1: \documentclass[twocolumn,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}
2: %\documentclass[preprint,amsmath,amssymb]{revtex4}
3: \usepackage{graphicx}% Include figure files REVTEX
4: \usepackage{dcolumn}% Align table columns on decimal point REVTEX
5: \usepackage{color}
6: \def\AD#1{{\textcolor{magenta}{#1}}}
7: \def\ADD#1{{\textcolor{blue}{#1}}} % addition
8: \def\NOTE#1{{\textcolor{red}{#1}}} % note
9: \topmargin -15pt %REVTEX
10: \newcommand{\beq}{\begin{equation}}
11: \newcommand{\eeq}{\end{equation}}
12:
13: \begin{document}
14:
15:
16: \title{Anisotropic fluxes and nonlocal interactions in MHD turbulence}
17:
18: \author{A. Alexakis, B. Bigot, H. Politano}
19: \affiliation{Laboratoire Cassiop\'ee, UMR 6202,
20: Observatoire de la C\^ote d'Azur,
21: BP 4229, Nice Cedex 4, France}
22:
23: \begin{abstract}
24:
25: {We investigate the locality or nonlocality of the energy transfer
26: and of the spectral interactions involved in the cascade for decaying
27: magnetohydrodynamic (MHD) flows in the presence of a uniform magnetic field
28: $\bf B$ at various intensities.
29: The results are based on a detailed analysis of
30: three-dimensional numerical flows at moderate Reynold numbers.
31: The energy transfer functions, as well as the global and partial fluxes,
32: are examined by means of different geometrical wavenumber shells.
33: %are examined for three different collections of wavevector sets which result in
34: %spherical, cylindrical and plane shells.
35: On the one hand, the transfer functions of the two conserved Els\"asser energies $E^+$ and $E^-$
36: are found local in both the directions parallel ($k_\|$-direction)
37: and perpendicular ($k_\perp$-direction) to the magnetic guide-field,
38: whatever the ${\bf B}$-strength.
39: %The energy cascade thus involves nearby wavenumber shells.
40: On the other hand, from the flux analysis,
41: the interactions between the two counterpropagating Els\"asser waves
42: become nonlocal.
43: Indeed, as the ${\bf B}$-intensity is increased,
44: local interactions are strongly decreased and
45: the interactions with small $k_\|$ modes dominate the cascade.
46: Most of the energy flux in the $k_\perp$-direction is due to modes in the plane at $k_\|=0$,
47: while the weaker cascade in the $k_\|$-direction is due to the modes with
48: $k_\|=1$. The stronger magnetized flows tends thus to get closer to the weak turbulence
49: limit where the three-wave resonant interactions are dominating.
50: Hence, the transition from the strong to the weak turbulence regime
51: occurs by reducing the number of effective modes in the energy cascade.
52: }
53: \end{abstract}
54:
55:
56: \pacs{47.27.ek, 47.65.-d, 47.35.Tv}
57:
58:
59: \maketitle % REVTEX
60:
61: \section{Introduction}
62:
63: The existence of magnetic fields is known in many astrophysical
64: objects such as interstellar medium, galaxies, accretion discs, star and
65: planet interiors, or solar wind (see e.g. \cite{Zeldovich}).
66: In most of these systems, the magnetic fields
67: are strong enough to play a significant dynamical role.
68: The involved kinetic and magnetic Reynolds
69: numbers in these astrophysical bodies are large enough so that
70: the flows exhibit a turbulent behavior with a large
71: continuous range of excited scales, from the largest where energy is injected towards the
72: finest where energy is dissipated.
73: In many cases, a strong large-scale magnetic field is present and induces dynamic anisotropy.
74: Direct numerical simulations that examine in detail the turbulent processes
75: in geo- and astrophysical plasmas are very
76: difficult to achieve, for only rather modest Reynolds numbers can be reached
77: with nowadays computers. One way around this difficulty is to model
78: the small spatial and temporal scales to reproduce the large-scale behavior of
79: turbulent flows. A more basic understanding of turbulence is thus needed
80: to adequately model the flows, in particular when a uniform magnetic
81: field, constant both in space and time, is applied.
82:
83: As a first approximation, the incompressible magnetohydrodynamic
84: (MHD) equations can be used to describe the evolution of both
85: velocity, ${\bf u}$, and magnetic field, ${\bf b}$, fluctuations.
86: In the presence of an uniform magnetic field ${\bf B}$ (magnetic fields are here expressed in velocity units),
87: the Els\"asser
88: formulation of the MHD equations, with constant unit mass density,
89: reads
90: \begin{equation}
91: \label{MHD}
92: \partial_t {\bf z}^\pm = \pm {\bf B} \cdot \nabla {\bf z}^\pm
93: - {\bf z}^\mp \cdot \nabla {\bf z}^\pm -\nabla P + \nu \nabla^2 {\bf z}^\pm
94: \end{equation}
95: together with $\nabla \cdot {\bf z}^\pm=0$, where
96: ${\bf z}^\pm={\bf u\pm b}$ are the Els\"asser fluctuations, and
97: $P$ is the total (kinetic plus magnetic) pressure.
98: We assume here equal molecular viscosity
99: $\nu$ and magnetic diffusivity $\eta$, in other words a unit magnetic Prandtl number ($P_r=\nu/\eta=1$).
100: Hereafter, the direction along the ${\bf B}$ magnetic field
101: is referred to as the parallel direction and the projection of the wavevectors along this
102: direction is denoted $k_\|$, while the two directions of planes perpendicular to ${\bf B}$
103: are referred to as perpendicular directions, the wavevector projection onto such planes
104: being denoted ${\bf k}_\perp$ with norm $ k_\perp \equiv |{\bf k}_\perp|$.
105:
106: For periodic boundary conditions, equations (\ref{MHD}) have two independent invariants
107: in the absence of molecular viscosity and magnetic diffusivity, namely the Els\"asser energies :
108: %\[
109: \beq
110: \label{InvE+-}
111: E^{\pm}=\frac{1}{2}\int {{\bf z}^\pm}^2({\bf x}) \ dx^3.
112: %\]
113: \eeq
114: However, when very small viscosity and magnetic diffusivity are present, it is
115: expected that the nonlinear terms cascade the energies between scales, in
116: the so-called inertial range, up to smallest ones
117: where dissipation becomes effective and removes energy from the system.
118: The rate at which large scales lose
119: energy is then controlled by the nonlinear terms ${\bf z}^\mp \cdot
120: \nabla {\bf z}^\pm -\nabla P$ that are responsible for coupling
121: different scales and cascading the energy towards smaller and smaller
122: scales. The nature of the interactions among various scales in
123: turbulent flows that lead to this cascade is a long standing problem.
124: Understanding the involved mechanisms
125: is very important to predict evolution of the large-scale flow behavior,
126: and to estimate global quantities in astrophysical systems, such as the transport of
127: angular momentum, and accretion rates in accretion discs.
128:
129: High-Reynolds-number hydrodynamic
130: turbulence, often investigated in the framework of
131: statistically homogeneous and isotropic turbulence (which can be
132: questionable in natural flows), is described to first order by
133: the Kolmogorov theory \cite{Kolmogorov1941}. In this phenomenological description,
134: interactions between eddies of similar size give the dominant contribution
135: to the energy cascade. This assumption leads to an
136: energy spectrum in $k^{-5/3}$ and an energy cascade rate
137: proportional to $u_{rms}^3/L$, where $u_{rms}$ is the root-mean-square
138: of the velocity at large scale and $L$ is the typical (large) flow scale.
139:
140: The cascade in MHD turbulence is more complex, especially in the presence
141: of a background magnetic field.
142: Even in the simplest case of zero or small intensity of the ${\bf B}$-field,
143: so that isotropy could be recovered,
144: whether the MHD energy cascade can be described by a phenomenology {\it \` a la}
145: Kolmogorov is still an open question.
146: {In particular, the assumption that interactions between similar size eddies
147: (local interactions) are responsible for the cascade of energy to smaller scales
148: has been questioned in turbulent MHD flows
149: both by theoretical arguments \cite{Vermarev04,Verma2005,Yousef2007} and the use of numerical simulations
150: \cite{Alexakis2005,Mininni2005,Carati2006}.
151: It has been shown for mechanically forced MHD turbulence
152: there is a strong nonlocal coupling between the
153: forced scales and the small scales of the inertial range.}
154: Moreover, the large-scale magnetic field
155: generated by the dynamo action can also locally affect the small scales
156: by suppressing the cascade rate
157: in the same manner that an initially imposed uniform magnetic field would.
158: %%
159: In the other limit, a strong ${\bf B}$-field
160: can lead to the flow bi-dimensionnalization, with a drastic reduction of
161: the nonlinear transfers along the uniform magnetic field.
162: For a ${\bf B}$-intensity (denoted $B$) well above the {\it rms} level of kinetic and magnetic fluctuations,
163: the MHD turbulence may be dominated by the Alfv\'en waves dynamics,
164: leading to wave (or weak) turbulence
165: where the energy transfer, stemming from three-wave resonant interactions,
166: can only increase perpendicular components of the wavevectors, {\it i.e.} components in planes perpendicular to
167: the ${\bf B}$-direction ($k_{\perp}$-direction), the nonlinear transfers along ${\bf B}$
168: ($k_\|$-direction)
169: being completely inhibited \cite{Galtier2000,Galtier2002}.
170: How MHD turbulence moves from the weak turbulence limit,
171: $B \gg u_{rms}$, to the strong turbulence limit, $B \sim u_{rms}$ and $B \sim 0$ (where isotropy
172: could be recovered), is an open question.
173:
174: Various authors have tried to give a physical description of the
175: strong turbulence regime with $B \sim u_{rms}$.
176: Iroshnikov \cite{Iroshnikov1963} and Kraichnan \cite{Kraichnan1965}
177: first proposed a phenomenological description that
178: takes into account the effect of a large-scale magnetic field
179: by reducing the rate of
180: the cascade due to the short time duration of individual collisions of
181: $z^{\pm}$ wave packets.
182: % different wave packets.
183: %%In more detail the time scale of the
184: %cascade of energy due to the collision of two eddies of scale $l$
185: %is reduced to $\tau \sim l B/z_{l}^2$.
186: %
187: %As a result
188: %the energy will cascade in a rate
189: %$\epsilon \sim z_l^4/lB$
190: %and therefore $z_l \sim (\epsilon B)^{1/4}l^{1/4}$ .
191: The resulting 1-D energy spectrum is then given by
192: $E(k)\sim (\epsilon B)^{1/3} k^{-3/2}$.
193: However, this description assumes isotropy and,
194: while the effect of the large-scale field is taken into
195: account by reducing the effective amplitude of the interactions, the
196: interactions themselves are considered to be local.
197: %
198: In order to take into account
199: anisotropy in strong turbulence, a scale dependent anisotropy has been
200: proposed \citep{Goldreich1995},
201: the turbulent $z^{\pm}_l$-eddies being such that the associated Alfv\'en
202: $\tau_{_A}\sim l_\|/B$ and nonlinear
203: $\tau_{_{NL}}\sim l_\perp/z $ times are equal (the so called critical balance),
204: where $l_\|$ and $l_\perp$ are the typical length scales respectively parallel and
205: perpendicular to the mean magnetic field.
206: Repeating the Kolmogorov arguments, one ends up with a
207: $E(k_\|,k_\perp)\sim k_\perp^{-5/3}$ energy spectrum with
208: $k_\|\sim k_\perp^{2/3}$.
209: %
210: %In order to take into account
211: %anisotropy \citep{Goldreich1995} proposed that in strong turbulence
212: %there is scale dependent anisotropy and the
213: %eddies in a turbulent flow are such that the Alfven time scale
214: %$\tau_{_A}\sim l_\|/B$ is of the same order with the non-linear time
215: %scale $\tau_{_{NL}}\sim l_\perp/z $ (so called critical balance
216: %relation), where $l_\|$ and $l_\perp$ are the parallel and
217: %perpendicular to the mean magnetic field typical length scales.
218: %Repeating the Kolmogorov arguments then one ends up with a
219: %$E(k_\|,k_\perp)\sim k_\perp^{-5/3}$ energy spectrum with $l_\|\sim
220: %l_\perp^{2/3}$.
221: %
222: Recently, this result has been generalized
223: in an attempt to model MHD
224: turbulence both in the weak and the strong limit,
225: the ratio of the two time scales $\tau_{_A}/\tau_{_{NL}}$ being
226: kept fixed but not necessarily of order one \cite{Galtier2005}.
227: {In an other approach to obtain the transition from the
228: strong to the weak turbulence limit \citep{Matthaeus1989,Zhou2004} suggested
229: time scale for the energy cascade is given by
230: the inverse average between the Alfv\'en and the nonlinear time scale
231: $\tau^{-1}=\tau_{_A}^{-1}+\tau_{_{NL}}^{-1}$.}
232: All these models however assume locality of interactions that
233: are also in question in anisotropic MHD turbulence \cite{Bhatta2001}.
234: A nonlocal model for anisotropic turbulence
235: has been recently proposed by one of the authors \cite{Alexakis2007};
236: it assumes that the energy cascade
237: is due to interactions between eddies
238: with different parallel sizes and similar perpendicular scales,
239: while a non-universal behavior is expected for moderate Reynold numbers.
240:
241: Although very useful in getting a first order understanding
242: of the processes involved in a turbulent cascade, cascade-energy models have to be unavoidably
243: based on assumptions that need to be tested. To this respect, numerical simulations
244: of the MHD equations are very valuable because they provide information about the evolution of
245: the fields in the whole space, something not easily obtained from observations.
246: Many numerical investigations have been performed
247: during the last two decades
248: \citep{Mason2007,Muller2005,Ng2003,Dmitruk2003,Maron2001,Cho2000,Biskamp2000,Ng1996,Oughton94}
249: and, at the achieved Reynolds numbers, they have demonstrated that
250: different power-law exponents are obtained depending on the chosen forcing.
251: In this work, we use the results of numerical simulations
252: of free decaying MHD flows at moderate Reynolds number to
253: investigate the MHD interactions for various intensities of
254: the external magnetic field.
255: In particular, we try to investigate
256: whether the transfer of energy in the parallel and perpendicular direction is local
257: (i.e. the two energies $E^\pm$ cascade between nearby wavenumbers) or nonlocal
258: (i.e. distant wavenumbers are involved in the cascade),
259: and whether the coupling between the two oppositely moving waves ${\bf z}^+$ and ${\bf z}^-$
260: (that do not exchange energy) is local or not; and if not, which modes are responsible for
261: the energy cascade.
262:
263: The paper is organized as follows. In the next section,
264: we give the precise definitions of the transfer functions and partial fluxes
265: used to analyze the nature of the energy cascade.
266: The details of the numerical simulations are given in Section IIIa.
267: In Section IIIb, we investigate the locality or nonlocality of the energy transfers,
268: and in section IIIc, we examine the nature of the interactions
269: between the two ${\bf z}^+$ and ${\bf z}^-$ fields.
270: We conclude and discuss our results in Section IV.
271:
272:
273: \section{Definitions}
274:
275: Our goal is to investigate the interactions among different scales.
276: To define the notion of ``scale",
277: we use the field Fourier transforms :
278: \beq
279: \label{Fourier}
280: {\bf \widehat{z}^{\pm}(k) } = \frac{1}{({2\pi})^3}\int {\bf {z^{\pm}}(x)}e^{-i{\bf k\cdot x}} dx^3
281: \eeq
282: defined in a $2\pi$-periodic cube, such as
283: \beq
284: \label{Space}
285: {\bf z^\pm(x)} = \sum_{\bf k} {\bf \widehat{z}^{\pm}}({\bf k})e^{i{\bf k\cdot x}},
286: \eeq
287: Similar size eddies will be considered as the ones whose Fourier transform
288: contains similar wavenumbers.
289:
290: In any basic flow interaction, three wavevectors are involved.
291: For example, the evolution of
292: a given Fourier amplitude $\widehat{\bf z}^{+}({\bf k})$
293: will be coupled to a $\widehat{\bf z}^{-}({\bf p})$ one and cascade the energy
294: to the mode $\widehat{\bf z}^{+}({\bf q})$ such that the wavevectors satisfy ${\bf k+p+q}=0$.
295: Note that the mode $\widehat{\bf z}^{-}({\bf p})$ does not gain or lose energy
296: from this interaction since the two energies $E^+$ and $E^-$ are separately
297: conserved. To obtain the cascade mean rate,
298: one needs to average over all possible triadic interactions.
299: To get a phenomenological understanding of the processes
300: at play in MHD turbulence, we need to know if :
301: i) most of the energetic exchanges occur between wavenumbers such that
302: $|{\bf k}|\sim |{\bf q}|$ and ii) the energy flux is a result of spectral
303: interactions of the two fields ${\bf z}^\pm$ with similar wavenumbers or not
304: {($|{\bf k}|\sim |{\bf p}|$)}.
305:
306:
307: To address these questions, let us
308: consider a partition of the wavevectors into non-overlapping sets $S^\pm_{_K}$
309: such that $S^\pm=\bigcup_{K=1}^{\infty}S^\pm_{_K}=\mathbb{Z}^3$.
310: For example $S^\pm_{_K}$ could be the spherical shells of unit width and radius $K$,
311: {\it i.e.} set of wavevectors {\bf k} that have $K<|{\bf k}|\le K+1$.
312: We now define the filtered fields ${\bf z}^\pm_{_K}({\bf x})$
313: so that only modes in the set $S^\pm_{_K}$ are kept:
314: \beq
315: \label{filterz}
316: {\bf z^\pm_{_K}(x)} = \sum_{{\bf k} \in S^\pm_{_K}} {\bf \widehat{z}^{\pm}}({\bf k})e^{i{\bf k \cdot x}}.
317: \eeq
318: Clearly, one gets
319: \beq
320: {\bf z^\pm(x) = \sum_{K} z^\pm_{_K}(x)}.
321: \eeq
322:
323:
324: The triadic interactions among the different sets, say $S^\pm_{_K}$,$S^\mp_{_P}$ and $S^\pm_{_Q}$, are given by:
325: \beq
326: \label{triadic}
327: \mathcal{T}^\pm_3(K,P,Q)=-{\int {\bf z^\pm_{_K}} {\bf z^\mp_{_P}} \cdot \nabla {\bf z^\pm_{_Q}}dx^3}
328: \eeq
329: that express the rates at which $E^\pm$ energies are transfered from $S^\pm_{_Q}$ to
330: $S^\pm_{_K}$ sets due to the interactions with the modes belonging to $S^\mp_{_P}$ set.
331: Note that the collection of sets $S^+$ and $S^-$ need not to be necessarily the same;
332: for example, $S^+$ could be a collection of cylindrical shells while $S^-$
333: could be a collection of plane sheets.
334: Adding over the index $P$ (all sets in $S^\mp$), we obtain the transfer functions :
335: \beq
336: \label{transfer}
337: \mathcal{T}^\pm(K,Q)=\sum_P \mathcal{T}^\pm_3(K,P,Q)=
338: -{\int {\bf z^\pm_{_K}} {\bf z^\mp} \cdot \nabla {\bf z^\pm_{_Q}}dx^3}
339: \eeq
340: that give the $E^+$ and $E^-$ transfer rates from $S^\pm_{_Q}$ to
341: $S^\pm_{_K}$ sets due to all possible interactions.
342: Note that the ${\bf z}^+$ field is not exchanging energy with
343: the ${\bf z}^-$ field, and vice versa, but their interaction is responsible
344: for the redistribution of the energy among various sets.
345: $\mathcal{T}^\pm(K,Q)$ can give us information about the locality or nonlocality of the energy transfer,
346: {\it i.e.} whether the energy is exchanged by nearby sets or long-range transfers from the large
347: scales directly to the small scales are also involved.
348: %is energy being exchanged by nearby sets? or is there also long range transfer from the large
349: %scales directly to the small scales?
350:
351: However, the $\mathcal{T}^\pm(K,Q)$ transfer functions
352: do not give us direct information on the scales of the two fields ${\bf z}^+$ and ${\bf z}^-$ that interact
353: and contribute to the energy cascade.
354: To investigate the locality on nonlocality of the interactions between
355: the two Els\"asser counterpropagating waves, we introduce the partial fluxes
356: (see \cite{Alexakis2005b,Mininni2006})
357: defined as:
358: \begin{eqnarray}
359: \Pi^\pm_{_P}(K)=& \sum_{K'=0}^K \sum_{Q=0}^\infty \mathcal{T}^\pm_3(K',P,Q) \nonumber \\
360: =& -\sum_{K'=0}^K\int {\bf z^\pm_{_{K'}} {\bf z}_{_P}^\mp \cdot \nabla {\bf z}^{\pm}} dx^3
361: \label{partialflux}
362: \end{eqnarray}
363: that express the flux of energy out of the outer surface of the $S^\pm_{_K}$ shell due to the interactions
364: with the $S^\mp_P$ shell.
365: Summation over the whole $S^\mp$ collection of sets enable to recover the usual definition for the global fluxes :
366: %Adding over all the sets in $S^\mp$ we reoptain the usual definition for the total flux
367: \begin{eqnarray}
368: \Pi^\pm(K)=& \sum_{K'=0}^K \sum_{Q=0}^\infty \sum_{P=0}^\infty \mathcal{T}^\pm_3(K',P,Q) \nonumber \\
369: =& -\sum_{K'=0}^K\int {\bf z^\pm_{_{K'}} {\bf z}^\mp \cdot \nabla {\bf z}^{\pm}} dx^3
370: \label{flux}
371: \end{eqnarray}
372:
373: In the current work, we are going to use three different types of wavevector collections.
374: %In the current work, we are going to use three different types of wavenumber sets.
375: We first consider spherical shells
376: traditionally used in studies of isotropic turbulence so
377: that a set $S_{_K}$ contains the wavevectors $\bf k$ such that
378: $K\le {|\bf k|} < K+1$.
379: %As a result the sets $S_{_K}$ are spheres of radius $K+1$.
380: The second collection of sets are cylindrical
381: shells along the direction of the guiding magnetic field.
382: In this case, the set $S_{_K}$ contains the wavevectors $\bf k$ such that
383: $K\le k_{\perp} < K+1$ (with $k_{\perp}=\sqrt{k_x^2 +k_y^2}$).
384: Finally, we consider planes perpendicular to the
385: ${\bf B}$-direction, so that the set $S_{_K}$
386: contains the wavevectors $\bf k$ whose $k_\|$-component satisfies
387: $K\le |k_\|| <K+1$ (where $k_\|$ stands for $k_z$).
388: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
389:
390: \section{Numerical results}
391: \subsection{Numerical setup and initial conditions}
392: We integrate numerically the three-dimensional incompressible MHD equations
393: (\ref{MHD}), in a $2\pi$-periodic box using a pseudo-spectral method
394: with $256^3$ collocation points. The time marching uses an
395: Adams-Bashforth Cranck-Nicholson scheme, i.e. a second-order
396: finite-difference scheme. The initial kinetic and magnetic fields
397: correspond to spectra proportional to $k^2 exp(-k/2)^2$ for
398: $k=[1,8]$, which means a flat modal spectrum for wavevector ${\bf
399: k}$ up to $k=2$, to prevent any favored wavevector at time $t=0$, and
400: the associated kinetic and magnetic energies are chosen equal, namely
401: $E_v(t=0)=E_b(t=0)=1/2$, as in previous numerical studies (see
402: \cite{Milano} and references therein).
403: {Moreover, the correlation between the velocity and magnetic field fluctuations, as measured by
404: the cross-correlation coefficient defined by $2 \int {\bf v}({\bf x}) \cdot {\bf b}({\bf x)} \ dx^3/(E_v+E_b)$,
405: is initially less than $1\%$.}
406: At scale injection, the
407: initial kinetic and magnetic Reynolds numbers are about $800$ for
408: flows at $\nu=\eta \sim 4 \times 10^{-3}$, with $u_{rms}=b_{rms}=1$
409: and an isotropic integral scale $L = 2\pi \int{k^{-1} E_v(k) dk}/
410: \int{E_v(k) dk}$ of about $\pi$. The dynamics of the flow is then let to freely
411: evolve. The parametric study according to the intensity of the
412: background magnetic field ${\bf B}$ is performed for four different
413: values : $B = 0, 1, 5$ and $15$.
414: All the simulations are run up to a computational time $t_{max}=15$,
415: at which the loss of the total energy (kinetic plus magnetic) is about $95\%$ for the simulation with $B=0$,
416: $90\%$ for $B=1$, and $83\%$ for the $B=5$ and $B=15$ runs.
417:
418: Figure \ref{fig1} shows the time evolution of the total energy, $E(t)=(E^+(t) +E^-(t))/2$,
419: and the total enstrophy,
420: $\Omega(t)=1/2 \int [{{\bf w}}^2({\bf x},t)+ {{\bf j}}^2({\bf x},t)] \ dx^3$
421: (where ${\bf w}=\nabla \times {\bf u}$ stands for the vorticity field,
422: and ${\bf j}= \nabla \times {\bf b}$ for the current),
423: for the four different simulations.
424: {One can note that the influence of the strength of the external magnetic field
425: clearly slows down the flow dynamics}.
426: Note that the $B=5$ and $B=15$ flows present a very similar temporal behavior
427: in energy as well as in enstrophy.
428: \begin{figure}
429: \includegraphics[width=8cm]{Fig1.eps}
430: \caption{Temporal evolutions of the total energy (top panel) and of
431: the total enstrophy (bottom panel) for the four examined intensities of
432: the ${\bf B}$ applied field.
433: \label{fig1}}
434: \end{figure}
435: {
436: The analysis that follows in the next subsection is based
437: on the outputs of the runs at $t = 4$ where
438: the spectra are fully developed and all runs wave roughly the same enstrophy.
439: }
440: {At this time, the cross-correlation coefficient, that one can also write
441: $(E^+-E^-)/(E^++E^-)$, is about $3.6\%$
442: and $2.5\%$ respectively for the $B=0$ and $B=1$ flows, while it is
443: close to $1.6\%$ for $B=5$ and $1\%$ for $B=15$, with thus a lesser increase
444: in the stronger magnetized flows.}
445: The energy spectra
446: in the perpendicular direction to the uniform ${\bf B}$-field
447: $E(k_\perp)=1/2 \int [\hat{\bf z}^-(k)]^2+[\hat{\bf z}^+(k)]^2 k_\perp dk_\| $
448: and in the parallel one
449: $E(k_\|)=1/2 \int [\hat{\bf z}^-(k)]^2+[\hat{\bf z}^+(k)]^2 k_\perp dk_\perp$
450: are shown in Figure \ref{fig2} at the same time.
451: Clearly as the magnetic field intensity is increased, the spectrum in the
452: $k_\|$-direction becomes steeper.
453: Because the planes at $k_\|=0$ and $k_\|=1$ are shown to play an important role
454: in the cascade, we mention here their properties in more details.
455: In absence of the applied magnetic field, the modes with $k_\|=0$ contain $32\%$ of
456: the total energy and $8\%$ of the total enstrophy, and the $k_\|=1$ modes have
457: $30\%$ of the total energy and $10\%$ of the total enstrophy.
458: In the strongly anisotropic case, $B=15$,
459: the $k_\|=0$ modes contain $55\%$ of the total energy and
460: $34\%$ of the total enstrophy while the $k_\|=1$ modes have
461: $37\%$ of the total energy and $35\%$ of the total enstrophy.
462: \begin{figure}
463: \includegraphics[width=8cm]{Fig2.eps}
464: \caption{Total energy spectra in the perpendicular (top panel)
465: and the parallel (bottom) directions.
466: \label{fig2}}
467: \end{figure}
468:
469: In our investigation, we focus on the cascade
470: of the $E^-$ energy. The $E^+$ cascade has also been analyzed
471: and it gives qualitatively similar results.
472: We consider separately
473: the cascades in the perpendicular and parallel directions relatively to the applied
474: magnetic field.
475: For this reason, we examine three different types of flux;
476: (i) the flux across spheres of radius $k \equiv |{\bf k}|$ that
477: corresponds to an isotropic analysis,
478: (ii) the flux across cylinders of radius $k \equiv k_{\perp}$ that
479: corresponds to the flux in the perpendicular direction
480: and (iii) the flux across planes located at $k \equiv |k_\||$ that corresponds to the flux
481: in the direction parallel to the ${\bf B}$-field direction.
482: Figure \ref{fig3} shows these three fluxes, as a function of $k$,
483: for various ${\bf B}$-intensities.
484: It is clear that as the amplitude of the large-scale magnetic field is increased,
485: the parallel flux is strongly reduced.
486: For $B=5$, this flux is reduced by more than one order of magnitude
487: when compared to the case with $B=0$.
488: For $B=15$, the parallel flux across planes is very small and
489: it even takes negative values.
490: \begin{figure}
491: \includegraphics[width=8cm]{Fig3.eps}
492: \caption{Fluxes $\Pi_S^{-}(k)$ across : (i) spheres (solid line),
493: (ii) cylinders (dashed line), (iii) planes (doted line), for $B=0$ (top panel),
494: $B=1$ (mid panel) and $B=5$ (bottom panel).
495: \label{fig3}}
496: \end{figure}
497:
498: \subsection{Energy transfers}
499:
500: We now examine the locality or nonlocality
501: of energy transfers from our numerical data.
502: For two different values of the uniform magnetic field, namely $B=0$ and $B=5$,
503: Figure \ref{fig4} shows a shadow-graph of the transfer function $\mathcal{T}^-(K,Q)$ between
504: ${\bf z^-_K}$ and ${\bf z^-_Q}$, defined in Eqs. (\ref{filterz}) and (\ref{transfer}),
505: for energy exchanges across cylindrical shells (perpendicular cascade),
506: while Figure \ref{fig5} shows the transfer function $\mathcal{T}^-(K,Q)$
507: for energy exchanges across plane sheets (parallel cascade).
508: In all cases, the transfer is concentrated along the diagonal
509: $K=Q$ line. This indicates that the cascade happens through a local energy exchange.
510: Similar results are obtained from the two other simulations at $B=1$ and $B=15$ (not shown).
511: Note the highly non-linear color bar used for the parallel cascade in the $B=5$ case.
512: This choice is due to the extremely fast decrease of the amplitude of $\mathcal{T}^-(K,Q)$
513: as the wavenumbers $K$ and $Q$ become large.
514: \begin{figure}
515: \includegraphics[width=8cm]{Fig4.eps}
516: \caption{The transfer function $\mathcal{T}^-(K,Q)$
517: that demonstrates the energy exchange between {\it cylindrical shells}
518: of radius $K$ and $Q$. Solid lines show the diagonal $K=Q$.
519: The top panel shows the $B=0$ case and the bottom panel the $B=5$ case.
520: \label{fig4}}
521: \end{figure}
522: \begin{figure}
523: \includegraphics[width=8cm]{Fig5.eps}
524: \caption{The transfer function $\mathcal{T}^-(K,Q)$
525: that demonstrates energy exchanges between {\it plane sheets}
526: located at distance $K$ and $Q$ from origin, for
527: $B=0$ (top panel) and $B=5$ (bottom panel) cases.
528: Solid lines indicate the $K=Q$ diagonal.
529: \label{fig5}}
530: \end{figure}
531: From Figures \ref{fig4} and \ref{fig5}, it can be seen that most of the energy exchange
532: happens close to the diagonal line ($K=Q$). This implies that waves traveling in the same direction
533: exchange energy between similar size wavenumbers.
534: In the strong $B$ flow, some inverse cascade is also visible in the parallel cascade
535: (Fig. \ref{fig5}) as
536: indicated by the dark lines below the diagonal and the bright ones above the diagonal.
537:
538: To get a better understanding of the $\mathcal{T}^\pm(K,Q)$ transfer functions, we look at
539: a single wavenumber $Q$. Figure \ref{fig6} displays
540: $\mathcal{T}^-(K,Q)$ for the perpendicular
541: cascade (cylinders) at $Q=10$ as a function of $K$, whereas Figure \ref{fig7} shows it for
542: the parallel cascade (planes) at $Q=10$.
543: To compare the results obtained from the different $B$ cases,
544: the $\mathcal{T}^-(K,Q)$ amplitudes are normalized so that
545: all transfers are of the same order of magnitude.
546: Positive values of $\mathcal{T}^-(K,Q)$ imply that the shell $K$ receives
547: energy from the shell $Q=10$ (Fig. \ref{fig6}, perpendicular case) and
548: (Fig. \ref{fig7}, parallel case) while negative values of $\mathcal{T}^-(K,Q)$
549: mean that the shell $K$ gives energy to the shell $Q=10$.
550: \begin{figure}
551: \includegraphics[width=8cm]{Fig6.eps}
552: \caption{The energy transfer function $\mathcal{T}^-(K,Q)$ for the perpendicular
553: cascade ({\it cylinders}) for $Q=10$ as a function of $K$,
554: from the runs with
555: $B=0$ (solid line), $B=1$ (dashed line) and $B=5$ (dotted line).
556: Amplitudes of $\mathcal{T}^-(K,Q)$ are normalized to have
557: the same order of magnitude than in the $B=0$ case.
558: \label{fig6}}
559: \end{figure}
560: \begin{figure}
561: \includegraphics[width=8cm]{Fig7.eps}
562: \caption{The energy transfer function $\mathcal{T}^-(K,Q)$ for the parallel
563: cascade ({\it planes}) for $Q=10$ as a function of $K$,
564: from the data at
565: $B=0$ (solid line), $B=1$ (dashed line) and $B=5$ (dotted line).
566: Amplitudes of $\mathcal{T}^-(K,Q)$ are normalized to have
567: the same order of magnitude than in the $B=0$ case.
568: %$B=0$ solid line, $B=1$ dashed line, $B=5$ doted line.
569: %The amplitude of $\mathcal{T}^\pm(K,Q)$ eve been normalized so that
570: %they are of the same order of magnitude with the $B=0$ case.
571: \label{fig7}}
572: \end{figure}
573:
574: For the perpendicular cascade, the shell $Q=10$ receives most energy from slightly smaller
575: wavenumbers than $K=10$ and it gives energy to slightly larger wavenumbers.
576: This implies a locality in the energy transfer,
577: since it is mostly the nearby cylindrical shells that exchange energy.
578: The parallel cascade presents a similar behavior; the shell $Q=10$ receives energy from slightly smaller
579: wavenumbers than $K=10$ and it gives energy to slightly larger wavenumbers.
580: Note however that for the $B=5$ flow, there is also some trace of an inverse cascade
581: (energy transfer from the wavenumber $Q=10$ to the wavenumber $K=8$).
582: This local behavior has also been found
583: in isotropic {($B=0$)} decaying MHD turbulence simulations \cite{Debliquy2005}.
584: Nevertheless, we need to note that in forced MHD turbulence
585: where the magnetic field is generated by dynamo action, strong nonlocal transfers also exist
586: \cite{Alexakis2005,Carati2006}.
587: Whether these nonlocal transfers are present in the forced anisotropic
588: regime still needs further studies.
589:
590: \subsection{Nonlinear interactions between ${\bf z}^+$ and ${\bf z}^-$}
591: %\subsection{$z^+,\,\,\, z^-$ interactions}
592:
593: The analysis of the energy transfer functions has thus shown that the energy cascades locally.
594: As a result,
595: each Els\"asser field, ${\bf z}^+$ or ${\bf z}^-$,
596: %moving in the same direction
597: {
598: exchanges energy between waves traveling in the same direction of similar size}.
599: Nonetheless, this does not mean that interactions
600: among oppositely traveling waves are local. In the limit of very
601: large intensities of the background magnetic field, where the weak turbulence theory is valid,
602: the energy cascade is due to interactions with the modes in the plane at $k_\|=0$.
603: Therefore, modes with $k_\|\ge 1$ interact with modes $k_\|\ll1$ to cascade the energy.
604: To that respect, the interactions are nonlocal since short waves (large $k_\|$) interact
605: with long waves (small $k_\|$ ) to cascade the energy.
606: To investigate how close to the weak turbulence regime we are,
607: we plot in Figure~\ref{fig8} the total energy flux $\Pi^{-}(K)$, defined in Eq.~(\ref{flux}),
608: across cylinders together with the partial
609: flux $\Pi_{P=0}^{-}(K)$, defined in Eq.~(\ref{partialflux})
610: due to interactions with just $k_\|=0$ modes.
611: \begin{figure}
612: \includegraphics[width=8cm]{Fig8.eps}
613: \caption{The total energy flux $\Pi^{-}(K)$ (solid line)
614: across {\it cylinders} of radius $K$ together with the
615: partial flux $\Pi_{P=0}^{-}(K)$ (dashed line)
616: for the four different values of $B$, from $B=0$ (top panel) up to
617: $B=15$ (bottom panel).
618: \label{fig8}}
619: \end{figure}
620: As the strength of the uniform magnetic field is increased,
621: the flux due to the interactions with the $k_\|=0$ modes
622: become more and more dominant.
623: In the $B=15$ flow, the global and partial fluxes
624: {across cylinders}
625: become almost indistinguishable suggesting that
626: interactions with the modes in the plane at $k_\|=0$
627: are responsible for the energy cascade. {This means that the flow dynamics tends to be closer
628: to a weak turbulence regime where the three-wave resonant interactions are dominating.}
629:
630: {A different behavior is obtained for the parallel energy cascade.}
631: When a mode $\widehat{\bf z}^-({\bf k})$ interacts with a mode $\widehat{\bf z}^+({\bf p})$,
632: %with $\bf p_\|=0$
633: the $\widehat{\bf z}^-({\bf k})$ energy will move to the wavevector $\bf q$
634: so that the relation $\bf k+p+q=0$ holds.
635: If however $\bf p$ belongs to the wavevector set
636: with $p_\|=0$, this relation then reads
637: %$\bf p_\|=0$ then this relation reads
638: $k_\|+q_\|=0$ in the parallel direction, i.e. $|k_\||=|q_\||$. Therefore, the energy
639: remains in spectral planes located at the same distance from the origin. As a result,
640: interactions with
641: the $k_\|=0$ modes cannot contribute to the energy cascade in the parallel direction.
642: In this case, the closest modes to the $k_\|=0$ modes are the ones that gives most of the
643: energy flux.
644: Figure \ref{fig9} shows the total energy flux across planes and the partial
645: flux only due to interactions with the modes in the plane at $k_\|=1$ (the closest to
646: the $k_\|=0$ plane).
647: As the amplitude of the ${\bf B}$-field is increased,
648: most of the parallel flux comes from modes with $k_\|=1$.
649: Here, we need to note that the flux in the parallel direction
650: is much noisier than the flux in the perpendicular direction
651: and that it often presents negative values
652: (absolute values are plotted in the bottom panel of Fig. \ref{fig9}).
653: A thorough analysis
654: of the parallel cascade would require to average many data outputs
655: which is not possible in the case of a freely decaying flows.
656: Such an analysis is left for future work.
657: \begin{figure}
658: \includegraphics[width=8cm]{Fig9.eps}
659: \caption{The total energy flux $\Pi^{-}(K)$ (solid line)
660: across {\it planes} at $k_\|=K$ together with the
661: partial flux $\Pi_{P=1}^{-}(K)$ (dashed line)
662: for the four different values of $B$
663: from $B=0$ (top panel) up to $B=15$ (bottom panel).
664: %in an increasing order
665: %($B=0$ top panel $B=15$ bottom panel).
666: \label{fig9}}
667: \end{figure}
668:
669:
670: \section{Conclusion and Discussion}
671:
672: In this work we examine the energy cascade and
673: the interactions between different scales for freely decaying MHD flows
674: in the presence of a uniform magnetic field.
675: Our analysis is based on data obtained from direct numerical
676: simulations of the MHD equations with four different intensities of the applied magnetic field,
677: in an attempt to study the transition from strong to weak turbulence limit.
678: One clearly established result is that,
679: as the strength of the uniform magnetic field is increased,
680: the energy spectrum becomes anisotropic with most of the energy concentrated in the
681: small $k_\|$ wavenumbers, as already known \cite{Oughton94}.
682: It is further shown that the energy flux in the parallel direction (relatively to the uniform
683: magnetic field) is also strongly suppressed
684: when the guiding field in introduced.
685:
686: To investigate the locality or nonlocality of the spectral interactions,
687: we measure the transfer functions for the parallel and the perpendicular cascade.
688: %and measured the flux in the parallel and the perpendicular direction
689: %due to the coupling of opposite polarity modes in different planes.
690: The transfer functions in the parallel and perpendicular directions
691: are found local whatever the strength of the external magnetic field.
692: As a result, the coupling between modes that travel in the same direction
693: is local and the energy exchange occurs between similar size eddies.
694: This behavior has been shown to hold in decaying isotropic MHD turbulence
695: simulations {(with $B=0$)} \cite{Debliquy2005}.
696: However, in the presence of a mechanical forcing,
697: strong nonlocal interactions have been observed with a direct energy transfer
698: from the forced scale to the inertial range scales \cite{Alexakis2005,Carati2006}.
699: If this nonlocal behavior persists in the anisotropic case still needs
700: further investigations.
701:
702: The locality or nonlocality of the interactions between oppositely moving
703: waves ($z^+$ and $z^-$), that do not exchange energy, is measured by means of
704: partial fluxes in the parallel and the perpendicular directions due
705: to the coupling in different spectral planes.
706: This coupling between oppositely propagating modes does not appear local.
707: As the amplitude of the applied magnetic field is increased,
708: most of the interactions occur with the $k_\|=0$ modes that are dominant in cascading
709: the energy.
710: Most of the energy flux is thus in the perpendicular direction,
711: since the $k_\|=0$ modes do not contribute to the energy cascade in the parallel direction.
712: {Hence, the stronger magnetized flows tends to present a dynamics close to the weak turbulence
713: limit where the three-wave resonant interactions are responsible for the cascade process.}
714: This also partly explains the similar temporal evolution in the $B=5$ and $B=15$ regimes
715: (see Figure \ref{fig1}) since, in both cases, most of the cascade
716: is due to the $k_\|=0$ modes.
717:
718: For the parallel cascade, the interactions are slightly different. As already said, this
719: is due to the inability of the $k_\|=0$ modes to cascade the energy in the parallel direction.
720: In that case, the modes with the smallest but not zero $k_\|$ ($k_\|\simeq1$) are the
721: ones responsible for the cascade.
722: This behavior is in qualitative agreement with the description of a recent phenomenological model
723: \cite{Alexakis2007}. However, the lack of resolution does not allow to pursue
724: a quantitative comparison.
725:
726: Finally, we would like to emphasize that we analyze here numerical data
727: of freely decaying MHD flows submitted to an external magnetic field
728: whose amplitude is varied, while all the other parameters are kept unchanged
729: (periodic boundary conditions, unit magnetic Prandtl number, initial conditions and Reynolds number).
730: Thus, one should be cautious in any attempt to generalize the obtained results, e.g.
731: forced turbulence could lead to different behaviors and should be studied separately.
732: The results could also be dependent on the kinetic Reynolds number as well on the magnetic Prandtl
733: number.
734: {Furthermore the use of a refined spectral grid in the parallel direction, allowing
735: the presence of more modes with $k_\|\simeq0$, could alter the energy cascade.
736: }
737: %Furthermore the use of "longer" computational boxes, allowing more modes
738: %with $k_\|\simeq0$ to be present, could alter the energy cascade.
739:
740: \acknowledgments
741:
742: This work was supported by INSU/PNST and /PCMI Programs and CNRS/GdR Dynamo. Computation
743: time was provided by IDRIS (CNRS) Grand No. 070597, and SIGAMM mesocenter
744: (OCA/University Nice-Sophia).
745:
746:
747:
748: \begin{thebibliography}{}
749:
750: \bibitem[Zeldovich, Ruzmaikin \& Sokoloff (1990)]{Zeldovich}
751: Ya. B. Zeldovich, A.A. Ruzmaikin and D.D. Sokoloff
752: ``Magnetic Fields in Astrophysics" 1990 Gordon \& Breach Science Pub.
753:
754:
755: \bibitem[Kolmogorov (1941)]{Kolmogorov1941}
756: A.N. Kolmogorov,
757: Dokl. Akad. Nauk SSSR {\bf 30},299 (1941)
758:
759: \bibitem[Yousef, Rincon \& Schekochihin (2007)]{Yousef2007}
760: T.A. Yousef, F. Rincon, and A.A. Schekochihin, J. Fluid Mech. {\bf 575}, 111 (2007)
761:
762: \bibitem{Vermarev04}
763: M.K. Verma, Phys. Reports, {\bf 401}, 229 (2004).
764:
765: \bibitem[Verma, Ayyer, \& Chandra (2005)]{Verma2005}
766: M.K. Verma, A. Ayyer, A.V. Chandra,
767: Phys. Plasmas, {\bf 12}, 82307 (2005)
768:
769: \bibitem[Alexakis, Mininni \& Pouquet (2005)]{Alexakis2005}
770: A. Alexakis, P.D. Mininni and A. Pouquet,
771: Phys. Rev. E {\bf 72}, 046301 (2005)
772:
773: \bibitem[Mininni, Alexakis \& Pouquet (2005)]{Mininni2005}
774: P.D. Mininni, A. Alexakis and A. Pouquet,
775: 2005, Phys. Rev. E 72, 046302
776:
777: \bibitem[Carati, et al. (2006)]{Carati2006}
778: D. Carati, O. Debliquy, B. Knaepen, B. Teaca and M.K. Verma
779: J. Turb. {\bf 7}, 1, (2006)
780:
781:
782: \bibitem[Galtier et al.(2002)]{Galtier2002}
783: S. Galtier, S.V. Nazarenko, A.C. Newell and A. Pouquet,
784: \apj, {\bf 564}, L49 (2002)
785:
786: \bibitem[Galtier et al.(2000)]{Galtier2000}
787: S. Galtier, S.V. Nazarenko, A.C. Newell and A. Pouquet,
788: J. Plasma Phys., {\bf 63}, 447 (2000)
789:
790:
791:
792: \bibitem[Iroshnikov (1963)]{Iroshnikov1963}
793: P. Iroshnikov,
794: Soviet Astron., {\bf 7}, 566 (1963)
795:
796:
797: \bibitem[Kraichnan (1965)]{Kraichnan1965}
798: R. Kraichnan, Phys. Fluids, {\bf 8}, 1385 (1965)
799:
800: \bibitem[Goldreich \& Sridhar (1995)]{Goldreich1995}
801: P. Goldreich and S. Sridhar,
802: \apj, {\bf 438}, 763 (1995)
803:
804:
805: \bibitem[Galtier et al.(2005)]{Galtier2005}
806: S. Galtier, A. Pouquet, and A. Mangeney,
807: Phys. Plasmas {\bf 12}, 092310 (2003)
808:
809: \bibitem[Matthaeus \& Zhou (1989)]{Matthaeus1989}
810: W.H. Matthaeus and Y. Zhou,
811: Phys. Fluids B {\bf 1} 1929 (1989)
812:
813:
814: \bibitem[Zhou, Matthaeus \& Dmitruk (2004)]{Zhou2004}
815: Y. Zhou, W.H. Matthaeus, P. Dmitruk
816: Rev. Mod. Phys., {\bf 76}, 1015 (2004)
817:
818: \bibitem[Bhattacharjee \& Ng (2001)]{Bhatta2001}
819: A. Bhattacharjee and C.S. Ng,
820: \apj, {\bf 548}, 318 (2001)
821:
822: \bibitem[Alexakis (2007)]{Alexakis2007}
823: A. Alexakis,
824: { arXiv:0706.0816v1} (2007)
825:
826:
827: \bibitem[Biskamp \& M{\"u}ller (2000)]{Biskamp2000}
828: D. Biskamp, and W.C. M{\"u}ller,
829: Phys. Plasmas, {\bf 7}, 4889 (2000)
830:
831: \bibitem[Cho \& Vishniac (2000)]{Cho2000}
832: J. Cho and E.T. Vishniac,
833: \apj {\bf 539}, 273 (2000)
834:
835: \bibitem[Dmitruk, Gomez, \& Matthaeus (2003)]{Dmitruk2003}
836: P. Dmitruk, D.O. Gomez, and W.H. Matthaeus,
837: Phys. Plasmas {\bf 10}, 3584 (2003)
838:
839: \bibitem[Mason, Cattaneo \& Boldyrev]{Mason2007}
840: J. Mason, F. Cattaneo and S. Boldyrev,
841: astro-ph arXiv:0706.2003v1 (2007)
842:
843: \bibitem[Maron \& Goldreich (2001)]{Maron2001}
844: J. Maron and P. Goldreich,
845: \apj {\bf 554}, 1175 (2001)
846:
847: \bibitem[M{\"u}ller \& Grappin (2005)]{Muller2005}
848: W.C. M{\"u}ller, and R. Grappin, Phys. Rev. Lett. {\bf 95}, 114502 (2005)
849:
850: \bibitem[Ng \& Bhattacharjee (1996)]{Ng1996}
851: C.S. Ng and A. Bhattacharjee, \apj, {\bf 465}, 845 (1996)
852:
853: %\bibitem[Ng, Bhattacharjee (1997)]{Ng1997}
854: % Ng, C. S., Bhattacharjee, A. 1997, Phys. Plasmas, 4, 605
855:
856: \bibitem[Ng, Bhattacharjee, \& Germaschewski (2003)]{Ng2003}
857: C.S. Ng, A. Bhattacharjee, K. Germaschewski and S. Galtier,
858: Phys. Plasmas, {\bf 10}, 1954 (2003)
859:
860: \bibitem[Oughton, Priest, \& Mattaheus (1994)]{Oughton94}
861: S. Oughton, E.R. Priest and W.H. Mattaheus, J. Fluid Mech. {\bf 280}, 95 (1994).
862:
863:
864: \bibitem[Alexakis (2005)]{Alexakis2005b}
865: A. Alexakis, P.D. Mininni and A. Pouquet,
866: Phys. Rev. Lett. {\bf 95}, 264503 (2005)
867:
868: \bibitem[Mininni, Alexakis \& Pouquet (2005)]{Mininni2006}
869: P.D. Mininni, A. Alexakis and A. Pouquet,
870: Phys. Rev E {\bf 74} 016303 (2006)
871:
872: \bibitem[Milano et al.(2001)]{Milano}
873: L.J. Milano, W.H. Matthaeus, and P. Dmitruk, D.C. Montgomery
874: Phys. Plasmas {\bf 8}, 2673-2681 (2001).
875:
876: \bibitem[Debliquy, Verma \& Carati (2005)]{Debliquy2005}
877: O. Debliquy, M.K. Verma and D. Carati,
878: Phys. Plasmas {\bf 12}, 042309 (2005)
879:
880: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
881:
882: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
883:
884:
885:
886:
887: \end{thebibliography}
888:
889:
890: \end{document}
891:
892: %%
893: