu^2)}_{g_1} \frac{ d g^{\prime} }{\beta\left(g^{\prime}\right) } \right) },
31: \end{equation}
32: %
33: where $g_1$ is an arbitrary integration constant. $\Lqcd$ is a renormalisation scheme-dependent quantity,
34: although it is a renormalisation group invariant within one particular scheme. 
35: So it is not a real physical quantity. Still, its value is important for estimating the 
36: lowest bound of the domain of validity of perturbation theory. 
37: Knowing several first coefficients of the $\beta$-function we find from
38: the equation (\ref{Lqcd_peturbative_integral}):
39: %
40: \begin{equation}
41: \label{Lambda_pert_generic}
42: \Lqcd \left(\mu, g(\mu^2) \right) = \mu 
43: \exp\left[ \frac{1}{2\beta_0}\left( \frac{1}{g_1^2} - \frac{1}{g^2(\mu^2)} \right)
44: +\frac{\beta_1}{2\beta_0^2}\log\frac{g_1^2}{g^2(\mu^2)}
45: \right] + O(g^2).
46: \end{equation}
47: %
48: We see that there is an essential singularity when $g^2(\mu^2)\rightarrow 0$, and thus a perturbative
49: calculation of related quantities (for example, the string tension $\sqrt{\sigma} = c_\sigma \Lqcd$) 
50: is impossible. In the following sections we describe the method of calculation of $\Lqcd$
51: from lattice Green functions in Landau gauge. We start with a review of the purely perturbative results for Green
52: functions. The momentum range available on the lattice is situated at rather low energies where 
53: the non-perturbative power corrections are not negligible. The section~\ref{Green_OPE_section} is 
54: devoted to the estimation of the dominant power corrections. At the end of this chapter
55: we present the results of analysis of our lattice data.
56: 
57: %
58: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
59: \section{$\Lqcd$ and perturbative expressions for Green functions}
60: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
61: %
62: 
63: Different scalar factors of Green functions depend on the $\Lqcd$ parameter discussed in the previous 
64: subsection. These scalar factors can be calculated non-perturbatively in lattice simulations, and one can extract
65: $\Lqcd$ by fitting the lattice data in the ultraviolet domain to the corresponding 
66: perturbative formulae. Here we make a review of available perturbative Landau gauge
67: calculations for the ghost and gluon propagators in the MOM schemes.
68: 
69: If $\Gamma^{(N)}_R(p_i,g_R^2,\mu^2)$ is a renormalised proper vertex in Landau gauge, then the corresponding proper bare 
70: vertex function is independent of the renormalisation point $\mu$. This fact is reflected by the
71: Callan-Symanzik equation for the renormalised function:
72: %
73: \beq
74: \label{CALLAN_SYMANZIK}
75: \left( \frac{\d}{\d \ln{\mu^2} } + \beta\left( g_R(\mu^2) \right) \frac{\d}{\d g} - \frac{N}{2} \gamma\left(g_R(\mu^2)\right) \right)\Gamma^{(N)}_R(p_i,g_R^2,\mu^2) = 0
76: \eeq
77: %
78: where $\gamma\left(g_R(\mu^2)\right)$ is the anomalous dimension. 
79: In the Momentum subtraction schemes, the renormalisation conditions are defined by 
80: setting some of the two- and three-point functions to their tree-level values at the 
81: renormalisation point. Then (\ref{CALLAN_SYMANZIK}) simplifies to 
82: %
83: \beq
84: \label{gamma_3}
85: \lim_{a^{-1} \to \infty} \frac{d\ln(Z_{3,{\rm MOM}}(p^2=\mu^2,a^{-1})}{d\ln{\mu^2}} 
86: =\gamma_{3,{\rm MOM}}(g_{\rm MOM})   
87: \eeq
88: %
89: in the case of two-point Green functions, where $Z_3(\mu^2)$ is defined in (\ref{DEFINITION_Z3_Z3TILDE}),
90: $a^{-1}$ stands for the ultraviolet regularisation and $\gamma_{3,{\rm MOM}}(g_{\rm MOM})$ is the anomalous dimension.
91: A similar expression can be written for the ghost propagator renormalisation factor $\widetilde{Z_3}$.
92: As we have already seen in the section~\ref{GREEN_FUNCTIONS_IN_LANDAU_GAUGE},
93: there is an infinite number of MOM schemes differing by kinematic configurations at the substraction point.
94: We limit ourselves to the configurations defined by the subtraction of the transverse part 
95: of the three-gluon vertex ($\widetilde{\rm MOM}$) and that of the ghost-gluon vertex 
96: with vanishing gluon momentum ($\widetilde{\rm MOM}_{c}$) and vanishing incoming ghost momentum ($\widetilde{\rm MOM}_{c0}$),
97: discussed in the section~\ref{GREEN_FUNCTIONS_IN_LANDAU_GAUGE}.
98: 
99: Both anomalous dimensions for ghost and gluon propagators have been recently computed (\cite{Chetyrkin:2000dq},
100: \cite{Chetyrkin:2004mf}) in the $\ms$ scheme. The result at four-loop order reads
101: %
102: \begin{align}
103: \label{LnZ}  
104:   \begin{split}
105:  \frac{d\ln(Z_{3,\text{MOM}})}{d \ln \mu^{2}} &= 
106:  \frac{13}{2}\,h_{\ms} + \frac{3727}{24}\,h^{2}_{\ms} + 
107:  \left(\frac{2127823}{288} - \frac{9747}{16}\zeta_{3}\right) h^{3}_{\ms} \\
108:  &+ \left(\frac{3011547563}{6912} - \frac{18987543}{256}\zeta_{3} - 
109:    \frac{1431945}{64}\zeta_{5}\right) h^{4}_{\ms} 
110:   \nonumber 
111:   \end{split}
112:   \\
113:   \begin{split}
114:  \frac{d\ln(\widetilde{Z}_{3,\text{MOM}})}{d \ln \mu^{2}} &= 
115:  \frac{9}{4}\,h_{\ms} + \frac{813}{16}\,h^{2}_{\ms} + 
116:  \left(\frac{157303}{64} - \frac{5697}{32}\zeta_{3}\right) h^{3}_{\ms} \\
117:  &+ \left(\frac{219384137}{1536} - \frac{9207729}{512}\zeta_{3} - 
118:   \frac{221535}{32}\zeta_{5}\right) h^{4}_{\ms}   
119:   \end{split}
120: \end{align}  
121: %
122: where $h=g^2/(4\pi)^2$. In order to obtain the coefficients of the anomalous dimensions (\ref{gamma_3}) in 
123: a MOM scheme one has to express the above expressions in terms of the corresponding
124: coupling. We use the results of the article~\cite{Chetyrkin:2000dq} where the three-loop perturbative substraction 
125: of all the three-vertices appearing in the QCD Lagrangian for kinematic configurations with one 
126: vanishing momentum are given. In Landau gauge and in the pure Yang-Mills case one has the following 
127: relations between the couplings in different MOM schemes and $h_{\ms}$:
128: %
129: \begin{align}
130: \begin{split}
131: h_{\momg} = & h_{\ms} + \frac{70}{3} h^2_{\ms} + 
132: \left( \frac {51627}{576} - \frac{153}{4} \zeta(3) \right) h^3_{\ms} +
133: \\ & + \left(\frac{304676635}{6912} - \frac{299961}{64}\zeta_{3} -
134:                      \frac{81825}{64}\zeta_{5} \right)h^{4}_{\ms} \nonumber 
135: \end{split}
136: \\
137: \begin{split}
138: h_{\momc} = & h_{\ms} + \frac{223}{12} h^2_{\ms} + 
139: \left( \frac {918819}{1296} - \frac{351}{8} \zeta(3) \right) h^3_{\ms} + 
140: \\ & + \left(\frac{29551181}{864} - \frac{137199}{32}\zeta_{3} -
141:                     \frac{74295}{64}\zeta_{5} \right)h^{4}_{\ms}
142: \nonumber 
143: \end{split}
144: \\
145: \begin{split}
146: h_{\mom} = & h_{\ms} + \frac{169}{12} h^2_{\ms} + 
147: \left( \frac {76063}{144} - \frac{153}{4} \zeta(3) \right) h^3_{\ms} + 
148: \\ & + \left( \frac{42074947}{1728} - \frac{35385}{8}\zeta(3) - \frac{66765}{65}\zeta(5)  \right)
149: h^4_{\ms}.
150: \label{hmom}
151: \end{split}
152: \end{align}
153: %
154: Thus, inverting (\ref{hmom}) and substituting in (\ref{LnZ}), we obtain the anomalous dimensions  of the 
155: gluon and ghost propagator in the three above-mentioned renormalisation schemes. 
156: In a MOM scheme, the equations (\ref{LnZ}) may be integrated as functions of $h$ (cf. \ref{gamma_3}) \footnote{We omit the 
157: index specifying the renormalisation scheme both for $h$ and $\Lambda_{\text{QCD}}$ in the following formulae}:
158: %
159: \begin{align}
160:   %\label{eq:Zmom}
161:   \begin{split}
162:      \ln\left(\frac{Z_{\Gamma,\text{MOM}}}{Z_{0}}\right) & = 
163:      \log(h)\,\frac{\overline{\gamma}_{0}}{\beta_{0}} + 
164:      h\,\frac{\left(\beta_{0}\,\overline{\gamma}_{1}-\beta_{1}\,\overline{\gamma}_{0}\right)}{\beta_{0}^{2}} \\
165:      &+ h^{2}\,\frac{\left(\beta_{0}^{2}\,\overline{\gamma}_{2}-\beta_{0}\,\beta_{1}\,\overline{\gamma}_{1}-(\beta_{0}\,\beta_{2}-\beta_{1}^{2})\,\overline{\gamma}_{0}\right)}{2\beta_{0}^{3}} +  \nonumber
166:      \end{split}
167: \end{align}
168: %
169: \begin{align}
170:   \label{eq:Zmom}
171:   \begin{split}
172:      &+ h^{3}\,\bigl(\beta_{0}^{3}\,\overline{\gamma}_{3}-\beta_{0}^{2}\,\beta_{1}\,\overline{\gamma}_{2}+(\beta_{0}\,\beta_{1}^{2}-\beta_{0}^{2}\,\beta_{2})\,\overline{\gamma}_{1} \\
173:      &\qquad +(-\beta_{0}^{2}\,\beta_{3}+2\,\beta_{0}\,\beta_{1}\,\beta_{2}-\beta_{1}^{3})\,\overline{\gamma}_{0}\bigr)\frac{1}{3\beta_{0}^{4}} + \ldots
174:      \end{split}
175: \end{align}
176: %
177: where $\overline{\gamma}_i$ are the expansion coefficients of the anomalous dimension in a generic MOM type scheme and $Z_0$ 
178: is an integration constant.  The knowledge of the $\beta$-function 
179: %
180: \beq
181: \label{RENORMALISATION_GROUP_BETA}
182: \beta(h) \ = \frac{d h}{d\ln{\mu^2}}  = - \sum_{i=0}^{n} \beta_{i} \ h^{i+2} \ + \ 
183: {\cal O}\left(h^{n+3}\right) \ 
184: \eeq
185: %
186: at some order $n$ allows to calculate the momentum dependence of $h$. 
187: At four-loop order one has
188: %
189: \begin{align}
190:   \label{betainvert}
191:   \begin{split}
192:       h(t) &= \frac{1}{\beta_{0}t}
193:       \left(1 - \frac{\beta_{1}}{\beta_{0}^{2}}\frac{\log(t)}{t}
194:      + \frac{\beta_{1}^{2}}{\beta_{0}^{4}}
195:        \frac{1}{t^{2}}\left(\left(\log(t)-\frac{1}{2}\right)^{2}
196:      + \frac{\beta_{2}\beta_{0}}{\beta_{1}^{2}}-\frac{5}{4}\right)\right)+ \\
197:      &+ \frac{1}{(\beta_{0}t)^{4}}
198:  \left(\frac{\beta_{3}}{2\beta_{0}}+
199:    \frac{1}{2}\left(\frac{\beta_{1}}{\beta_{0}}\right)^{3}
200:    \left(-2\log^{3}(t)+5\log^{2}(t)+
201: \left(4-6\frac{\beta_{2}\beta_{0}}{\beta_{1}^{2}}\right)\log(t)-1\right)\right),
202:      \end{split}
203: \end{align}
204: %
205: where $t=\ln{\frac{\mu^2}{\Lambda^2_{\text{QCD}}}}$.  The last equation together with (\ref{eq:Zmom} ) 
206: allows us to write the ghost and gluon propagators as functions of the momentum and $\Lqcd$. The numerical 
207: coefficients for the $\beta$-function in (\ref{RENORMALISATION_GROUP_BETA}) are summarised in the Table~\ref{betacoefs}:
208: %
209: \begin{table}[!h]
210: \centering
211: \begin{tabular}{c|ccc}
212: \hline \hline
213:  & $\widetilde{MOM}$ & $\widetilde{MOM}_c$ & $\widetilde{MOM}_{c0}$
214: \\ \hline
215: $\beta_0$ & \multicolumn{3}{c}{$11$}
216: \\ \hdashline[0.4pt/1pt]
217: $\beta_1$ & \multicolumn{3}{c}{$102$} 
218: \\ \hdashline[0.4pt/1pt]
219: $\beta_2$ & $2412.16$ & $2952.73$ & $3040.48$
220: \\ \hdashline[0.4pt/1pt]
221: $\beta_3$ & $84353.8$ & $101484$ & $100541$
222: \\ \hline \hline
223: \end{tabular}
224: \caption{\footnotesize\it The numerical coefficients for the $\beta$-function for different MOM schemes~\cite{vanRitbergen:1997va}}
225: \label{betacoefs}
226: \end{table}
227: %
228: 
229: %
230: %
231: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
232: \section{OPE for the Green functions and dominant power corrections}
233: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
234: %
235: %
236: \label{Green_OPE_section}
237: 
238: The momentum dependence of the QCD Green functions at low energies is modified by non-perturbative effects.
239: These effects show up by presence of power-corrections to logarithmic series or, in other words,
240: by non-zero values of corresponding condensates. For example, such a non-perturbative object as instanton 
241: has a weight $\propto \exp{-\frac{8\pi^2}{g^2(p^2)}}$, giving at leading order a 
242: power correction $\propto \frac{1}{p^2}$. It is argued in (\cite{Boucaud:2000ey},\cite{Boucaud:2001qz}) 
243: that non-perturbative lattice gluonic two- and three-point functions include such 
244: contributions up to quite large energies of around $10$~GeV. For a systematic study of $\Lqcd$ one has to
245: know the influence of power corrections on the Green functions.
246: 
247: A powerfull tool to study the dependence of Green functions on the non-pert\-urb\-ati\-ve condensates is the 
248: Operator Product Expansion (OPE)~\cite{Wilson:1969zs}. This method is applicable to the problems having a specific energy
249: hierarchy, or two very different characteristic energy scales. For example, in QCD it may be applied to the study
250: of the influence of some background semi-classical field configurations. We recall here the idea of this method 
251: on the example of a two-point correlation function of a generic field $\phi$
252: %
253: \begin{equation}
254: G(x) = 
255: \left\langle
256:       \phi\left(\frac{x}{2}\right) 
257:       \phi\left(-\frac{x}{2}\right)
258: \right\rangle.
259: \end{equation}
260: %
261: It is postulated that when $x\rightarrow 0 $ the product of the fields may be expanded as
262: %
263: \begin{equation}
264:  \phi\left(\frac{x}{2}\right)  \phi\left(-\frac{x}{2}\right) = 
265:  \sum_{n=0}^{\infty}\sum_{i} w^{n}_{i}(x) \mathcal{O}^{[2n]}_i (0),
266: \end{equation}
267: %
268: where the second sum is performed on all local operators $\mathcal{O}^{[2n]}_i$ of mass dimension $2n$ having the same quantum 
269: number than the l.h.s.  The OPE suggests that all the features of the short-distance behaviour are stored in the Wilson coefficients 
270: %
271: \begin{equation}
272: w^{n}_{i}(x) \sim \left( x^2 \right)^{(n-1)} \times \Big[ \text{series in } \alpha_s \Big] ,
273: \end{equation}
274: %
275: that can be calculated in perturbation theory. In Fourier space they behave as
276: %
277: \begin{equation}
278: \widetilde{w}^{n}_{i}(p) \sim \left( \frac{1}{p^2} \right)^{(n+1)} \times \Big[ \text{series in } \alpha_s \Big],
279: \end{equation}
280: %
281: and thus 
282: %
283: \begin{equation}
284: \label{OPE_example}
285: \widetilde{G}(p) = \sum_{n=0}^{\infty}\sum_{i} \widetilde{v}^{n}_{i}\left(\alpha_s, \log{\left(p^2 / \mu^2\right) ,a^{-1}} \right)
286:  \frac{ \left\langle \mathcal{O}^{[2n]}_i \right\rangle } {\left( p^2 \right) ^{n+1}},
287: \end{equation}
288: %
289: where the coefficients $\widetilde{v}^{n}_{i}$ are computed in perturbation 
290: theory, and $\langle \mathcal{O}^{[2n]}_i \rangle$ are \textbf{vacuum condensates}. 
291: At $n=0$, corresponding to the trivial basic operator $\id$, we find an ordinary 
292: perturbative series for $\widetilde{G}$. But other condensates may lead to the appearance 
293: of non-perturbative power corrections. Usually 
294: this method is applied to gauge-invariant product of currents, and involves
295: only gauge invariant quantities (for a recent review see~\cite{Ioffe:2002ee}).
296: However it can be extended to gauge-dependent operators (like QCD propagators) and 
297: involve gauge-variant condensates (\cite{Lavelle:1992yh},\cite{Ahlbach:1991ws}).
298: We do not discuss here the subtile question of the renormalisation of condensates
299: and of calculation of their anomalous dimensions. On the lattice the MOM-type renormalisation 
300: process is non-ambiguous (\cite{Boucaud:2000nd},\cite{Boucaud:2001st}), 
301: because the non-perturbative value for the l.h.s in (\ref{OPE_example}) is available. This allows to define 
302: the condensates at fixed ultraviolet cut-off. Then one can apply a MOM renormalisation
303: prescription on the both sides of (\ref{OPE_example}) and thus renormalise 
304: the condensates $\left\langle \mathcal{O}^{[2n]}_i \right\rangle$.
305: 
306: In the following paragraph we will discuss the dominant power corrections, and corresponding 
307: condensates, in the case of the gluon and ghost correlators.
308: 
309: %
310: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
311: \subsection{The dominant OPE power correction for the gluon propagator}
312: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
313: %
314: 
315: The basis of operators in the pure Yang-Mills case is 
316: %
317: \begin{equation}
318: \underline{~\id_{~} }
319: \quad
320: A_\mu^a
321: \quad
322: c^a
323: \quad
324: \d_\mu A^a_\nu
325: \quad
326: \bar{c}^a c^b
327: \quad
328: A^a_\mu c^b 
329: \quad
330: \underline{A^a_\mu A^b_\nu}
331: \quad
332: \d_\mu c^a
333: \quad 
334: \bar{c}^a\bar{c}^b
335: \quad
336: c^a c^b
337: \quad
338: \ldots
339: \end{equation}
340: %
341: At the leading order (a $\propto 1/p^2$ power 
342: correction compared to perturbation theory) only underlined operators contribute~\cite{Boucaud:2000nd} to the gluon propagator, because 
343: operators with an odd number of fields cannot satisfy colour and Lorentz invariance and $\bar{c} c$ does not contribute because of the particular
344: structure of the ghost-gluon vertex (cf. Figure \ref{CONDENSAT_DIAGRAMS}(b)). 
345: %
346: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
347: \begin{figure}[ht]
348: \begin{center}
349: \begin{tabular}{cc}
350: \includegraphics[width=7cm]{EPS/condensat_GLUON}\hspace{0.35cm} &
351: \hspace{0.35cm} \includegraphics[width=7cm]{EPS/condensat_GHOST}
352: \\ 
353: (a) & (b)
354: \end{tabular}
355: \end{center}
356: \caption{\footnotesize\it
357: (a) Contribution of the gluon $A^2-$condensate (represented as soft external gluons) to the gluon two-point function
358: (b) Contribution of the ghost $\bar{c}c$ condensate (represented as soft external ghosts) to the gluon and ghost  two-point functions. These contributions vanish because 
359: they are proportional to ($\sim$zero) momentum of the outgoing ghost is the ghost-gluon vertex.}
360: \label{CONDENSAT_DIAGRAMS}
361: \end{figure}
362: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
363: %
364: We write then for the gluon propagator:
365: %
366: \begin{equation}
367: \begin{split}
368: \left( \widetilde{G}^{(2)} \right)^{ab}_{\mu\nu}(p^2) & \equiv  
369: \left\langle 
370: T\left( \widetilde{A}^a_\mu(-p) \widetilde{A}^b_\nu(p)\right) 
371: \right\rangle 
372: = \\ & =
373: \left( V_0 \right)^{ab}_{\mu\nu}(p^2) + 
374: \left( V_2 \right)^{ab a^\prime b^\prime}_{\mu\nu \mu^\prime \nu^\prime}(p^2) 
375: \delta^{a^{\prime} b^{\prime} }
376: \delta_{\mu^{\prime} \nu^{\prime}  }
377: \frac{\left\langle : A^c_\rho(0)A^c_\rho(0):  \right\rangle}{4(N_c^2-1)}+\ldots,
378: \end{split}
379: \end{equation}
380: %
381: where $\langle \bullet \rangle$ is a v.e.v with respect to the non-perturbative vacuum and $:A^c_\rho(0)A^c_\rho(0):$ is
382: a free-field normal product. In the perturbative vacuum the v.e.v. of all the normal 
383: products give zero, and thus only $V_0$ is non-vanishing. Hence
384: %
385: \begin{equation}
386: \left( V_0 \right)^{ab}_{\mu\nu}(p^2) = \left( \widetilde{G}^{(2)}_{\text{pert}} \right)^{ab}_{\mu\nu} (p^2).
387: \end{equation}
388: %
389: The coefficient $V_2$ is obtained at the tree-level order from
390: %
391: \begin{equation}
392: \langle g\arrowvert  
393: : A^c_\rho(0)A^c_\rho(0):  
394: \arrowvert g\rangle = 2 + O(\alpha_s)
395: \end{equation}
396: %
397: and
398: %
399: \begin{equation}
400: \langle g\arrowvert  
401: T\left( \widetilde{A}^a_\mu(-p) \widetilde{A}^b_\nu(p)\right) 
402: \arrowvert g\rangle_{\text{connected}} = 
403: \left( V_2 \right)^{ab a^\prime b^\prime}_{\mu\nu \mu^\prime \nu^\prime}(p^2) 
404: \langle g\arrowvert  
405: : \widetilde{A}^{a^\prime}_{\mu^\prime}(0) \widetilde{A}^{b^\prime}_{\nu^\prime}(0) : 
406: \arrowvert g\rangle,
407: \end{equation}
408: %
409: where $\arrowvert g\rangle$ is a soft gluon state. So, using the LSZ rule to cut the soft external gluons, we obtain
410: %
411: \begin{equation}
412: \label{wilson_coeff_gluon}
413: \left( V_2 \right)^{ab a^\prime b^\prime}_{\mu\nu \mu^\prime \nu^\prime}(p^2)  = 
414: \frac{1}{2}
415: \frac{
416: 	\left\langle 
417: 		\widetilde{A}^{t}_{\tau}(0)  
418: 		\widetilde{A}^{a}_{\mu}(-p)
419: 		\widetilde{A}^{b}_{\nu}(p)
420: 		\widetilde{A}^{s}_{\sigma}(0)
421: 	\right\rangle}{ \left( G^{(2)}_{\text{pert}}(0) \right)^{t a^\prime}_{\tau\mu^\prime}   \left( G^{(2)}_{\text{pert}} (0) \right)^{s b^\prime}_{\sigma\nu^\prime} },
422: \end{equation}
423: %
424: which can be computed in perturbation theory (cf. Figure \ref{CONDENSAT_DIAGRAMS}(a)). Finally,
425: %
426: \begin{equation}
427: \left( \widetilde{G}^{(2)} \right)^{ab}_{\mu\nu}(p^2) = 
428: \frac{1}{p^2} \left( \delta_{\mu\nu} - \frac{p_\mu p_\nu}{p^2}\right)
429: \left( 
430: 	p^2 \widetilde{G}^{(2)}_{\text{pert}} (p^2) 
431: 	+ N_C\frac{g_0^2 \langle A^2 \rangle}{4\left(N_C^2 -1\right)}\frac{1}{p^2} + O(g^4,p^{-4})
432: \right).
433: \end{equation}
434: %
435: A MOM-type renormalisation prescription may be defined non-perturbatively. This allows an easy renormalisation 
436: procedure for the $A^2-$condensate~\cite{Boucaud:2000nd}. Here we do not include the 
437: effects of the anomalous dimension of the $A^2$ operator~\cite{Boucaud:2002jt} and hence 
438: we apply the MOM prescription by imposing the tree-level value to the Wilson coefficient at the 
439: renormalisation point $p^2=\mu^2$ for the last equation. This allows to factorise the perturbative gluon propagator giving finally
440: %
441: \begin{equation}
442: \label{Z3gluon}
443: {Z_3}(\mu^2) \ = \ {Z_{\rm 3,pert}}(\mu^2) \
444: \left(  1 + \frac{N_C}{\mu^2} \frac{g^2_R \langle A^2 \rangle_R} {4 \left(N_C^2-1\right)} + O(g_R^4,\mu^{-4}) \right).
445: \end{equation}
446: %
447: 
448: 
449: %
450: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
451: \subsection{The dominant OPE power correction for the ghost propagator}
452: \label{OPEsection}
453: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
454: %
455: 
456: In the case of the ghost propagator the set of basic operators is the same, the ghost condensate $\bar{c}c$ does not contribute for the same reasons as for the
457: gluon propagator (cf. Figure~\ref{CONDENSAT_DIAGRAMS}(b) ). Thus, applying the OPE to the ghost two-point function, we obtain:
458: %
459: \beq\label{OPE1}
460: \widetilde{F}^{(2)a b}(p^2) &=& (\widetilde{V}_0)^{a b}(p^2) \ + \ \left( \widetilde{V}_2 \right)^{a b \sigma \tau}_{s t}(p^2)
461: \langle : A_\sigma^s(0) A_\tau^t(0): \rangle \ + \ \dots \nonumber \\ 
462: &=& F^{(2)a b}_{\rm pert}(p^2) \ + \ 
463: w^{a b} \ \frac{\langle A^2 \rangle}{4 (N_c^2-1)} \ + \ \dots 
464: \eeq
465: %
466: where, in analogy with (\ref{wilson_coeff_gluon}), the Wilson coefficient reads
467: %
468: \begin{small}
469: \beq\label{OPE3}
470: w^{a b} \ &=& \ \left( \widetilde{V}_2 \right)^{a b \sigma \tau}_{s t} \delta^{s t} \delta_{\sigma \tau} \ = \ 
471: \frac 1 2 \ \delta^{s t} \delta_{\sigma \tau} \frac{\int d^4x e^{i p \cdot x} \
472: \langle \Am{\tau'}{t'}{0} \ T\left( c^a \overline{c^b} \right) \ \Am{\sigma'}{s'}{0} \rangle_{\rm connected}}
473: {{G^{(2)}}_{\sigma \sigma'}^{s s'}(0) {G^{(2)}}_{\tau \tau'}^{t t'}(0) }
474: % \nonumber \\
475: %&=&   2 \times \rule[0cm]{0cm}{1.7cm} \ghost,
476: \eeq
477: \end{small}
478: %
479: which is equal to twice the diagram represented on the Figure~\ref{CONDENSAT_DIAGRAM_A2_GHOST_PROPAGATOR}
480: %
481: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
482: \begin{figure}[ht]
483: \begin{center}
484: \includegraphics[scale=1]{EPS/condensat_A2_GHOST}
485: \end{center}
486: \caption{\footnotesize\it Contribution of the gluon $A^2-$condensate (external soft gluons) to the ghost propagator} 
487: \label{CONDENSAT_DIAGRAM_A2_GHOST_PROPAGATOR}
488: \end{figure}
489: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
490: %
491: that describes the coupling of  the ghost propagator to the gluon $A^2-$condensate. Hence
492: %
493: \begin{equation}
494: w^{a b} = \frac{1}{2}\delta^{st}\delta_{\sigma\tau}
495: \cdot 2 \frac{\delta^{aa_1}}{p^2}\left( i g_0 f^{a_2 t a_1}\right)
496: \frac{\delta^{a_2 a_3}}{p^2}
497: \left( ig_0 f^{a_4 s a_3}\right)
498: \frac{\delta^{a_4 b} }{p^2} = N_C \frac{g_0^2}{p^2}\widetilde{F}^{(2)ab}_{\text{tree}}(p^2).
499: \end{equation}
500: %
501: This gives the leading non-perturbative contribution, because the first Wilson coefficient 
502: trivially gives the perturbative ghost propagator. Finally,
503: %
504: \beq\label{Fin1}
505: \widetilde{F}^{(2)a b}(p^2) \ = \ \widetilde{F}_{\rm pert}^{(2)a b}(p^2) \
506: \left( 1 + \frac{N_C}{q^2} \ 
507: \frac{g^2_0 \langle A^2 \rangle} {4 \left(N_C^2-1\right)}  + \ {\cal O}\left(g_0^4,q^{-4} \right) \right) \
508: \eeq
509: %
510: where all quantities are bare. Performing the MOM renormalisation 
511: we obtain for the renormalisation factor:
512: %
513: \beq\label{Z3fantome}
514: \widetilde{Z_3}(\mu^2) \ = \ \widetilde{Z_{\rm 3,pert}}(\mu^2) \
515: \left(  1 + \frac{N_C}{\mu^2} \frac{g^2_R \langle A^2 \rangle_R} {4 \left(N_c^2-1\right)}  
516: + O(g_R^4,\mu^{-4}) \right) \ ,
517: \eeq
518: %
519: where the $A^2$-condensate is renormalised as in the case of the gluon propagator.
520: We see that the dominant multiplicative correction to the 
521: perturbative $\widetilde{Z_{\rm 3,pert}}$  is identical to the one obtained 
522: in the previous section for the gluon propagator (\ref{Z3gluon}).
523: 
524: 
525: %
526: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
527: \subsection{Constraints on the Wilson coefficients from the Slavnov-Taylor identity}
528: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
529: %
530: \label{subsection_on_Wilson_coeff_from_ST}
531: 
532: The gauge-dependent power corrections due to the $\langle A^2\rangle$-condensate
533: are obviously absent in gauge-invariant quantities. Because of this the Wilson
534: coefficients for the $\langle A^2\rangle$-condensate in different Green functions
535: are not independent. Some relations may be obtained from 
536: the Slavnov-Taylor identity (\ref{STid}) but their role in the MOM renormalisation
537: constants is not obvious.
538: %
539: %For example, considering the limit
540: %$r\rightarrow 0$ in (\ref{STid}) (see section~\ref{section_Slavnov_Taylor_IR}
541: %for details, equation (\ref{WIhab})) one obtains a relation between the 
542: %MOM renormalisation factors of the three-gluon vertex and the ghost-gluon vertex:
543: %%
544: %\begin{equation}
545: % \label{ST_A2_ghost_gluon_vertex}
546: % \frac{ Z_1^{\widetilde{\text{\footnotesize MOM}}}(\mu^2) }
547: % { \widetilde{Z_1}^{\widetilde{\text{\footnotesize MOM}}_{c}}(\mu^2) }
548: % = \frac{Z_3(\mu^2)}{\widetilde{Z_3}(\mu^2)}.
549: % \end{equation}
550: % %
551: % The equality of the dominant power corrections found in the previous subsections
552: % allows to conclude that the corrections to the three-gluon vertex in
553: % an asymmetric kinematic configurations~\cite{DeSoto:2001qx} and to the
554: % ghost-gluon vertex with vanishing gluon momentum are also equal. Thus 
555: % the ghost-gluon vertex contains power corrections since the thee-gluon vertex does~\cite{DeSoto:2001qx}.
556: %
557: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
558: \begin{figure}[!h]
559: \begin{center}
560: \includegraphics[width=0.3\linewidth]{EPS/1PI_ghost-gluon-vertex}
561: \end{center}
562: \caption{\footnotesize\it The $\langle A^2 \rangle$ contribution to the ghost-gluon vertex with 
563: vanishing entering momentum. The above diagram is zero in Landau gauge because of the projector in the gluon propagator.}
564: \label{CONDENSAT_DIAGRAM_A2_GHOST_GLUON_VERTEX_1PI}
565: \end{figure}
566: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
567: %
568: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
569: \begin{figure}[!h]
570: \begin{center}
571: \includegraphics[width=0.3\linewidth]{EPS/condensat_MOMc_vertex}
572: \end{center}
573: \caption{\footnotesize\it Non-zero dominant $\langle A^2 \rangle$ contribution to the ghost-gluon vertex with 
574: vanishing entering momentum. This term contribute to the external ghost propagator.}
575: \label{CONDENSAT_DIAGRAM_A2_GHOST_GLUON_VERTEX}
576: \end{figure}
577: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
578: %
579: % Unfortunately, it is not possible to use (\ref{STid}) in order 
580: % to relate the Wilson coefficient in the case of the ghost-gluon vertex 
581: % with vanishing ghost momentum to any other vertex. 
582: 
583: It is interesting to know if there are any power corrections
584: to this vertex, because perturbation theory predicts the non-renormalisation of
585: this vertex, i.e. it is equal to $1$ to all orders (\cite{Taylor:1971ff},\cite{Chetyrkin:2000dq}).
586: If the $\langle A^2\rangle$ power corrections are present they will constitute
587: the main contribution at low energies. One can directly evaluate the Wilson coefficient
588: to the ghost-gluon vertex $\widetilde{\Gamma}_{\mu}^{abc}(-p,0;p)$ with 
589: vanishing entering ghost momentum.
590: 
591: The only non-zero correction to the ghost-gluon vertex with one vanishing ghost momentum is
592: the one that contributes to the external ghost propagator (see Figure~\ref{CONDENSAT_DIAGRAM_A2_GHOST_GLUON_VERTEX}).
593: But all the diagrams with the condensate interaction attached to different external legs
594: are zero in Landau gauge (see Figure~\ref{CONDENSAT_DIAGRAM_A2_GHOST_GLUON_VERTEX_1PI}).
595: Thus the ghost-gluon vertex in this particular kinematic configuration does not contain
596: the $\frac{1}{p^2}$ power-corrections, and thus the non-renormalisation theorem holds at this order.
597: However, this it is not true if the external ghost momentum is not \emph{exactly} zero.
598: 
599: %
600: %
601: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
602: \section{Data analysis}
603: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
604: %
605: %
606: 
607: We calculated, using the techniques described in the chapter~\ref{chapter_Lattice_Green_functions},
608: the ghost and the gluon propagators of the Landau gauge $SU(3)$ gauge theory at different lattice volumes and different values of 
609: the $\beta$ parameter (cf. Table~\ref{tab:simulation})~\cite{Boucaud:2005np}.
610: %
611: \begin{table}[h]
612:   \centering
613: \begin{tabular}[h]{rcccc}
614:   \hline
615:   $\beta$ & $V$ &  $a^{-1}$ (GeV) & $V_{\text{phys}}$ ($\text{fm}^4$) & \# Configurations \\
616:   \hline
617:   $\rightarrow6.0$   &  $16^{4}$ & $1.96$ & $6.73$ & $1000$ \\
618:   $6.0$   &  $24^{4}$ & $1.96$ & $33.17$ & $500$ \\
619:   $\rightarrow6.2$   &  $24^{4}$ & $2.75$ & $8.43$ & $500$  \\
620:   $\rightarrow6.4$   &  $32^{4}$ & $3.66$ & $8.85$ & $250$ \\
621:   \hline
622: \end{tabular}
623: \caption{\footnotesize\it Lattice setup parameters. The lattice spacings are taken from Table~3 in \cite{Bali:1992ru} 
624: with a physical unit normalised by $\sqrt{\sigma}=445$ MeV. The lattices marked by the ``$\rightarrow$" symbol
625: have similar physical volume. }
626: \label{tab:simulation}
627: \end{table}
628: %
629: The lattices marked by the ``$\rightarrow$" symbol correspond to similar physical volume. The produced data allow
630: us to study the propagators in the momentum range $[\approx 2\text{GeV},\, \approx 6.5\text{GeV}]$.
631: 
632: This section is organised in the following way. In the first subsection we 
633: present the fits of the ghost and the gluon propagators separately,
634: and compare the fitted values for $\Lqcd$. Thus we test the self-consistency of the method. 
635: Non-perturbative effects are quite important in the energy interval accessible to us.
636: This is why another motivation is to study the asymptoticity of the perturbative series.
637: The latter is done by comparing the results in different renormalisation schemes 
638: ($R=\ms,\widetilde{\text{MOM}},\widetilde{\text{MOM}_c},\widetilde{\text{MOM}_{c0}}$) 
639: and at different orders (from two to four loops). In the second subsection
640: we use the analytical result of the previous section namely that the
641: dominant non-perturbative effects are the same for the ghost and for the gluon propagators,
642: and hence the ratio of the gluon and the ghost dressing functions is 
643: better described by perturbation theory at low energies. We shall see that lattice data support this claim.
644: 
645: 
646: %
647: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
648: \subsection{Fitting the gluon and the ghost propagators}
649: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
650: %
651: 
652: We extracted $\Lqcd$ from the dressing functions of our lattice propagators by fitting them to the 
653: formula (\ref{eq:Zmom}) (with $h$ given by (\ref{betainvert}) ) in different MOM renormalisation schemes.
654: There are two parameters of the fit - the wanted $\Lqcd$ and the integration constant $Z_0$.
655: An example of such a fit if presented at Figure~\ref{exemple_fits_Zn}. 
656: %
657: \begin{figure}[h]
658: \centering
659: \psfig{figure=EPS/Zn, width=0.9\linewidth}
660: \caption{\footnotesize\it Extrapolated lattice data at $\beta=6.4$ for $G(p)$ (left)
661: and $F(p)$ (right). The solid line is the fit at
662: four-loop order in the $\ms$ scheme. The vertical dotted lines
663: delimit the window of each fit.}
664: \label{exemple_fits_Zn}
665: \end{figure}
666: %
667: The obtained value of $\Lambda_{R}$ in a scheme $R$ is converted to 
668: the $\ms$ scheme using the exact~\footnote{A relation between the values of $\Lqcd$ in 
669: different schemes $\textit{A}$ and $\textit{B}$ reads 
670: %
671: \begin{equation}
672: \label{conversion_asympt_a_une_boucle}
673: \frac{\Lambda^{\textit{B}}}{\Lambda^{\textit{A}}} = \exp{\left[\frac{1}{2\beta_0} 
674: \left( \frac{1}{g_{\textit{A}}^2(\mu^2)} -\frac{1}{g_{\textit{B}}^2(\mu^2)} \right) + O\left(g^2(\mu^2)\right) \right]}.
675: \end{equation}
676: %
677: If $g^2_{\textit{B}}=g^2_{\textit{A}}\left(1+ \zeta\,\frac{g^2_{\textit{A}}}{4\pi}+\ldots\right)$ then the asymptotic freedom
678: gives $\Lambda^{\textit{B}}=\Lambda^{\textit{A}}e^{\frac{\zeta}{2\beta_0}}$. Thus the exact conversion coefficient
679: is given by an one-loop calculation~\cite{Celmaster:1979km}. } asymptotic formulae
680: %
681: \begin{equation}
682: \label{conversion}
683: \begin{array}{l}
684: \Lambda_{\ms} = 0.346\, \Lambda_{\widetilde{\text{MOM}} }
685: \\
686: \Lambda_{\ms} = 0.429\, \Lambda_{\widetilde{\text{MOM}_c} }
687: \\
688: \Lambda_{\ms} = 0.428\, \Lambda_{\widetilde{\text{MOM}_{c0}} }
689: \end{array}
690: \end{equation}
691: %
692: The results in $\ms$ are given in Tables~\ref{tab:ms_values},\ref{tab:mom_tilde_values1},\ref{tab:mom_tilde_values2}. 
693: The errors include the statistical error, extrapolation errors and the bias due to the choice of the fit window.
694: %
695: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%
696: \begin{table}[!h]
697: \centering
698: \begin{tabular}{c||c;{0.4pt/1pt}c|c;{0.4pt/1pt}c}
699: \hline
700: $\beta$ & $\Lambda^{(3)}_{\ms,gluon}$ & $\Lambda^{(3)}_{\ms,ghost}$ &
701: $\Lambda^{(4)}_{\ms,gluon}$ & $\Lambda^{(4)}_{\ms,ghost}$ \\
702: \hline
703: 6.0 &  $519(6)^{+12}_{-4~}$ & $551(12)^{+33}_{-16}$ & $441(4)^{+8~}_{-4~}$ & 
704: $461(10)^{+29}_{-14}$ \\
705: 6.2 &  $509(6)^{+17}_{-27}$ & $550(8)^{+27}_{-63}~$ & $435(6)^{+11}_{-19}$ & 
706: $465(8)^{+33}_{-36}~$ \\
707: 6.4 &  $476(7)^{+44}_{-40}$ & $549(7)^{+55}_{-51}~$ & $410(4)^{+33}_{-29}$ & 
708: $468(7)^{+48}_{-40}~$ \\
709: \hline
710: \end{tabular}
711: \caption{\footnotesize\it Three-loop and four-loop physical values of $\Lambda_{\ms}$ in MeV
712: extracted from fits in the $\ms$ scheme.}
713: \label{tab:ms_values}
714: \end{table}
715: %%%%%%%%%%%%%%%%%%%%%%%%%%%
716: %
717: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%
718: \begin{table}[!h]
719: \centering
720: \begin{tabular}{c||c;{0.4pt/1pt}c|c;{0.4pt/1pt}c}
721: 	\hline
722: 	$\beta$ & $\Lambda^{(3)}_{\ms,gluon}$ & $\Lambda^{(3)}_{\ms,ghost}$ & $\Lambda^{(4)}_{\ms,gluon}$ & $\Lambda^{(4)}_{\ms,ghost}$ \\
723: 	\hline
724: 	6.0 &  324(2)$^{+2~}_{-5~}$ & 322(8)$^{+20}_{-16}$ & --- & --- \\
725: 	6.2 &  320(2)$^{+8~}_{-14}$ & 326(5)$^{+26}_{-33}$ & --- & 331(8)$^{+21}_{-16}$ \\
726: 	6.4 &  312(1)$^{+9~}_{-25}$ & 331(4)$^{+42}_{-35}$ & 320(4)$^{+6~}_{-4~}$ & 353(9)$^{+17}_{-38}$ \\
727: 	\hline
728: \end{tabular}
729: \caption{\footnotesize\it Three-loop physical values of $\Lambda_{\ms}$ in MeV
730: converted from fits in the $\widetilde{\text{MOM}}$ scheme.
731: }
732: \label{tab:mom_tilde_values1}
733: \end{table}
734: %
735: %
736: \begin{table}[!h]
737: \centering
738: \begin{tabular}{c||c;{0.4pt/1pt}c|c;{0.4pt/1pt}c}
739: 	\hline
740: 	$\beta$ & $\Lambda^{(3)}_{\ms,gluon}$ & $\Lambda^{(3)}_{\ms,ghost}$ & $\Lambda^{(4)}_{\ms,gluon}$ & $\Lambda^{(4)}_{\ms,ghost}$ \\
741: 	\hline
742: 	6.0 &  345(3)$^{+4~}_{-4~}$ & 369(9)$^{+3~}_{-2~}$ & --- & --- \\
743: 	6.2 &  341(2)$^{+6~}_{-7~}$ & 364(8)$^{+11}_{-19}$ & 344(4)$^{+9~}_{-6~}$ & 357(10)$^{+8~}_{-16}$ \\
744: 	6.4 &  323(2)$^{+17}_{-11}$ & 354(8)$^{+28}_{-20}$ & 332(2)$^{+14}_{-30}$ & 351(8)$^{+23}_{-25}$ \\
745: 	\hline
746: \end{tabular}
747: \caption{\footnotesize\it Three-loop physical values of $\Lambda_{\ms}$ in MeV
748: converted from fits in the $\widetilde{\text{MOM}}_c$ scheme.
749: }
750: \label{tab:mom_tilde_values2}
751: \end{table}
752: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%
753: %
754: % % % %
755: % % % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%
756: % % % \begin{table}[!h]
757: % % % \centering
758: % % % \begin{tabular}{lr}
759: % % % 	\begin{tabular}{c|c;{0.4pt/1pt}c}
760: % % % 	\hline
761: % % % 	$\beta$ & $\Lambda^{(3)}_{\ms,gluon}$ & $\Lambda^{(3)}_{\ms,ghost}$ \\
762: % % % 	\hline
763: % % % 	6.0 &  324(2)$^{+2~}_{-5~}$ & 322(8)$^{+20}_{-16}$  \\
764: % % % 	6.2 &  320(2)$^{+8~}_{-14}$ & 326(5)$^{+26}_{-33}$   \\
765: % % % 	6.4 &  312(1)$^{+9~}_{-25}$ & 331(4)$^{+42}_{-35}$  \\
766: % % % 	\hline
767: % % % 	\end{tabular}
768: % % % & \hspace*{0.1\linewidth}
769: % % % 	\begin{tabular}{c|c;{0.4pt/1pt}c}
770: % % % 	\hline
771: % % % 	$\beta$ & $\Lambda^{(3)}_{\ms,gluon}$ & $\Lambda^{(3)}_{\ms,ghost}$ \\
772: % % % 	\hline
773: % % % 	6.0 &  345(3)$^{+4~}_{-4~}$ & 369(9)$^{+3~}_{-2~}$  \\
774: % % % 	6.2 &  341(2)$^{+6~}_{-7~}$ & 364(8)$^{+11}_{-19}$  \\
775: % % % 	6.4 &  323(2)$^{+17}_{-11}$ & 354(8)$^{+28}_{-20}$  \\
776: % % % 	\hline
777: % % % 	\end{tabular}
778: % % % \end{tabular}
779: % % % \caption{\footnotesize\it Three-loop physical values of $\Lambda_{\ms}$ in MeV
780: % % % converted from fits in the $\widetilde{\text{MOM}}$ scheme (left) and in
781: % % % the $\widetilde{\text{MOM}}_c$ scheme(right).
782: % % % }
783: % % % \label{tab:mom_tilde_values}
784: % % % \end{table}
785: % % % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%
786: % % % %
787: We see from these tables that at a given order and in a given renormalisation scheme the
788: values obtained from the gluon and ghost propagators are consistent within the error bars, and
789: are quite independent of the ultraviolet cut-off. The results from a direct fit in the $\ms$ scheme
790: (Table~\ref{tab:ms_values}) confirm the old claim that we are still far from asymtoticity 
791: in the considered momentum interval in this scheme~\cite{Becirevic:1999hj}.
792: %
793: \begin{figure}[ht]
794: \centering
795: \psfig{figure=EPS/asympt_lambda_ratio_orders, width=0.9\linewidth}
796: \caption{\footnotesize\it $\frac{\Lambda^{(n+1)}_{R}}{\Lambda^{(n)}_{R}}$ for
797: $n=2$ (dashed lines) and $n=3$ (solid lines), for the
798: gluon propagator in the $\ms$ scheme (a), $\widetilde{\text{MOM}}$ scheme (c) and
799: $\widetilde{\text{MOM}}_c$ scheme (e), and for the ghost propagator in the $\ms$
800: scheme (b), $\widetilde{\text{MOM}}$ scheme (d) and $\widetilde{\text{MOM}}_c$ scheme (f).}
801: \label{asympt_lambda_ratio}
802: \end{figure}
803: %
804: Our analysis suggests that the perturbative series 
805: become asymptotic at the NNLO in the case of $\widetilde{\text{MOM}}$ and $\widetilde{\text{MOM}}_c$ 
806: renormalisation schemes. However, the property of asymptoticity is only 
807: approximate at considered momenta. To see this one can use the perturbative expression 
808: (analogue to (\ref{Lambda_pert_generic})) for $\Lambda_{R}$ in terms of the coupling $h_R$ 
809: to the order four
810: %
811: \begin{equation}
812: \label{Lambda_n_n_plus_1}
813: 2\ln\Lambda^{(4)}_{R} = \ln\mu^{2} - \frac{1}{\beta_{0}h_{R}} - 
814: \frac{\beta_{1}}{\beta_{0}^{2}}\ln(\beta_{0}h_{R}) - 
815: \frac{\beta_{0}\beta_{2}-\beta_{1}^{2}}{\beta_{0}^{3}}h_{R} -
816: \frac{\beta_{0}^{2}\beta_{3} - 2\beta_{0}\beta_{1}\beta_{2} +
817: \beta_{1}^{3}}{2\beta_{0}^{4}}h_{R}^{2},
818: \end{equation}
819: %
820: and plot the ratio (Figure~\ref{asympt_lambda_ratio}) of the 
821: consecutive orders $\frac{\Lambda^{(n+1)}_{R}}{\Lambda^{(n)}_{R}}$.
822: There is a qualitative agreement between the ratios presented at 
823: Figure~\ref{asympt_lambda_ratio} and our results 
824: (see Tables~5-10 in \cite{Boucaud:2005np}). The influence of truncation,
825: responsible for the differences between different orders and renormalisation schemes,
826: is mostly due to the large value of the effective coupling at considered energies~\cite{Boucaud:2005np}.
827: In fact, as shown in~\cite{Boucaud:2000ey}, the real value of the coupling constant may
828: be smaller, because of the power correction discussed in the section~\ref{Green_OPE_section}. 
829: Indeed, according to the OPE analysis the effective coupling constant is modified by a factor 
830: %
831: \begin{equation}
832: \alpha_s\rightarrow\alpha_s\left(  1 + \text{const} \cdot \frac{\langle A^2 \rangle}{p^2} \right).
833: \end{equation}
834: %
835: According to the results of the section~\ref{Green_OPE_section}, one can eliminate the dominant power correction 
836: by considering the ratio of the propagators. In this case one expects a better
837: behaviour of perturbative series at low momenta. We discuss this strategy in the following subsection.
838: 
839: %
840: %
841: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
842: \subsection{Fit of the ratio}
843: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
844: %
845: 
846: Given that at the leading order the non-perturbative power 
847: corrections factorise (\ref{Z3gluon}),(\ref{Z3fantome}) and are identical 
848: in the case of the ghost and gluon propagators, we can fit the ratio
849: %
850: \beq
851: \label{ratioNP}
852: \frac{\widetilde{Z_3}(q^2,\Lambda_{R},\langle A^2 \rangle)}
853: {Z_3(q^2,\Lambda_{R},\langle A^2 \rangle)} = 
854: \frac{\widetilde{Z_{3,\text{pert}}}(q^2,\Lambda_{R})}{Z_{3,\text{pert}}(q^2,\Lambda_{R})},
855: \eeq
856: %
857: to the ratio of \emph{perturbative} formulae in scheme $R$ given by (\ref{eq:Zmom}), and
858: then convert $\Lambda_{R}$ to $\Lambda_{\ms}$ using (\ref{conversion}). It is interesting to notice 
859: that non-perturbative corrections cancel out in this ratio even in
860: the unquenched case with $n_f \neq 0$ flavours of dynamical quarks. The $\Lambda_{\rm QCD}$-parameter extracted from this ratio 
861: is free from non-perturbative power corrections up to contributions related to the operators of dimension four.
862: In Table~\ref{best-fits} the best-fit parameters for the three schemes are presented and we plot in 
863: Figure~\ref{ratio} the lattice data and the $\widetilde{\text{MOM}}$ best-fit 
864: curve for the ratio (\ref{ratioNP}).
865: %
866: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
867: \begin{figure}[ht]
868: \begin{center}
869: \includegraphics[width=0.6\linewidth]{EPS/ratioMOMg}
870: \end{center}
871: \caption{\footnotesize\it Plot (in the $\widetilde{\text{MOM}}$ scheme) of the $\frac{Z_3(p^2)}{\widetilde{Z}_3(p^2)}$ for the best fit parameter $\Lambda_{\ms}=269(5)$ MeV. }
872: \label{ratio}
873: \end{figure}
874: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
875: %
876: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
877: \begin{figure}[ht]
878: \begin{center}
879: \includegraphics[width=0.6\linewidth]{EPS/kmin}
880: \end{center}
881: \caption{\footnotesize\it The determination of the optimal window fit (from 3 GeV to $k_{\rm max} a \le \pi/2$) 
882: results from the search for 
883: some ``plateau'' of $\Lambda_{\ms}$ when one changes the low bound of the fit window. Fits are done in the 
884: $\widetilde{\text{MOM}}$ scheme.}
885: \label{kmin}
886: \end{figure}
887: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
888: %
889: %
890: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
891: \begin{table}[ht]
892: \begin{center}
893: \begin{tabular}{c|c;{0.4pt/1pt}c|c;{0.4pt/1pt}c|c;{0.4pt/1pt}c}
894: \hline 
895: scheme & $\Lambda_{\ms}^{(2)}$ & $\chi^2$/n.d.f & $\Lambda_{\ms}^{(3)}$ & $\chi^2$/n.d.f & $\Lambda_{\ms}^{\text{4 loops}}$ & $\chi^2$/n.d.f 
896: \\ \hline
897: $\widetilde{\text{MOM}}$ \rule[0cm]{0cm}{0.5cm} &  $324(6)$ & $0.33$ & $269(5)$ & $0.34$ & $282(6)$ & $0.34$
898:  \\
899: $\widetilde{\text{MOM}}_c$ \rule[0cm]{0cm}{0.5cm} &  $351(6)$ & $0.33$ & $273(5)$ & $0.34$ & $291(6)$ & $0.33$
900:  \\
901: $\widetilde{\text{MOM}}_{c0}$  \rule[0cm]{0cm}{0.5cm} &  $385(7)$ & $0.33$ & $281(5)$ & $0.34$ & $298(6)$ & $0.33$
902:  \\ 
903: \hline 
904: \end{tabular}
905: \end{center}
906: \caption{\footnotesize\it The best-fitted values of $\Lambda_{\ms}$ for the three considered renormalisation 
907: schemes. As discussed in the text, $\momc$ seems to be 
908: the one showing the best asymptotic behaviour.}
909: \label{best-fits}
910: \end{table}
911: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
912: %
913: In Figure~\ref{loops_a} we show the evolution of the fitted parameter $\Lambda_{\ms}$ 
914: when changing the order of perturbation theory used in the fitting formula.
915: One can conclude from Figure~\ref{loops_a} that the $\widetilde{\text{MOM}}$ scheme 
916: at three loops gives the most stable result for $\Lambda_{\ms}$. It can also be seen from
917: the ratio of four to three loops contributions (see Figure~\ref{loops_b}) 
918: for the perturbative expansion of $\ln{Z_3}$,
919: %
920: \beq\label{LnZ-loops}
921: \ln{Z_3} \ = \ r_0 \ln{h_R} + \sum_{i=1} r_i h_R^i \ ,
922: \eeq
923: %
924: where the coefficients $r_i$ are to be computed from those 
925: in equations (\ref{LnZ}-\ref{betainvert}) using the Table~\ref{betacoefs}. 
926: The same is done for $\ln{\widetilde{Z_3}}$. 
927: 
928: According to our analysis, and in agreement with the result of the separate fit, three loops seems to 
929: be the optimal order for the asymptoticity~\footnote{Note that the the asymptoticity property is better verified in
930: the case of the ratio, see Figure~\ref{loops_a}.}. Indeed, the values of $\Lambda_{\ms}$ for the three considered 
931: renormalisation schemes practically match each other at three loops (see Figure~\ref{loops_a}). 
932: The approximate value
933: %
934: \beq
935: \label{Lambda-final}
936: \Lambda_{\ms}=269(5)^{+12}_{-9}
937: \eeq
938: %
939: could be presented as the result for the fits of the ratio of dressing functions to perturbative 
940: formulae.
941: 
942: The results of the previous subsection and~\cite{Boucaud:2005np} suggest that our present 
943: systematic uncertainty may be underestimated (narrowness of the momentum interval, truncation of the
944: perturbative series, etc.), that is why we prefer simply to quote $\Lambda_{\ms} \approx 270 \text{ MeV}$
945: for future reference. This value is pretty smaller than the 
946: value of $\approx330\text{ MeV}$ obtained by independent fits of dressing 
947: functions (see Tables~\ref{tab:mom_tilde_values1},\ref{tab:mom_tilde_values2}). In light of our
948: OPE analysis and previous results~\cite{Boucaud:2001st}, this argues in favour of presence of low-order
949: non-perturbative corrections to the ghost and gluon propagators in the momentum range $[2\text{ GeV},6\text{ GeV}]$.
950: %
951: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
952: \begin{figure}[h]
953: \vspace*{0.51cm}
954: \begin{center}
955: \includegraphics[width=0.6\linewidth]{EPS/lambda}
956: \end{center}
957: \caption{\footnotesize\it Evolution of the parameter $\Lambda_{\ms}$, extracted from fits 
958: of the ratio \ref{ratioNP} and propagators alone (rhombus and star markers, extracted from 
959: Tables~\ref{tab:mom_tilde_values1},\ref{tab:mom_tilde_values2}~\cite{Boucaud:2005np}) to perturbative 
960: formulae, as function of the order of perturbation theory. The solid line corresponds to the value (\ref{Lambda-final}). 
961: Only statistical error is quoted.} 
962: \label{loops_a}
963: \end{figure}
964: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
965: %
966: %
967: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
968: \begin{figure}[h]
969: \begin{center}
970: \includegraphics[width=0.4\linewidth]{EPS/loops}
971: \end{center}
972: \caption{\footnotesize\it (b) Ratio of four-loop to three-loop 
973: contributions (and of three-loop to two-loops for the sake of comparison) for the perturbative 
974: expansion of $\log{Z_3}$ and $\log{\widetilde{Z_3}}$ (in $\widetilde{\text{MOM}}$) in \ref{LnZ-loops},
975: plotted versus the momenta inside our fitting window.} 
976: \label{loops_b}
977: \end{figure}
978: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
979: %
980: 
981: 
982: %
983: %
984: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
985: \subsection{Comparing the results}
986: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
987: %
988: %
989: 
990: We showed that perturbation theory is quite successfull in describing (up to NNNLO)
991: lattice propagators in the momentum range $[2\,\text{GeV},6\,\text{GeV}]$, yielding
992: compatible values of $\Lqcd$. The separate fit of the ghost and gluon propagators,
993: and the fit of their ratio favours the existence of $\propto \frac{1}{p^2}$ power
994: corrections and validates the OPE approach in the case of ghost and gluon propagators.
995: The difference with previous approaches is that we have not introduced any additional fit parameter, and have only used
996: perturbation theory. Our method can also be used to calculate $\Lqcd$ from propagators alone 
997: in the unquenched case. In principle, it can be used to estimate the value of the 
998: $\langle A^2 \rangle$ condensate.
999: 
1000: The main limitation of the application of perturbation theory to lattice 
1001: Green functions in the accessible energy interval is the lack of asymptoticity
1002: and the truncation of the series. In fact, even the conversion formula 
1003: (\ref{conversion_asympt_a_une_boucle}) is not exact at considered momenta.
1004: We estimate the accuracy of our results at around $10\%$. It can be improved by
1005: performing the simulations at $\beta=6.6, 6.8$ on the lattices of sizes 
1006: $48^4,64^4$, respectively. The choice of parameters is motivated by
1007: the necessity to have the same physical volume of the lattice in order
1008: to control the finite-size effects.
1009: 
1010: