1: % Contribution MHD-d z "RG'2005" do: << J.Phys. A >>
2: %
3: \documentclass{iopart}
4: \usepackage[dvips]{graphicx}
5: \begin{document}
6:
7: \title[D-dimensional developed MHD turbulence: Double expansion model]
8: {D-dimensional developed MHD turbulence: \\ Double expansion model}
9:
10: \author{M Jurcisin \footnote{Present address: JINR Dubna, Moscow Region, Russia}
11: and M Stehlik}
12:
13: \address{Institute of Experimental Physics SAS, Ko\v{s}ice, Slovakia}
14: \eads{\mailto{jurcisin@thsun1.jinr.ru}, \mailto{stehlik@saske.sk}}
15:
16: \begin{abstract}
17: Developed magnetohydrodynamic turbulence near two dimensions $d$ up
18: to three dimensions has been investigated by means of
19: renormalization group approach and double expansion regularization.
20: A modification of standard minimal subtraction scheme has been used
21: to analyze the stability of the Kolmogorov scaling regime which is
22: governed by the renormalization group fixed point. The exact
23: analytical expressions have been obtained for the fixed points.
24: The continuation of the universal value of the inverse Prandtl number
25: $u=1.562$ determined at $d=2$ up to $d=3$ restores the value of $u=1.393$
26: which is known in the kinetic fixed point from usual $\epsilon$-expansion.
27: The magnetic stable fixed point has been calculated and its
28: stability region has been also examined.
29: This point losses stability: (1) below critical value of dimension
30: $d_c=2.36$ (independently on the $a$-parameter of a magnetic
31: forcing) and, (2) below the value of $a_c=0.146$ (independently on
32: the dimension).
33: \end{abstract}
34: \pacs{ 47.27.ef, 52.65.Kj, 11.10.Hi} %47.27.$-$i,
35: \submitto{\JPA}
36: %\maketitle
37:
38:
39: \section{Introduction}
40:
41: The renormalization group (RG) methods have been widely used to
42: the analysis of fully developed hydrodynamic (HD) turbulence
43: beginning from pioneering papers \cite{FoNeSt77,DomMar79} based on
44: \cite{MaSiRo73,BaJaWa76}. It gives possibility to reply upon some
45: principal questions, e.g., on the fundamental description of the
46: infrared (IR) scale invariance, as well as it is useful for
47: calculation of many quantities, e.g., critical dimensions of the
48: fields and their gradients, viscosity, etc. (see, e.g.
49: \cite{AdAnVa96-99,Vasiliev98-04,McComb90-95,Davidson04}).
50:
51: Then many authors begin to use Wilson's scheme or some adequate
52: generalized renormalization scheme to study of HD turbulence
53: \cite{YakOrs86-87} as well as of magnetohydrodynamic (MHD)
54: turbulence \cite{FoSuPo82,CamTas92}. This time Vasiliev's team
55: have used functional formulation of the field-theoretic RG
56: \cite{Vasiliev76,Zinn89} to legalize of the Kolmogorov scaling
57: regime of HD turbulence \cite{AdVaPi83,AdVaHn84}. They consider
58: (as used in present paper) the functional quantum field RG
59: approach \cite{AdAnVa96-99} rather then Wilson's RG technique
60: \cite{WilKog74}. It assigns a field action to the stochastic
61: problem and makes possible to use elegant and very well developed
62: RG procedure in quantum field theory to investigate infrared
63: asymptotic regimes of a stochastic system. Then this
64: RG method has been applied in MHD turbulence
65: \cite{AdVaHn85,AdVaHn87}. Note here that this functional RG method
66: allow a straightforward extension of the pertubative calculation
67: to an higher order loops without a principal difficulty (see
68: \cite{AdAnKoVa03,AdAnBa01}, for example).
69:
70: Considerable effort had been devoted to application of adequate field-theoretical
71: methods in the MHD turbulence (see recent review of Verma \cite{Verma04}, for example).
72: Authors in \cite{FoSuPo82,CamTas92} have used the 'classical' Yakhot-Orszag scheme
73: \cite{YakOrs86-87}.
74: In last years Verma \cite{Verma01a,Verma01b} performed detailed RG calculation of
75: MDH turbulence using McComb's alternative field-theoretic RG procedure
76: \cite{McComb90-95} and they reached notable progress in calculation of some
77: renormalized parameters of MHD turbulence.
78: Here we will not present full discussion of all methods used in the full developed
79: turbulence theory such as calculation of Alfven ratio, magnetic resistivity
80: \cite{Verma01b} or a problem of magnetic dynamo in helicit MHD \cite{AdVaHn87}
81: because it goes out of the frame of present paper (but see some remarks and
82: discussion in section 4 and section 5).
83: %Here we restrict ourselves to the remark that us used RG method fixes already
84: %only stable scaling regimes with a balanced kinetic and/or magnetic energy density
85: %in stationary turbulent regime due to consistent problem formulation.
86:
87: Present paper deals with an investigation of the existence and range of stability
88: of the 'magnetic' scaling regime (i.e. the magnetic fixed point for zero
89: inverse Prandtl number, see below) in the non-helical $d$-dimensional MHD turbulence.
90: %%%==
91: The existence of two different anomalous scaling regimes in three dimensions,
92: which are known as kinetic and magnetic ones, was established in the pioneering
93: papers \cite{FoSuPo82,AdVaHn85}. These two points correspond to two
94: IR stable fixed points of the RG. On the other hand, it was also
95: supposed that in two dimensions the magnetic fixed point does not
96: exist as a result of nonexistence of the IR stable magnetic fixed
97: point. But the conclusions about two dimensional fixed points
98: cannot be consider without doubts in these papers due to the
99: problems with renormalization in two dimensions which were not taken
100: into account \cite{HonNal96} (see also \cite{AdAnVa96-99}). In
101: \cite{Liang93,Kim99} two dimensional case was studied too but again
102: with shortcomings, therefore their results cannot be considered
103: completely conclusive. Within the our field theoretic RG approach the
104: problem is related to the existence of additional divergences which
105: arise in two dimensions.
106:
107: The first correct treatment of the two dimensional case of the
108: stochastically forced MHD equations with the proper account of these
109: additional divergences was done in \cite{HnHoJu01}. It was
110: accomplished within a two-parameter expansion (double expansion) of
111: scaling exponents and scaling functions \cite{HonNal96} where,
112: besides the parameter which characterize the deviation of the
113: exponent of the powerlike correlation function of random forcing
114: from its critical value, the additional parameter of the deviation
115: of the spatial dimension was introduced. The using of this double
116: expansion method has allowed them to confirm the basic conclusions
117: of the previous works \cite{FoSuPo82,AdVaHn85}, namely, the
118: nonexistence of the magnetic scaling regime near two dimensions.
119:
120: The authors of the paper \cite{HnHoJu01} also tried to restore the
121: stability of the magnetic fixed point when moving from two
122: dimensions in direction of three dimensions. This possibility was
123: achieved by using of the special choice of finite renormalization
124: which allowed them to keep track of the effect of the additional
125: divergences near two dimensions. Technically, it was done by
126: introducing of another uniform UV cutoff in all propagators which
127: does not affect the large-scale properties of the model. This setup
128: is similar to that of Polchinski \cite{Polchinski}. As a result, the
129: borderline dimension between stable and unstable magnetic fixed
130: point was found and it leads to the possibility of the uniform
131: description of two and three dimensional cases of stochastic MHD.
132:
133: Another possibility how to solve the problem of the additional
134: divergences in two dimensions together with the problem of
135: restoration of the stability of the corresponding fixed point when
136: going from a two dimensional system to a three dimensional one was
137: proposed in \cite{HnJoJuSt01}. They suggest to apply a modified
138: minimal substraction (MS) scheme in which the $d$-dependence of the
139: tensor structures of the UV divergent parts of the corresponding
140: diagrams are kept. It was successfully used in the fully developed
141: Navier-Stokes turbulence with weak uniaxial anisotropy to restore
142: the stability of the Kolmogorov scaling regime which is unstable in
143: two dimensions and stable in three dimensions.
144:
145: In what follows, we shall apply the double expansion method together
146: with modified MS scheme introduced in \cite{HnJoJuSt01} to the
147: stochastic MHD equations. Our aim is to investigate if it is
148: possible to describe correctly and uniformly the two dimensional and
149: the three dimensional systems and to compare our results with that
150: of \cite{HnHoJu01} where the different method was used (see above).
151: Thus, we carry out an analysis of the randomly forced MHD equations
152: with the proper account of the additional UV divergences which are
153: appeared in $d=2$. We apply a modified minimal subtraction scheme
154: based on the fact that the tensor structure of counter-terms is left
155: generally $d$-dependent in the calculations of divergent parts of
156: Green's functions. It will be shown that it allows us to investigate
157: behavior of the system under continual transition to $d=3$ beginning
158: from $d=2$. We have also confirmed the earlier conclusions made in
159: \cite{FoSuPo82,AdVaHn85,Verma01a}
160: that near two dimensions a scaling regime
161: driven by the velocity fluctuations may exist, but no magnetically
162: driven scaling regime can occur. We have also investigated the
163: long-range asymptotic behavior of the model in the double expansion
164: framework and found, in particular, that in this case thermal
165: fluctuations of the magnetic scaling regime may occur and that the
166: value of the borderline dimension is significantly lower
167: ($d_c=2.36$) than in the $\epsilon$ expansion \cite{FoSuPo82}
168: ($d_c=2.85$) and rather lower than in the 'modified' double
169: expansion introduced in \cite{HnHoJu01} ($d_c=2.46$)
170: but it is rather higher then value ($d_c=2.2$) calculated in the frame of
171: the McComb's renormalization \cite{Verma04}.
172: The discrepancy between the value of inverse Prandtl number $u$ which corresponds
173: to nontrivial stable fixed point of the RG in the three dimensions, which has been
174: obtained in the double expansion scheme in earlier paper \cite{HnHoSt94} and that
175: obtained by the usual $\epsilon$-expansion scheme \cite{FoSuPo82,AdVaHn85}
176: and also that obtained by Verma \cite{Verma01a,Verma01b} by McComb's procedure,
177: was one more reason of the present analysis.
178: Here we show
179: that the continuous transition from the universal value of the inverse Prandtl
180: number $u=1.562$ determined at $d=2$ restores the value of $u=1.393$ at $d=3$
181: which is known from usual $\epsilon$-expansion.
182:
183: The paper is organized as follows: In section~\ref{sec2} the functional
184: field theoretic formulation of the model is present in detail. In
185: section~\ref{sec3} the renormalization of the model is discussed. In
186: section~\ref{sec4} detailed analysis of the possible scaling regimes
187: is done. In section~\ref{sec5} conclusions and discussion of the
188: results are given.
189:
190:
191:
192: \section{Functional formulation of double expansion model}\label{sec2}
193:
194: In the present paper we study the universal statistical features of
195: the model of stochastic MHD which is described by the system of
196: equations for the fluctuating velocity field of an incompressible
197: conducting fluid $\bi{v}(x)$, $x\equiv (\bi{x},t)$,
198: $\bi{\nabla}\bi{\cdot}\bi{v} = 0$ and the magnetic induction ${\bf
199: B}=(\rho \mu)^{1/2} \bi{b}(x)$ (where $\rho$ is density of the fluid
200: and $\mu$ is its permeability) \cite{FoSuPo82,AdVaHn85,HnaSte91}:
201: \begin{eqnarray}
202: \partial_t \bi{v} + (\bi{v}\bi{\cdot}\bi{\nabla})\bi{v} -
203: (\bi{b}\bi{\cdot}\bi{\nabla})\bi{b} - \nu_0 \nabla^2 \bi{v} =
204: \bi{f^v} \,,
205: \label{NSR} \\
206: \partial_t \bi{b} + (\bi{v}\bi{\cdot}\bi{\nabla}) \bi{b} -
207: (\bi{b}\bi{\cdot}\bi{\nabla})\bi{v} -\nu_0 u_0\nabla^2\bi{b}
208: =\bi{f^b}\,, \label{magnet}
209: \end{eqnarray}
210: with the incompressibility conditions $\bf{\nabla}\bi{\cdot}\bi{f^v}
211: =0$ and $\bf{\nabla}\bi{\cdot}\bi{f^b} =0$ and the field $\bi{b}$ is
212: suppose to be solenoidal too, $\bi{\nabla}\bi{\cdot}\bi{b}=0$. The
213: statistics of $\bi{v}$, $\bi{b}$ is completely determined by both
214: the non-linear equations (\ref{NSR},\ref{magnet}) and the statistics
215: of the external inter-correlated large-scale random forces
216: $\bi{f^v}$, $\bi{f^b}$. The dissipation is controlled by the
217: parameter of kinematic viscosity $\nu_0$, and $u_0$ denotes inverse
218: Prandtl number (hereafter all parameters with a subscript $0$ denote
219: bare parameters of unrenormalized theory; see below). Note here that
220: the term $(\bi{b}\bi{\cdot}\bi{\nabla})\bi{b}$ expresses the
221: transverse part of Lorentzian force and it can be omitted in the
222: case of magnetic field treated as a passive admixture.
223:
224: As usual \cite{FoSuPo82,AdVaHn85}, statistical properties of the
225: Gaussian forcing with zero mean values
226: ($\langle\,\bi{f^v}\,\rangle=0, \,\, \langle\,\bi{f^b}\,\rangle=0$)
227: are determined by relations:
228: \begin{eqnarray}\fl\quad
229: \langle \,f_j^v(1) f_s^v(2)\,\rangle = \delta(\tau) u_0\,\nu_0^3\,
230: \int\frac{\rmd^d\bi{k}}{(2\pi)^d}\,{\rm P}_{js}(\bi{k})
231: \rme^{\rmi\bi{k.x}} \left[ g_{v10}\, k^{2- 2\delta-2\epsilon} +
232: g_{v20}\,k^2\,\right] \,,
233: \label{corelv} \\
234: \fl\quad \langle\, f_j^b(1) f_s^b(2)\,\rangle = \delta(\tau)
235: u_0^2\,\nu_0^3\, \int\frac{\rmd^d\bi{k}}{(2\pi)^d}\,{\rm
236: P}_{js}(\bi{k}) \rme^{\rmi\bi{k.x}} \left[ g_{b10}\, k^{2- 2\delta-2
237: a\epsilon} + g_{b20}\,k^2\,\right] \,, \label{corelb}
238: \end{eqnarray}
239: where the argument $1\equiv x_1$, $\tau=t_1 - t_2$,
240: $\bi{x}=\bi{x_1}-\bi{x_2}$, ${\rm P}_{js}(\bi{k})=\delta_{js}-k_j
241: k_s/k^2 $, the parameter $\epsilon$ determines the powerlike falloff
242: of the long-range forcing correlations, and the parameter
243: $\delta=(d-2)/2$ describe the deviation from spatial dimension
244: $d=2$. The free parameter $a$ controls the power form of magnetic
245: forcing. Note that parameters $\epsilon=2, a=1$ are the natural
246: "physical" values in our "massless" power-law energy injection. The
247: introduction of the local correlations (proportional to the new
248: couplings $g_{v20}$, and $g_{b20}$) which are described by the
249: analytic in $k^2$ terms in the correlation functions (\ref{corelv}),
250: and (\ref{corelb}) is related to the existence of additional
251: divergences of this structure (see below in the text) in the two
252: dimensional model which cannot be removed by corresponding nonlocal
253: terms \cite{HonNal96,Honkonen98,HnHoHoSe99}. At the same time, the
254: localness of the counterterms is the fundamental feature of a model
255: to be multiplicatively renormalizable \cite{Collins85,Zinn89}. For
256: example, it was not taken into account in the analysis of the model
257: in \cite{FoSuPo82,AdVaHn85}.
258:
259: Using the well-known Martin-Siggia-Rose formalism
260: \cite{MaSiRo73,BaJaWa76}, one can transform the stochastic problem
261: (\ref{NSR})-(\ref{magnet}) with correlators (\ref{corelv}), and
262: (\ref{corelb}) into the field theoretic model of the doubled set of
263: fields $\Phi\equiv\{\bi{v},\bi{b},\bi{v^{\prime}},\bi{b^{\prime}}\}$
264: with the following action functional
265: \begin{eqnarray}\fl
266: \quad S = {1\over2}\int\rmd x_1 \rmd x_2\,
267: \Big\{ v^{\prime}_j(1)\,\langle\,f_j^v(1)f_s^v(2)\,\rangle_0\,v^{\prime}_s(2)
268: + b^{\prime}_j(1)\,\langle\,f_j^b(1)f_s^b(2)\,\rangle_0\,b^{\prime}_s(2) \Big\}+
269: \nonumber\\
270: + \int\rmd x\ \bi{v^{\prime}} \bi{\cdot} \left( -\partial_t\bi{v} +
271: \nu_0\,\nabla^2\bi{v} -(\bi{v}\bi{\cdot}\bi{\nabla})\bi{v}
272: -(\bi{b}\bi{\cdot}\bi{\nabla})\bi{b} \right)
273: \nonumber\\
274: + \int\rmd x\ \bi{b^{\prime}}\bi{\cdot}\left(-\partial_t\bi{b} +u_0\,\nu_0\nabla^2\bi{b}+
275: (\bi{b}\bi{\cdot}\bi{\nabla})\bi{v}
276: -(\bi{v}\bi{\cdot}\bi{\nabla})\bi{b}\right)\,,
277: \label{action}
278: \end{eqnarray}
279: where $\bi{v^{\prime}}$, and $\bi{b^{\prime}}$ are independent of
280: $\bi{v}$, and $\bi{b}$ auxiliary incompressible fields, which we
281: have to introduce when transforming the stochastic problem into a
282: functional form.
283:
284: The dimensional constants $g_{v10}, g_{b10}, g_{v20}$, and
285: $g_{b20}$, which control the amount of randomly injected energy
286: given by (\ref{corelv}), (\ref{corelb}), play the role of the
287: coupling constants in the perturbative expansion.
288: %Their universal values have been determined after the parameters
289: %$\epsilon,\delta$ have been chosen to give the desired power form of
290: %forcing and desired dimension.
291: For the convenience of further calculations the factors $\nu_0^3
292: u_0$ and $\nu_0^3 u_0^2$ including the "bare" (molecular) viscosity
293: $\nu_0$ and the "bare" (molecular or microscopic) magnetic inverse
294: Prandtl number $u_0$ have been extracted. As was mentioned already
295: the bare (non-renormalized) quantities are denoted by subscript "0".
296:
297: The most important measurable quantities in the study of a fully
298: developed turbulence and related problems are considered to be the
299: statistical objects represented by correlation and response
300: functions (Green functions) of the fields. Standardly, the
301: formulation through the action functional (\ref{action}) replaces
302: the statistical averages of random quantities in the stochastic
303: problem (\ref{NSR})-(\ref{corelb}) with equivalent functional
304: averages with weight $\exp S(\Phi)$. Generating functionals of total
305: Green functions G(A) and connected Green functions W(A) are then
306: defined by the functional integral
307: \begin{equation}
308: G(A)=e^{W(A)}=\int {\cal D}\Phi \,\, e^{S(\Phi) +
309: A\Phi},\label{green}
310: \end{equation}
311: where $A(x)=\{{\bf A^{v}}, {\bf A^{b}}, {\bf A^{v^\prime}},{\bf
312: A^{b^{\prime}}}\}$ represents a set of arbitrary sources for the set
313: of fields $\Phi$, ${\cal D}\Phi \equiv {\cal D}\theta{\cal
314: D}\theta^{\prime}{\cal D}{\bf v}{\cal D}{\bf v^{\prime}}$ denotes
315: the measure of functional integration, and linear form $A\Phi$ is
316: defined as
317: \begin{equation}
318: \hspace{-2cm} A \Phi= \int d\,x [{\bf A^{v}}(x)\cdot {\bi v}(x)+
319: {\bf A^{b}}(x)\cdot {\bi b}(x) + {\bf A^{v^{\prime}}}(x)\cdot {\bi
320: v^{\prime}}(x) + {\bf A^{b^{\prime}}}(x)\cdot {\bi
321: b^{\prime}}(x)].\label{form}
322: \end{equation}
323:
324: The functional formulation gives the possibility of using the field
325: theoretic methods, including the RG technique to solve the problem.
326: By means of the RG approach it is possible to extract large-scale
327: asymptotic behavior of the correlation functions after an
328: appropriate renormalization procedure which is needed to remove UV
329: divergences. The functional formulation is advantageous also because
330: the Green functions of the Fourier-decomposed stochastic MHD can be
331: calculated by means of Feynman diagrammatic technique.
332:
333: Action (\ref{action}) is given in a form convenient for a
334: realization of the field theoretic perturbation analysis with the
335: standard Feynman diagrammatic technique. Free (bare) propagators
336: ${\hat\Delta}$ can be found from the quadratic part of the action
337: (\ref{action}) written in the form $\,\,-(1/2)\Phi\hat{{\cal
338: K}}\Phi$ and by using the definition $\,\,\hat{{\cal
339: K}}{\hat\Delta}=\hat1$, where $\hat1$ denotes the diagonal matrix
340: whose diagonal elements are the transverse projectors (our fields
341: are solenoidal). One obtains
342: \begin{eqnarray}
343: {\hat\Delta}_{js} &=&
344: \left(
345: \begin{array}{cccc}
346: \Delta^{vv}_{js} & 0 & \Delta^{vv'}_{js} & 0 \\
347: 0 & \Delta^{bb}_{js} & 0 & \Delta^{bb'}_{js} \\
348: \Delta^{v'v}_{js} & 0 & 0 & 0 \\
349: 0 & \Delta^{b'b}_{js} & 0 & 0
350: \end{array} \right)
351: \end{eqnarray}
352: with the elements (wave-number-frequency representation)
353: \begin{eqnarray}
354: \Delta^{vv'}_{js}({\bf k},\omega)&=& \Delta^{v^{\prime}v}_{js}(-{\bf
355: k},-\omega) =\frac{{\rm P}_{js}({\bf k})}{-i\omega+\nu_0\, k^2}\,,
356: \nonumber\\
357: \Delta^{bb'}_{js}({\bf k},\omega)&=& \Delta^{b^{\prime}b}_{js}(-{\bf
358: k},-\omega) =\frac{{\rm P}_{js}({\bf k})}{-i\omega+u_0\,\nu_0\,
359: k^2}\,,
360: \nonumber\\
361: \Delta^{vv}_{js}({\bf k},\omega) &=&u_0\,\nu_0^3\,k^2\
362: \frac{g_{v10}\,k^{-2\delta-2\epsilon}+g_{v20}}
363: {|-i\omega+\nu_0\, k^2|^2}\,{\rm P}_{js}({\bf k})\,,
364: \nonumber\\
365: \Delta^{bb}_{js}({\bf k},\omega)&=&u_0^2\,\nu_0^3\,k^2 \
366: \frac{g_{b10}\,k^{-2\,a\delta-2\epsilon}+ g_{b20}}
367: {|-i\omega+ \,u_0\,\nu_0\,k^2|^2}\,{\rm P}_{js}({\bf k}) \, .
368: \label{prop}
369: \end{eqnarray}
370:
371: The model has three triple (interaction) vertices
372: \begin{eqnarray}
373: - \bi{v^{\prime}}(\bi{v}\bi{\cdot}\bi{\nabla})\bi{v}&=&v^{\prime}_j
374: V_{jkl} v_k v_l\,, \\ -
375: \bi{v^{\prime}}(\bi{b}\bi{\cdot}\bi{\nabla})\bi{b}&=&v^{\prime}_j
376: V_{jkl} b_k b_l\,, \\
377: \bi{b^{\prime}}[(\bi{b}\bi{\cdot}\bi{\nabla})\bi{v}-
378: (\bi{v}\bi{\cdot}\bi{\nabla})\bi{b} ]&=&b^{\prime}_j U_{jkl} b_k
379: v_l\,,
380: \end{eqnarray}
381: where the tensor structure of the vertices in wave-number-frequency
382: representation are
383: \begin{equation}
384: V_{jkl} = i(\delta_{jk} p_l +\delta_{jl} p_k)\,,\,\,\,
385: U_{jkl}=i(\delta_{jl} p_k -\delta_{jk} p_l),
386: \end{equation}
387: where momentum ${\bf p}$ is flowing into the vertex via the
388: auxiliary fields $\bi{v^{\prime}}$, and $\bi{b^{\prime}}$.
389:
390:
391:
392:
393: \section{Renormalization}\label{sec3}
394: \subsection{Divergences of the model}
395:
396: It can be shown \cite{AdVaHn85} that for any fixed space dimension
397: $d>2$, the superficial UV divergences can exit only in the following
398: one-particle irreducible (1PI) Green functions:
399: $\Gamma^{vv^{\prime}}, \Gamma^{bb^{\prime}}$, and $\Gamma^{v'bb}$.
400: They lead to local counterterms of the form $\propto
401: \bi{v^{\prime}}\nabla^2\bi{v}$,
402: $\propto\bi{b^{\prime}}\nabla^2\bi{b}$, and
403: $\propto\bi{v^{\prime}}(\bi{b}\bi{\cdot}\bi{\nabla})\bi{b}$ which
404: are already present in the action (\ref{action}), therefore, the
405: model is multiplicatively renormalizable (the analytic terms in
406: $k^2$ proportional to $g_{v20}$, and $g_{v20}$ in (\ref{corelv}),
407: and (\ref{corelb}) are not needed in this case, and the model can be
408: formulated without them).
409:
410: The situation is more complicated in the two dimensional case, where
411: additional UV divergences appear. They are related to the 1PI Green
412: functions $\Gamma^{v^{\prime}v^{\prime}}$, and
413: $\Gamma^{b^{\prime}b^{\prime}}$. In this situation the formulation
414: of the model without local (analytic in $k^2$) terms cannot give, in
415: general, multiplicatively renormalizable model because the nonlocal
416: terms of the action is not renormalized since the divergences
417: produced by the loop integrals of the diagrams are always local in
418: space and time (see, e.g., \cite{Zinn89}). Thus, the simplest way
419: how to restore the renormalizability of the model (or how to include
420: the corresponding local counterterms $\propto
421: \bi{v^{\prime}}\nabla^2\bi{v^{\prime}}$, and
422: $\propto\bi{b^{\prime}}\nabla^2\bi{b^{\prime}}$ in the
423: renormalization) is to add corresponding local terms to the force
424: correlation functions. It is shown explicitly in (\ref{corelv}), and
425: (\ref{corelb}). In language of classical hydrodynamics the forcing
426: contribution $\propto k^2$ corresponds to the appearance of large
427: eddies convected by small and active ones and it is represented by
428: the local term of $\bi{v^{\prime}}\nabla^2\bi{v^{\prime}}$. In its
429: analogy the term $\bi{b'}\nabla^2\bi{b'}$ is added to the magnetic
430: forcing.
431:
432: Thus, in two dimensions, the model (\ref{action}) is renormalizable
433: by the standard power-counting rules, and for limits
434: $\epsilon\rightarrow 0, \ \delta\rightarrow 0$ possesses the
435: ultraviolet (UV) divergences which are present in five
436: aforementioned 1PI Green functions. It means that model is
437: regularized using a combination of analytic and dimensional
438: regularization with the parameters $\epsilon$, and $\delta=(d-2)/2$.
439: As a result, the UV divergences appear as poles in $\epsilon$,
440: $\delta$, and their following combinations: $2\epsilon +\delta$, and
441: $(a+1)\epsilon + \delta$. The UV divergences may be removed by
442: adding needed counterterms to the basic action $S_B$ which is
443: obtained from the unrenormalized one (\ref{action}) by the
444: substitution of the renormalized parameters for the bare ones:
445: $g_{v10} \rightarrow \mu^{2\epsilon} g_{v1}$, $g_{v20} \rightarrow
446: \mu^{-2 \delta} g_{v2}$, $g_{b10} \rightarrow \mu^{2 a \epsilon }
447: g_{b1}$, $g_{b20} \rightarrow \mu^{-2\delta } g_{b2}$, $\nu_0
448: \rightarrow \nu$, $u_0 \rightarrow u$, where $\mu$ is a
449: scale-setting parameter having the same canonical dimension as the
450: wave number.
451:
452: In what follows, we shall work with, in our case the most
453: convenient, minimal subtraction (MS) scheme, i.e., we are
454: interesting only in the singular (pole) parts of divergent 1PI Green
455: functions which are included in the renormalization constants. They
456: give rise to the counterterms added to the basic action to make the
457: Green functions of the renormalized model UV finite. In our model,
458: the counterterms have the form
459: \begin{eqnarray}\fl
460: \qquad S_{count}=&\int\,\rmd x\, \big[ \nu\, \left(1- Z_1\right)
461: \bi{v}'\nabla^2\, \bi{v}+
462: u \nu\,\left(1- Z_2\right) \bi{b}'\nabla^2\, \bi{b}
463: \nonumber\\
464: \fl &+\frac{1}{2}\,(\,Z_4-1) u {\nu^3} g_{v2}\,{\mu}^{-2\delta}\,
465: \bi{v}'{\nabla^2} \bi{v}' +\frac{1}{2}\, ( Z_5-1 )\,u^2 {\nu^3}
466: g_{b2}\,{\mu}^{-2\delta}\, \bi{b}'{\nabla^2} \bi{b}'
467: \nonumber\\
468: \fl &+(1-Z_3)\,\bi{v}'(\bi{b\cdot\nabla})\bi{b} \big]\,,
469: \label{counter}
470: \end{eqnarray}
471: where the renormalization constants $Z_i, i=1,2,4,5$ renormalizing
472: the unrenormalized parameters
473: $e_0=\{g_{v10},g_{v20},g_{b10},g_{b20},\nu_0,u_0\}$, and the
474: renormalization constant $Z_3$ renormalizing the fields ${\bi b}$,
475: and ${\bi b^{\prime}}$. They are chosen to cancel the UV divergences
476: appearing in the Grren functions constructed using the basic action.
477: The remaining fields $\bi{v^{\prime}}$, and ${\bi v}$ are not
478: renormalized due to the Galilean invariance of the model
479: (\ref{action}).
480:
481: Renormalized Green functions are expressed in terms of the
482: renormalized pa\-ra\-me\-ters
483: \begin{eqnarray}
484: g_{v1} &= g_{v10}\,{\mu}^{-2\epsilon}\, Z_1^2 Z_2,\qquad &
485: g_{v2} = g_{v20}\,{\mu}^{2\delta}\, Z_1^2 Z_2 Z_4^{-1},
486: \nonumber\\
487: g_{b1} &= g_{b10}\,{\mu}^{-2\,a\epsilon}\,Z_1 Z_2^{2} Z_3^{-1},\qquad &
488: g_{b2} = g_{b20}\,{\mu}^{2\delta}\, Z_1 Z_2^{2} Z_3^{-1} Z_5^{-1},
489: \label{Zkaka1}\\
490: \nu &= \nu_0\, Z_{1}^{-1}, \qquad &
491: u\ =\ u_0\, Z_2^{-1} Z_1
492: \nonumber
493: \end{eqnarray}
494: appearing in the renormalized action $S^R$ connected with the action
495: (\ref{action}) by the relation of multiplicative renormalization:
496: $ S^R \{\bi{e}\} = S \{\bi{e_0}\}$ . The renormalized action $S^{R}$,
497: which depends on the renormalized parameters $e(\mu)$, yields
498: renormalized Green functions without UV diver\-gences.
499: The RG is mainly concerned with the prediction of the asymptotic
500: behavior of correlation functions expressed in terms of anomalous
501: dimensions $\gamma_j$ by the use of $\beta$ functions, both defined
502: via differential relations
503: \begin{equation}\fl
504: \qquad \gamma_j=\mu \frac{\partial \ln Z_j}{\partial \mu}\Big|_{e_0}\,,\,\,
505: \,\,\,
506: \beta_g=\mu \frac{\partial g}{\partial\mu }\Big|_{e_0}\,,\qquad\mbox{with}
507: \,\,\,
508: g\equiv\{g_{v1},g_{v2},g_{b1},g_{b2},u\}\,.
509: \label{defbeta}
510: \end{equation}
511: These definitions with expressions (\ref{Zkaka1}) yield the $\gamma$-functions
512: \begin{eqnarray}
513: \gamma_{gv1}&= -2\gamma_1-\gamma_2\,,
514: \qquad& \gamma_{gb1}= -\gamma_1-2\gamma_2+\gamma_3)\,,
515: \nonumber\\
516: \gamma_{gv2}&= -2\gamma_1-\gamma_2+\gamma_4\,,
517: \qquad& \gamma_{gb2}= -\gamma_1-2\gamma_2+\gamma_3+\gamma_5\,,
518: \label{gamy1}\\
519: \gamma_\nu &= \gamma_1\,, \qquad \gamma_b = {1\over2} \gamma_3\,,
520: \qquad &\ \ \gamma_u= -\gamma_1+\gamma_2)
521: \nonumber
522: \end{eqnarray}
523: and then $\beta$-functions
524: \begin{eqnarray}
525: \fl\qquad\ \beta_{gv1}&=g_{v1}\,(-2\epsilon+2\gamma_1+\gamma_2)\,,
526: \quad&
527: \beta_{gb1}=g_{b1}\,(-2\,a\epsilon+\gamma_1+2\gamma_2-\gamma_3)\,,\,\,
528: \nonumber \\
529: \fl\qquad\ \beta_{gv2}&=g_{v2}\,(2\delta+2\gamma_1+\gamma_2-\gamma_4)\,,
530: \quad&
531: \beta_{gb2}=g_{b2}\,(2\delta+\gamma_1+2\gamma_2-\gamma_3-\gamma_5)
532: \label{bety} \\
533: \fl\qquad\ \ \ \beta_u&= u\,(\gamma_1-\gamma_2).
534: \nonumber
535: \end{eqnarray}
536:
537:
538: \subsection{RG equations}
539:
540: Correlation functions of the fields are expressed in terms of
541: scaling functions of the variable $s=(k/\mu)$,
542: $s\in\langle0,1\rangle$. Then the asymptotic behaviour and the
543: universality of MHD statistics stem from the existence of a stable
544: IR fixed point. The continuous RG transformation is an operation
545: linking a sequence of invariant parameters ${\overline g}(s)$
546: determined by the Gell-Mann-Low equations
547: \begin{equation}\fl
548: \qquad \frac{d\overline g(s)}{d\ln s}= \beta_g\left({\overline g}(s)\right)\,\,\,
549: \,\mbox{with the abbrevation}\,\,\,{\overline g}\equiv \{ {\overline g_{v1}},
550: {\overline g_{v2}}, {\overline g_{b1}}, {\overline g_{b2}}, {\overline u} \}\,,
551: \label{Gell}
552: \end{equation}
553: where the scaling variable $s$ parameterizes RG flow with the
554: initial conditions ${\overline g}|_{s=1}\equiv g$ (the critical
555: behaviour corresponds to IR limit $s\rightarrow 0$). The expression
556: of the $\beta({\overline g}(s))$ function is known in the framework
557: of the $\delta$, $\epsilon$ expansion (see (\ref{gama}) and also
558: (\ref{bety})). The fixed point $g^{\ast}(s\rightarrow 0)$ satisfies
559: a system of equations $\beta_{g}(g^{\ast})=0, $ while a IR stable
560: fixed point, weakly dependent on initial conditions, is defined by
561: positive definiteness of the real part of the matrix $\Omega=
562: (\partial \beta_g/\partial g)|_{g^{\ast}}$ (the matrix of the first
563: derivatives taken at the fixed point). In other words, a fixed point
564: is stable if all the trajectories $g(s)$ in its vicinity approach
565: the value of the fixed point.
566:
567: The initial conditions ${\overline g}|_{s\rightarrow 1}=g$ of the
568: equations (\ref{Gell}), dictated by a micromodel, are insufficient
569: since our aim is the large-scale limit of statistical theory, where
570: $g^{\ast}\equiv {\overline g}|_{s\rightarrow 0}$. As was mentioned
571: already, the RG fixed point is defined by the equation
572: \begin{equation}\beta(\,g^{\ast}\,)= 0\,.\label{beta01}\end{equation}
573: For ${\overline g}(s)$ close to $g^{\ast}$ we obtain a system
574: of linearized equations
575: \begin{equation}
576: \left( I\,s\,\frac{\mbox{d}}{\mbox{d} s} - \Omega\,
577: \right) \,({\overline g} - g^{\ast})\,= 0,
578: \end{equation}
579: where $I$ is $(5\times 5)$ unit matrix. Solutions of this system
580: behave like ${\overline g}=g^{\ast}+{\cal O}( s^{{\xi}_j}) $
581: if $s\rightarrow 0$.
582: The exponents $\xi_j$ are the elements of the diagonalized
583: matrix $\Omega^{diag} =
584: (\,\xi_1,\,\xi_2,\,\xi_3,\,\xi_4, \xi_5 \,)$
585: and can be obtained as roots of the characteristic polynomial
586: ${\mbox{\small Det}}(\Omega - \xi I).$
587: The positive defineteness of $\Omega$ represented by the conditions
588: $\mbox{Re}_j(\xi)\geq 0, j=1,2,...5$ is the test of the IR
589: asymptotical stability of discussed theory.
590:
591:
592: \subsection{One-loop order calculation}
593:
594: In the standard MS scheme \cite{Hoft73} the renormalization constants have the general form
595: \begin{equation}
596: Z_i=1-F_i P^{\delta,\epsilon},
597: \end{equation}
598: where the terms $P^{\delta,\epsilon}$ are given by the linear combinations of the poles
599: and the amplitudes $F_i$ are some functions of $g_{v1},g_{v2},g_{b1},g_{b2}$, and $u$,
600: but are independent of $\delta$ and $\epsilon$. The amplitudes $F_i=F_i^{(1)}
601: F_i^{(2)}$ are a product of two multipliers $F_i^{(1)}, F^{(2)}_i.$
602: One of them, say, $F_i^{(1)}$ is a multiplier originating from the
603: divergent part of the Feynman diagrams, and the second one,
604: $F_i^{(2)}$ is connected only with the tensor nature of the
605: diagrams (see discussion in \cite{HnJoJuSt01} for details).
606:
607: It can be explained by the following simple example \cite{HnJoJuSt01}
608: (the example is taken from a problem with anisotropy, i.e., where another
609: arbitrary unit vector ${\bf n}$ exists but the conclusions are the same).
610: Consider a UV-divergent integral
611: $$ I({\bf k}, {\bf n}) \equiv n_i n_j k_l k_m\int d^d {\bf q}
612: \frac{1}{(q^2+m^2)^{1+2\delta}} (\frac{q_iq_jq_lq_m}{q^4}
613: -\frac{\delta_{ij}q_lq_m+\delta_{il}q_jq_m+\delta_{jl}q_iq_m}{3q^2})$$
614: (summations over repeated indices are implied) where $m$ is an
615: infrared mass. It can be simplified in the following way: $$ I({\bf
616: k}, {\bf n}) \equiv n_i n_j k_l k_m S_{ijlm}\int_0^{\infty} d q^2
617: \frac{q^{2\delta}}{2(q^2+m^2)^{1+2\delta}},$$ where
618: $$S_{ijlm}=\frac{S_d}{d(d+2)}(\delta_{ij}\delta_{lm}+\delta_{il}\delta_{jm}+
619: \delta_{im}\delta_{jl}-\frac{(d+2)}{3}(\delta_{ij}\delta_{lm}+
620: \delta_{il}\delta_{jm}+\delta_{im}\delta_{jl})),$$
621: $$\int_0^{\infty} d q^2
622: \frac{q^{2\delta}}{2(q^2+m^2)^{1+2\delta}}=\frac{\Gamma{(\delta
623: +1)}\Gamma{(\delta)}}{2 m^{2\delta}\Gamma(2\delta+1)}, $$ and
624: $S_d=2\pi^{d/2}/\Gamma(d/2)$ is the surface of unit the
625: d-dimensional sphere. The purely UV divergent part manifests itself
626: as the pole in $2\delta=d-2$; therefore, we find $$\mbox{UV div.
627: part of}\,\,\,\, I= \frac{1}{2\delta}(F^{(2)}_1 k^2+F^{(2)}_2 ({\bf
628: n k})^2),$$ where $F^{(2)}_1=F^{(2)}_2/2 = (1-d)S_d/3d(d+2)$
629: ($F_1^{(1)}=F_2^{(1)}=1$).
630: It has to be mentioned that in spite of the above simple example in our
631: calculation we shall introduce the needed IR regularization by restriction
632: on interval of integrations.
633:
634: In the standard MS scheme one puts $d=2$ in $F^{(2)}_1,F^{(2)}_2$,
635: therefore the d-dependence of these multipliers is ignored. As was
636: discussed in \cite{HnJoJuSt01}, for the theories with vector fields
637: and, consequently, with tensor diagrams, where the sign of values of
638: fixed points and/or their stability depend on the dimension $d$, the
639: procedure, which eliminates the dependence of multipliers of the
640: type $F^{(2)}_1,F^{(2)}_2$ on $d,$ is not completely correct because
641: one is not able to control the stability of the fixed point when
642: $d=3.$ Therefore, in \cite{HnJoJuSt01} it was proposed to slightly
643: modify the MS scheme in such a way to keep the d-dependence of $F$
644: in renormalization constants $Z_i$. Then the subsequent calculations
645: of the RG functions ($\beta-$ functions and anomalous dimensions
646: $\gamma_i$) allow one to arrive at the results which are in
647: qualitative agreement with the results obtained in the framework of
648: the simple analytical regularization scheme, i.e., one is able to
649: obtain the fixed point which is not stable for $d=2$, but whose
650: stability is restored for a borderline dimension $2<d_c<3$. In what
651: follows, it will be shown that it is really our case, thus we shall
652: apply this modified MS scheme in our calculations.
653:
654: Now we can return and continue with RG analysis. Using the RG
655: routine the anomalous dimensions
656: $\gamma_j(g_{v1},g_{v2},g_{b1},g_{b2})$ can be extracted from
657: one-loop diagrams. Thus, the extraction of the UV-divergent parts
658: from one-loop diagrams gives $Z$-constants in the form
659: \begin{eqnarray}
660: \fl\quad Z_1&=&1+\frac{S_d}{(2\pi)^d} \left[ \,u\,\lambda_5 \left(
661: \frac{ g_{v2}}{2\delta}-\frac{ g_{v1}}{2\epsilon}\right) + \lambda_6 \left(
662: \frac{ g_{b2}}{2\delta}-\frac{ g_{b1}}{2 a\epsilon}\right) \right]\,,
663: \nonumber\\
664: \fl\quad Z_2&=&1+\frac{S_d}{(2\pi)^d (u+1)} \left[\lambda_1 \left(
665: \frac{ g_{v2}}{2\delta}-\frac{ g_{v1}}{2\epsilon}\right) +
666: \lambda_3 \left(\frac{ g_{b2}}{2\delta}-\frac{ g_{b1}}{2 a \epsilon} \right)\right]\,,
667: \nonumber\\
668: \fl\quad Z_3&=&1+\frac{S_d}{(2\pi)^d} \,\lambda_7 \left(
669: \frac{ g_{v1}}{2\epsilon}-\frac{ g_{v2}}{2\delta} -
670: \frac{ g_{b1}}{2a\epsilon}+\frac{ g_{b2}}{2\delta}\right)\,,
671: \label{zetka}\\
672: \fl\quad Z_4&=&1+\frac{S_d}{(2\pi)^d} \frac{\lambda_4}{g_{v2}}\left(
673: \frac{u g_{v1}^2}{2\delta+4\epsilon}+\frac{2u g_{v1}g_{v2}}{2\epsilon}
674: -\frac{u g_{v2}^2}{2\delta} +
675: \frac{g_{b1}^2}{2\delta+4a\epsilon}+\frac{2 g_{b1}g_{b2}}{2a\epsilon}
676: -\frac{g_{b2}^2}{2\delta} \right)\,,
677: \nonumber\\
678: \fl\quad Z_5&=&1+\frac{S_d}{(2\pi)^d} \frac{\lambda_2}{(u+1)g_{b2}} \left(
679: \frac{g_{v1}\,g_{b1}}{2\delta+2\epsilon(1+a)}+
680: \frac{g_{v1}\,g_{b2}}{2\epsilon}+\frac{g_{v2}\,g_{b1}}{2 a\epsilon}
681: -\frac{g_{v2}\,g_{b2}}{2\delta} \right) \,,
682: \nonumber
683: \end{eqnarray}
684: and, in consequence, the lowest order $\gamma$-functions are
685: \begin{eqnarray}
686: \gamma_1&=&\widetilde{S_d}\,\left( u\,\lambda_5\,g_v + \lambda_6\,g_b \right)\,,
687: \quad\ \,\gamma_2=\widetilde{S_d}\,\frac{(\lambda_1\,g_v + \lambda_3\,g_b)}{u+1}\,,
688: \nonumber\\
689: \gamma_3&=&\widetilde{S_d}\,\lambda_7\, ( -g_v + g_b )\,,
690: \qquad\quad\,\gamma_4=\widetilde{S_d}\,\frac{\lambda_4}{g_{v2}}( u\,g_v^2 + g_b^2 )\,,
691: \label{gama}\\
692: \gamma_5&=&\widetilde{S_d}\,\frac{\lambda_2}{(1+u)}\frac{g_v\ g_b}{g_{b2}}\,,
693: \nonumber
694: \end{eqnarray}
695: where $\widetilde{S_d}=S_d/(2\pi)^d$, $S_d$ denote $d$-dimensional sphere
696: $S_d=2\pi^{d/2}/\Gamma(d/2)$, $g_v\equiv g_{v1}+g_{v2}$, $g_b\equiv g_{b1}+g_{b2}$,
697: and $\lambda$-coefficients depend only on dimension $d$:
698: \begin{eqnarray}\fl\qquad
699: \lambda_1&=\frac{d-1}{2d}\,, \quad\quad\ \lambda_2=\frac{d-2}{2d}\,, \quad &
700: \lambda_3=\frac{d-3}{2d}\,, \quad\ \ \lambda_4=\frac{d^2-2}{4d(d+2)}\,,
701: \nonumber \\
702: \fl\qquad
703: \lambda_5&=\frac{d-1}{4(d+2)}\,, \quad \lambda_6=\frac{d^2+d-4}{4d(d+2)}\,,\quad &
704: \lambda_7=\frac{1}{d(d+2)}\,.
705: \label{lambdy}
706: \end{eqnarray}
707: Substituting (\ref{gama}) into $\beta$-functions (\ref{bety}) one can
708: obtains $\beta$-functions in the one-loop order approximation.
709: Note that in two dimensions the $\gamma$-functions are
710: \begin{eqnarray}\fl\qquad
711: \gamma_1^{(2)}&=&{1\over32\pi} \left( u\,g_v + g_b \right)\,,\quad\,
712: \gamma_2^{(2)} = {1\over8\pi} \frac{(g_v - g_b)}{(u+1)}\,, \quad\,
713: \gamma_3^{(2)} = {1\over16\pi}\, (-g_v + g_b)\,,
714: \nonumber\\ \fl\qquad
715: \gamma_4^{(2)}&=&{1\over32\pi} \frac{( u\,g_v^2 + g_b^2 )}{g_{v2}}\,,\quad\,
716: \gamma_5^{(2)} = 0
717: \label{gamaD2}
718: \end{eqnarray}
719: and, in correspondence with \cite{HnHoJu01} $Z_5=1$, which is a specific property
720: of the two-dimensional MHD turbulence because there are no UV divergences in
721: the 1PI Green's function $\Gamma^{b'b'}$ in the one-loop approximation. Here we
722: emphasize that in general case of $d$ dimensions $\gamma_5\neq 0$ and $Z_5\neq 1$.
723:
724:
725:
726: \section{Fixed points}\label{sec4}
727: \subsection{Case of passive vector admixture}
728:
729: Here we briefly consider the case when the magnetic field can be
730: treated as a passive vector field in the developed HD turbulence.
731: Notation the "passive" magnetic field means that the Lorentz force
732: acting on conductive fluid can be neglected at large spatial scales,
733: thus, the Lorentzian term $(\bf{b\cdot\nabla})\bf{b}$ in the
734: Navier-Stokes equation can be omitted. Just then the vertex function
735: $\Gamma^{v'bb}$ is finite and the term containing $Z_3$ in
736: $S_{count}$ does not exists. Therefore, the magnetic field is not
737: renormalized and $\gamma_3=0$. Furthermore, some diagrams of
738: $\Gamma^{v'v}, \Gamma^{v'v'}, \Gamma^{b'b}$ containing the vertex
739: $\Gamma^{v'bb}$ can be omitted and $Z$-constants as well as
740: $\gamma$-functions are reduced. Resulting $\gamma$-functions take
741: the form
742: \begin{eqnarray}
743: \gamma_1 &=& \widetilde{S_d}\, u\,\lambda_5\,g_v \,,\quad\ \,
744: \gamma_2\ =\ \widetilde{S_d}\,\lambda_1\,\frac{g_v}{u+1}\,,
745: \nonumber\\
746: \gamma_4 &=& \widetilde{S_d}\,\lambda_4\,\frac{u}{g_{v2}}\,g_v^2 \,,\quad\,
747: \gamma_5\ =\ \widetilde{S_d}\,\frac{\lambda_2}{(1+u)}\frac{g_v\ g_b}{g_{b2}} \, .
748: \label{gamabez}
749: \end{eqnarray}
750: Substituting of $\gamma$-functions (\ref{gamabez}) and $\gamma_3=0$ into
751: $\beta$-equations (\ref{bety}) one obtains a system of fourth nonlinear equations
752: $\beta_{gv1}=\beta_{gv2}=\beta_{gb1}=\beta_{gb2}=0$ for $g_i$ and
753: one equation $\beta_u=0$ for $u$. The last one gives $u^\ast=0$, or, nonzero
754: universal inverse Prandtl number,
755: \begin{equation}
756: u^{\ast}={1\over2}\left( \sqrt{\frac{16+9 d}{d}}-1\right) \,.
757: \label{uu}
758: \end{equation}
759: In the first case of $u^\ast=0$ one obtains only two fixed points (with zeroth
760: $g_{b1}^\ast, g_{b2}^\ast$):\\
761: 1. $g_{v1}^\ast=0, g_{v2}^\ast=-2\delta/\lambda_1 \widetilde{S_d}\ $
762: which is non-physical (negative), and, \\
763: 2. $g_{v1}^\ast=2\epsilon/\lambda_1 \widetilde{S_d}, g_{v2}^\ast=0\ $ which in unstable.
764:
765: Let $u$ is given by (\ref{uu}). Then apart from the Gaussian fixed
766: point $g_{v1}^\ast=g_{v2}^\ast=g_{b1}^\ast=g_{b2}^\ast=0$, with no
767: fluctuation effect on the large-scale asymptotics, there are
768: following fixed points with $g_{b2}^\ast=0$:
769: \begin{eqnarray*}
770: \fl\qquad (1^\ast)\qquad g_{v1}^\ast=0,\quad
771: g_{v2}^\ast=-\frac{2(d-2)d^2(u^\ast+1)}{2d^2-3d+2}\widetilde{S_d}^{-1},\quad
772: g_{b1}^\ast=0;
773: \\
774: \fl\qquad (2^\ast)\qquad
775: g_{v1}^\ast=\frac{4\epsilon d(u^\ast+1)}{3(d-1)}\widetilde{S_d}^{-1},\quad
776: g_{v2}^\ast=0,\quad g_{b1}^\ast=0;
777: \\
778: \fl\qquad (3^\ast)\qquad
779: g_{v1}^\ast=\frac{4\epsilon(3d^3+d^2(4\epsilon-9)-6d(\epsilon-1)+4\epsilon)(u^\ast+1)}
780: {9(d+2\epsilon-2)(d-1)^2}\widetilde{S_d}^{-1},
781: \\
782: \fl\qquad\qquad\qquad g_{v2}^\ast=\frac{8\epsilon^2(d^2-2)(u^\ast+1)}
783: {9(d+2\epsilon-2)(d-1)^2}\widetilde{S_d}^{-1},\quad g_{b1}^\ast=0.
784: \end{eqnarray*}
785: Next three fixed points are the same as the last $(1^\ast)$--$(3^\ast)$
786: with different $g_{b2}^\ast$:\\
787: $(1a^\ast)$ $\ \ g_{b2}^\ast=(d^2-2)/d(d-2)$; \\
788: $(2a^\ast)\equiv(3a^\ast)$ $\ \ g_{b2}^\ast=3(d-1)(d+2\epsilon-2/2(d-2)\epsilon$.\\
789: The points $(2a^\ast)$ and $(3a^\ast)$ have the same $g_{b2}^\ast$ because
790: $g_{v1}^\ast$ of the point $(2^\ast)$ is equal to the sum $(g_{v1}^\ast + g_{v2}^\ast)$
791: of the point $(3^\ast)$. Note that $g_{b2}^\ast$ has discontinuity at $d=2$.
792:
793: The "thermal" point $(1^\ast)$ is generated by short-range correlations of the random
794: force \cite{HnHoJu01} and has negative $g_{v2}^\ast$. The second fixed point $(2^\ast)$
795: is unstable. The physical meaning has the third "kinetic" point $(3^\ast)$ whose
796: parameters $\{g_1, g_2, u\}$ dependence on the dimension $d$ is shown in figure~1.
797: for physical value of $\epsilon=2$.
798: %
799: % <<<<< Fig.1 <<<<< Kinet.bod G1,G2,U(d) <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
800: \begin{figure}[ht] % [ht] here, [t] top, [e] end
801: \vspace*{-3mm}
802: \begin{center}
803: \includegraphics[angle=-90,width=90mm]{Figure1.eps}
804: \vspace{-3mm}
805: \caption{Dependence of the parameters $\{g_{v1}, g_{v2}, u\}$ \\
806: on the dimension $d$ for $\epsilon=2$ at the kinetic fixed point (\ref{fixed2bez}).}
807: \end{center} \label{f1}
808: \end{figure} %<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
809: %
810: \\
811: Setting $\epsilon=2$ and $u^\ast$ from (\ref{uu}) one obtains
812: \begin{equation}\fl\ \
813: g_{v1}^\ast = \frac{(2\pi)^d}{S_d}
814: \frac{8(u^\ast+1)(3 d^3-d^2-6 d+8)}{9(d-1)^2(d+2)} \,,\quad
815: g_{v2}^\ast = \frac{(2\pi)^d}{S_d}
816: \frac{32(u^\ast+1)(d^2-2)}{9(d-1)^2(d+2)}\,.
817: \label{fixed2bez}
818: \end{equation}
819: In this case the sum of $g_{v1}^\ast+g_{v2}^\ast\equiv g_{v}^\ast =
820: (2\pi)^d\, 8 d (u^\ast+1)/3(d-1){S_d}$.
821: %
822: Detailed numerical calculations have shown that the region of stability
823: of this point is limited by the value of parameter $\ a<1\ $ and this
824: limiting value does not depend on the dimension $\ d$. This stable region
825: is denoted as region $A$ in figure~\ref{fig2}.
826: %
827: % <<<<< Fig.2 <<<<< Kinet.bod = Oblast stability <<<<<<<<<<<<<<<<<<<<<<<
828: \begin{figure}[ht] % [ht] here, [t] top, [e] end
829: \vspace*{-3mm}
830: \begin{center}
831: \includegraphics[angle=-90,width=90mm]{Figure2.eps}
832: \end{center}
833: \vspace{-6mm}
834: \caption{Stability regions of kinetic point and the critical dimension $d_c$ \\
835: dependence on parameter $a$. The region $A$ spreads down to $a=0$.}
836: \label{fig2}
837: \end{figure}
838:
839:
840: \subsection{Case of active vector admixture}
841:
842: In the full self-consistent system, the RG equations yield besides
843: the known fixed point in the kinetic regime also the nontrivial
844: magnetic fixed point. If the both are stable in the same region of
845: parameters then the choice between two possible critical regimes
846: will depend on initial conditions for RG equations, i.e. critical
847: behavior of the system is non universal.
848:
849: \subsubsection{Kinetic fixed point.}
850:
851: \hspace*{0mm}
852: The nontrivial stable kinetic fixed point of RG equations has been found to be the same
853: as in the previous case of passive magnetic field admixture because the $\beta$-functions
854: $\beta_{gv1}, \beta_{gv2}$ are the same for zero $g_{b1}, g_{b2}$. Only difference was
855: found in the stability region in dependence on parameter $a$: the stable region is
856: enlarged by new region $B$ unlike the case of passive magnetic field admixture, see
857: figure~\ref{fig2}. The critical dimension $d_c$ continuously decreases from $3$
858: to $2$ in dependence on value of parameter $a$ from the interval
859: $\langle 1.1595, 1.427\rangle$.
860: It confirms the results of \cite{HnHoJu01} that the stability of kinetic scaling regime
861: is strongly affected by the behavior of magnetic fluctuations.
862:
863: Figure~\ref{fig3ab} shows values of the charges $g_{v1}, g_{v2}$ which continuously
864: depend on value of nonzero $\epsilon\leq 2$, for two special case of $d$ equal
865: to $2$ and $3$. The right axle corresponds to the physical value of $\epsilon=2$.
866: While the both charges remain nonzero (positive) for $d=2$, in three dimensions
867: one of them, $g_{v2}$ rapidly decreases for $\epsilon\rightarrow 0$.
868: The stable as well as unstable regions depends on parameter $a$ and the critical
869: value of $a$ remains the same for $\epsilon=2$ following from figure~\ref{fig2},
870: or greater for $\epsilon<2$ (the critical $a$ increases for $\epsilon\rightarrow 0$).
871: %
872: % <<<<< Fig.3.ab <<<<< Kin.bod = zav. od epsilon (d=2; 3) <<<<<<<<<<<<<<<<<
873: \begin{figure}[ht] % [ht] here, [t] top, [e] end
874: \vspace*{-3mm}
875: \begin{center}
876: \begin{flushleft} \includegraphics[angle=-90,width=67mm]{Figure3a.eps}
877: \end{flushleft} \vspace{-50mm}
878: \begin{flushright} \includegraphics[angle=-90,width=67mm]{Figure3b.eps}
879: \end{flushright}
880: \end{center}
881: \vspace{-4mm}
882: \caption{Dependence of the parameters $\{g_{v1}, g_{v2}, u\}$ on value of $\epsilon$ \\
883: for $2$- and $3$-dimensions at the kinetic fixed point in the general case. \\
884: Dashed line shows the critical value of $a$ at the stability region limit.}
885: \label{fig3ab}
886: \end{figure}
887:
888:
889: \subsubsection{Magnetic fixed point.}
890:
891: \hspace*{0mm} We have shown in (\ref{gamaD2}) that in two dimensions
892: the function $\gamma_5$ vanishes and then both functions
893: $\beta_{gb1}$ and $\beta_{gb2}$ contain the same linear combination
894: of $\gamma$ functions. Thus, at least one of the magnetic charges
895: ($g_{b1}, g_{b2})$ must be zero in fixed point. But in the other
896: dimensions this restriction does not take place.
897:
898: Here we restrict ourselves only by finding nontrivial magnetic fixed point.
899: In \cite{HnHoJu01} there was mentioned that it is characterized by zero
900: $g_{v1}^\ast$ and $u^\ast$. Therefore, the set of five equations of zero
901: $\beta$-functions (\ref{bety}) is reduced to three equations. Applying $\ g_{v1}=u=0\ $
902: in (\ref{bety}), (\ref{gama}) and (\ref{lambdy}) one obtains the set
903: \begin{eqnarray}
904: a_1g_{v2} +a_2g_{v2}^2 +a_3g_{v2}g_b -a_4g_b^2 &=& 0 \ ,
905: \nonumber\\
906: -A_0 +a_5g_{v2} +a_6g_b &=& 0 \ ,
907: \nonumber\\
908: a_1g_{b2} +a_5g_{v2}g_{b2} +a_6g_{b2}g_b -a_7g_{v2}g_b &=& 0 \ ,
909: %\nonumber\\
910: %-g_{b1} +g_{b2} +g_b &=& 0 \ ,
911: \label{betared}
912: \end{eqnarray}
913: where
914: \begin{eqnarray}
915: A_0 &=& \frac{2a\epsilon}{S_d}\ ,\qquad\quad\ a_1=\frac{2(d-2)}{S_d}\ ,
916: \quad\quad a_2=\frac{(d-1)}{2d}\ ,
917: \nonumber \\
918: a_3 &=& \frac{(d^2-5)}{d(d+2)}\ ,\quad\ a_4=\frac{(d^2-2)}{4d(d+2)}\ ,
919: \quad\ a_5=\frac{(d^2+d-1)}{d(d+2)} ,
920: \nonumber \\
921: a_6 &=& \frac{(5d^2-3d-32)}{4d(d+2)}\ ,\qquad\qquad\qquad\quad
922: a_7=\frac{(d-2)}{2d} \ . \label{acoef}
923: \end{eqnarray}
924: Positive coefficients $a_1, a_7$ vanishes at $d=2$, $a_3$ and $a_6$ are positive
925: for $d>2.236$ and $d>2.848$, respectively.
926: %
927: % <<<<< Fig.4 <<<<< Mag.bod = Oblast stability <<<<<<<<<<<<<<<<<<<<
928: \begin{figure}[ht] % [ht] here, [t] top, [e] end
929: \vspace*{-4mm}
930: \begin{center}
931: \includegraphics[angle=-90,width=90mm]{Figure4.eps}
932: \end{center}
933: \vspace{-6mm}
934: \caption{The stability region of the magnetic fixed point \\
935: in the plane of $\{d,a\}$ for the physical value of $\ \epsilon=2$.}
936: \label{fig4}
937: \end{figure}
938: %
939: The set (\ref{betared}) can be analytically solved with respect to $g_{v2}, g_{b1},
940: g_{b2}$. Because all $g_i$ must be positive, the system (\ref{betared}) with
941: $\ g_{v1}=u=0\ $ gives the only solution,
942: \begin{eqnarray}
943: g_{v2} &=& \frac{A_0 - a_6 g_b}{a_5} \,, \qquad
944: g_{b1} = g_b -\frac{g_b(a_6g_b-A_0)-a_5a_7}{a_5(a_1+2a_6g_b-A_0)} \ ,
945: \nonumber\\
946: g_{b2} &=& \frac{g_b(a_6g_b-A_0)-a_5a_7}{a_5(a_1+2a_6g_b-A_0)} \ ,
947: \label{solut}
948: \end{eqnarray}
949: where
950: \begin{eqnarray}
951: g_b = \frac{-a_1a_5a_6 +a_3a_5A_0 -2a_2a_6A_0 +a_5\sqrt{D}}
952: {2(a_4a_5^2 +a_3a_5a_6 -a_2a_6^2)} \,,
953: \label{denominator}\\
954: D =a_1^2a_6^2 +4a_1a_4a_5A_0 +2a_1a_3a_6A_0 +a_3^2A_0^2 +4a_2a_4A_0^2\,.
955: \nonumber
956: \end{eqnarray}
957: Note that the parameters $a$ and $\epsilon$ appears in the solution
958: only as the product $a\epsilon$ in $A_0$. The physical value is
959: restricted by the inequality $a\epsilon\leq2$. Denominator in the
960: expression (\ref{denominator}) for $g_b$ is zero at $d_0=2.2628$
961: (and it is positive for $d>d_0$), therefore, at $d_0$ we can expect
962: discontinuity and/or divergence. Numerical analysis of the
963: expressions (\ref{solut}) shows that all $g_i$ have a discontinuity
964: at $d_0$, and, a physical solution cannot exist for any $a,
965: \epsilon\ $ if $d\leq d_0$. The stability region of the magnetic
966: fixed point and the corresponding critical dimension $d_c$ was
967: determined numerically and it is shown in figure~\ref{fig4}.
968: %
969: % <<<<< Fig.5.abc <<<<< Mag.bod = zav. Gv2,Gb1,Gb2(d) <<<<<<<<<<<<<<<<<<
970: \begin{figure}[ht] % [ht] here, [t] top, [e] end
971: \vspace*{-7mm}
972: \begin{center}
973: \includegraphics[width=90mm]{Figure5.eps}
974: \end{center}
975: \vspace{-13mm}
976: \caption{Dependence of the parameters $\{g_{v2}, g_{b1}, g_{b2}\}$ on the dimension $d$ \\
977: for $\epsilon=2$ at the magnetic fixed point (\ref{solut}) for $a=1, 0.5, 0$.
978: All $g_i$ have \\ discontinuity at $d=2.2628$ (chained vertical line).}
979: \label{fig5}
980: \end{figure}
981: \\
982: Figure~\ref{fig5} demonstrates the charges $g_{v2}, g_{b1}, g_{b2}$
983: dependence on dimension $d$ for several values of parameter $a$.
984: First, we have found that $g_{v2}, g_{b2}$ tend to infinity at limit
985: value $d_0$. For increasing dimension $d$ from $2$ up to $d_0$ the
986: charge $g_{b2}$ increase from a small positive value up to infinity
987: at $d_0$ and, therefore, $g_{v2}$ decrease here from a small
988: negative value to minus infinity at $d_0$ (because $g_{v2}\propto
989: -a_6 g_b$ and both $a_6$ and $g_b$ are negative in this dimensions).
990: The charge $g_{b1}$ rapidly decrease to zero at $d=2.352$ for
991: decreasing $d$ and continue to minus infinity at $d_0$. These
992: limiting value are in correspondence with numerical calculation of
993: the stability region - the system losses stability for the critical
994: dimension $d_c$ lower than approximately $2.36$ for arbitrary
995: parameter $a$.
996: % ----------------------------------------------------------------------
997:
998:
999: \section{Discussion and conclusions}\label{sec5}
1000:
1001: In this paper we revised the calculations of stability ranges of
1002: developed magnetohydrodynamic turbulence in the frame of double
1003: expansion scheme. The modified standard minimal subtraction scheme
1004: \cite{HnJoJuSt01} has been used in the dimension region of $ d\geq2
1005: $ up to $ d=3 $ in both cases of the magnetic field treated as a
1006: passive as well as active vector admixture.
1007: We confirm existence of the known "kinetic" fixed point (corresponding to
1008: Kolmogorov scaling regime) what is the same
1009: in the both considered cases and only difference is in the stability
1010: region: the critical dimension $d_c$ is achieved for a slightly
1011: higher value of $a$-parameter of a magnetic forcing in the case of
1012: active magnetic field. Limit value of the inverse Prandtl number at
1013: $d=3$ restores the value of $ u=1.393 $ which is known from usual
1014: $\epsilon$-expansion, and it fluently rises to $u=1.562$ at $d=2$,
1015: (figure~1).
1016:
1017: It was believed earlier that in the double expansion being defined for
1018: the space dimension to be closed to two the results obtained in two dimensions can not be
1019: applicable to opposite dimension interval end closed to three. Here we have showed that
1020: the double expansion in exact $d$-dimensional formulation gives some critical dimension
1021: $d_c$ above which the scaling regime is governed by the competition of the stable kinetic
1022: and magnetic fixed points which exists in three dimensions.
1023:
1024: A new nontrivial results of the present paper is connected with derivation of the exact
1025: analytical expression for the nontrivial "magnetic" stable fixed point with $u=g_{v1}=0$
1026: but nonzero $g_{v2}, g_{b1}$ and $g_{b2}$ as well as specification of the borderline
1027: dimension $d_c$. A physical region of the RG fixed point lies
1028: below the $a\epsilon=2$ line, see in figure~4. This point completely losses stability
1029: below the critical value of dimension $d_c=2.36$ (independently on the $a$-parameter)
1030: and also below the value of $a_c=0.146$ (independently on the dimension).
1031: Thus we confirm, in particular, that thermal fluctuations of the magnetic scaling
1032: regime may occur, and, in comparison with earlier results our value of the borderline
1033: dimension ($d_c=2.36$) is significantly lower than in the $\epsilon$ expansion
1034: \cite{FoSuPo82} ($d_c=2.85$) and rather lower than in the 'modified' double expansion
1035: introduced in \cite{HnHoJu01} ($d_c=2.46$) but it is rather higher then value
1036: ($d_c=2.2$) calculated in the frame of the McComb's renormalization \cite{Verma04}.
1037:
1038: Note that the stability of any regime determines the concrete Alfven ratio $r_A$
1039: (ratio of kinetic and magnetic energy density in MHD turbulence, see
1040: \cite{Verma01a,Verma01b}, for example).
1041: Once the stationary scaling regime becomes and stands, the Alfven ratio is fixed
1042: (i.e., it means that the fixed point is reached in the field RG terminology).
1043: Thus the injected energy necessary to steady the stationary scaling regime must
1044: have specific value, or, in another words, all "coupling constants" $g_i$ are
1045: fixed in scaling regime with values which are dependent on dimension $d$.
1046: In like manner the inverse Prandtl number $u\equiv\eta/\nu$ ($\eta$ is magnetic
1047: resistivity) is thus fixed.
1048: Verma \cite{Verma01b} has obtained $\eta/\nu=0.85/0.36=2.36$ in $3$-dimensions
1049: for large $r_A\approx5000$ (corresponding to region of the kinetic regime) and
1050: for zeroth normalized cross-helicity. For smaller $r_A$ this ratio decrease to
1051: $0.69$ for $r_A=1$, and, the both $\eta$ and $\nu$ vary approximately as $d^{-1/2}$
1052: \cite{Verma01a}. We have mentioned above that in our double expansion calculation
1053: in the kinetic point we have fixed the ratio $u\equiv\eta/\nu$ with its $d$-dependence
1054: showed in figure 1.
1055: The magnetic fixed point is characterized by decreasing value of $u$ to zero
1056: what is in correspondence with results of \cite{Verma01b}: his calculation gives
1057: for decreasing $r_A$ (magnetic regime) in $3$-dimensions also decreasing value
1058: of $\eta/\nu$ as one can expect in the magnetic fixed point.
1059:
1060:
1061: \ack %\section*{Acknowledgement}
1062: This work was supported by Science and Technology Assistance
1063: Agency under the contract No. APVT-51-027904, and by grant
1064: RFFI-RFBR 05-02-17603, and by SAS, project No. 2/6193.
1065:
1066: % ======================== BIBLIOGRAPHY ===========================
1067: \Bibliography{99}
1068: %1
1069: \bibitem{FoNeSt77}
1070: Forster D, Nelson D R and Stephen M J 1977 {\it Phys. Rev.} A {\bf 16} 732
1071: %2
1072: \bibitem{DomMar79}
1073: De Dominicis C and Martin P C 1979 {\it Phys.Rev.} A {\bf 19} 419
1074: %3
1075: \bibitem{MaSiRo73}
1076: Martin P C, Siggia E D and Rose H A 1973 {\it Phys. Rev.} A {\bf 8} 423
1077: %4
1078: \bibitem{BaJaWa76}
1079: Bausch R, Janssen H K and Wagner H 1976 {\it Z. Phys. B: Condens. Matter} {\bf 24} 113
1080: %5
1081: \bibitem{AdAnVa96-99}
1082: Adzhemyan L Ts, Antonov N V and Vasil'ev A N 1996 {\it Usp. Fiz. Nauk}
1083: {\bf 166} 1257 [1996 {\it Phys. Usp.} {\bf 39} 1193];
1084: Adzhemyan L Ts, Antonov N V and Vasil'ev A N 1999 {\it The Field
1085: Theoretic Renormalization Group in Fully Developed Turbulence}
1086: (London: Gordon $\&$ Breach)
1087: %6
1088: \bibitem{Vasiliev98-04}
1089: Vasil'ev A N 1998 {\it Quantum-Field Renormalization Group in the Theory of Critical Phenomena
1090: and Stochastic Dynamics} (St. Petersburg: St. Petersburg's Institute of Nuclear Physics)
1091: [in Russian; English translation: Gordon \& Breach, 2004]
1092: %7
1093: \bibitem{McComb90-95}
1094: McComb W D 1990 {\it The Physics of Fluid Turbulence} (Clarendon: Oxford Univ. Press);
1095: McComb W D 1995 {\it Rep. Prog. Phys.} {\bf 58} 1117
1096: %8
1097: \bibitem{Davidson04}
1098: Davidson P A 2004 {\it Turbulence} (Oxford, University Press)
1099: %9
1100: \bibitem{YakOrs86-87}
1101: Yakhot V and Orszag S A 1986 {\it J. Sci. Comput.} {\bf 1} 3;
1102: Dannevik W P, Yakhot V and Orszag S A 1987 {\it Phys. Fluids} {\bf 30} 2021
1103: %10
1104: \bibitem{FoSuPo82}
1105: Fournier J D, Sulem P L and Poquet A 1982 {\it J. Phys. A: Math. Gen.} {\bf 35} 1393
1106: %11
1107: \bibitem{CamTas92}
1108: Camargo S J and Tasso H 1992 {\it Phys. Fluids} B {\bf 4} 1199
1109: %12
1110: \bibitem{Vasiliev76}
1111: Vasil'ev A N 1976 {\it Functional Methods in Quantum Field Theory and Statistics}
1112: (Leningrad: Leningrad University Press)[in Russian]
1113: %13
1114: \bibitem{Zinn89}
1115: Zinn-Justin J 1989 {\it Quantum Field Theory and Critical
1116: Phenomena} (Oxford: Oxford Univ. Press)
1117: %14
1118: \bibitem{AdVaPi83}
1119: Adzhemyan L Ts, Vasil'ev A N and Pis'mak Yu M 1983 {\it Teor. Mat. Fiz.} {\bf 57} 268
1120: %15
1121: \bibitem{AdVaHn84}
1122: Adzhemyan L Ts, Vasil'ev A N and Hnatich M 1984 {\it Teor. Mat. Fiz.} {\bf 58} 72
1123: %16
1124: \bibitem{WilKog74}
1125: Wilson K and Kogut J 1974 {\it Phys. Rep.} {\bf 12C} 75
1126: %17
1127: \bibitem{AdVaHn85}
1128: Adzhemyan L Ts, Vasil'ev A N and Hnatich M 1985 {\it Teor. Mat. Fiz.} {\bf 64} 196
1129: %18
1130: \bibitem{AdVaHn87}
1131: Adzhemyan L Ts, Vasil'ev A N and Hnatich M 1987 {\it Teor. Mat. Fiz.} {\bf 72} 369
1132: %19
1133: \bibitem{AdAnKoVa03}
1134: Adzhemyan L Ts, Antonov V N, Kompaniets M V and Vasiliev A N 2003
1135: {\it Int. J. Mod. Phys.} B {\bf 17} 2137
1136: %20
1137: \bibitem{AdAnBa01}
1138: Adzhemyan L Ts, Antonov V N, Barinov V A, Kabrits Yu S and Vasiliev A N 2001
1139: {\it Phys. Rev.} E {\bf 64} 056306
1140: %21
1141: \bibitem{Verma04}
1142: Verma M K 2004 {\it Phys. Rep.} {\bf 401} 229
1143: %22
1144: \bibitem{Verma01a}
1145: Verma M K 2001 {\it Phys. Plasmas} {\bf 8} 3945
1146: %23
1147: \bibitem{Verma01b}
1148: Verma M K 2001 {\it Phys. Rev.} E {\bf 64} 026305
1149: %24
1150: \bibitem{HonNal96}
1151: Honkonen J and Nalimov M Yu 1996 {\it Z. Phys. B: Condens. Matter} {\bf 99} 297
1152: %25
1153: \bibitem{Liang93}
1154: Liang W Z and Diamond P H 1993 {\it Phys. Fluids} B {\bf 5} 63
1155: %26
1156: \bibitem{Kim99}
1157: Kim C B and Yang T J 1999 {\it Phys. Plasmas} {\bf 6} 2714
1158: %27
1159: \bibitem{HnHoJu01}
1160: Hnatich M, Honkonen J and Jurcisin M 2001 {\it Phys. Rev.} E {\bf 64} 056411
1161: %28
1162: \bibitem{Polchinski}
1163: Polchinski J 1984 {\it Nucl. Phys.} B {\bf 231} 269
1164: %29
1165: \bibitem{HnJoJuSt01}
1166: Hnatich M, Jonyova E, Jurcisin M and Stehlik M 2001 {\it Phys. Rev} E {\bf 64} 016312
1167: %30
1168: \bibitem{HnHoSt94}
1169: Hnatich M, Horvath D and Stehlik M 1994 Theory of the randomly forced MHD turbulence
1170: {\it Preprint JINR Dubna} E17-94-313
1171: %31
1172: \bibitem{HnaSte91}
1173: Hnatich M and Stehlik M 1992 In {\it Renormalization group '91} 204
1174: Eds. Shirkov D V and Priezzev V B (Singapore: World Scien. Pub.)
1175: %32
1176: \bibitem{Honkonen98}
1177: Honkonen J 1998 {\it Phys. Rev.} E {\bf 58} 4532
1178: %33
1179: \bibitem{HnHoHoSe99}
1180: Hnatich M, Honkonen J, Horv\'ath D and Seman\v{c}\'{\i}k R 1999
1181: {\it Phys. Rev.} E {\bf 59} 4112
1182: %34
1183: \bibitem{Collins85}
1184: Collins J 1985 {\it Renormalization} (Cambridge: Cambridge University Press)
1185: %35
1186: \bibitem{Hoft73}
1187: 't Hoft G 1973 {\it Nucl. Phys.} B {\bf 61} 455
1188:
1189: \endbib
1190: \end{document}
1191: % ===================================================================
1192: % E N D
1193: % ===================================================================
1194: