1: %\documentclass[preprint,floatfix,amsmath,amssymb]{revtex4}

2: \documentclass[twocolumn,floatfix,amsmath,amssymb]{revtex4}

3: %\documentclass[pre,preprint,showpacs,floatfix,amsmath,amssymb]{revtex4}

4: \usepackage{graphicx} \usepackage{dcolumn} \usepackage{color}

5: 

6: \def\NOTE#1{{\textcolor{red}{\bf [#1]}}}  % note

7: \def\ADD#1{{\textcolor{blue}{#1}}}        % question

8: \def\AD#1{{\textcolor{magenta}{#1}}}      % plug a value, a ref, ...

9: \def\DEL#1{{\textcolor{green}{ #1}}}      % suggested deletion in text

10: \def\BB#1{{\textcolor{blue}{\bf #1}}}  

11: 

12: % table definitions

13: \def\hfq{\hfill\quad} \def\cc#1{\hfq#1\hfq}

14: \def\tvi{\vrule height 12pt depth 5pt width 0pt}

15: \def\traithorizontal{\noalign{\hrule}}

16: \def\tv{\tvi\vrule}

17: %

18: \newcommand{\be}{\begin{equation}} \newcommand{\ee}{\end{equation}}

19: \newcommand{\uh}{\hat{u}}

20: \def\u{{\bf u}} \def\v{{\bf v}} \def\x{{\bf x}} \def\dv{\delta {\bf v}}

21: %

22: \def\eg{{\it e.g.}\ } \def\etal{{\it et al.}\ } \def\ie{{\it i.e.}\ }

23: \def\lhs{{\it l.h.s.}\ } \def\op{{\it op. cit.}\ } \def\resp{{\it resp.}\ }

24: \def\rhs{{\it r.h.s.}\ } \def\rms{{\it r.m.s.}\ }

25: \def\viz{{\it viz.}\ } \def\vs{{\it vs.}\ } \def\vsk{\vskip0.05truein}

26: \def \v{\vspace*{1.2cm}} \def \h{\hspace*{2.5cm}} \def\ni{\noindent}

27: %

28: \def\al{Alfv\'en\ waves\ } \def\els{Els\"asser variables\ }

29: \def\kol{Kolmogorov\ } \def\mhe{MHD equations\ }

30: \def\nse{Navier-Stokes equations\ }

31: %

32: \def \pmbmath{\mathpalette\pmbmathaux}

33: \def \pmbmathaux#1#2{ \pmbtext{$#1#2$}}

34: \def \pmbtext#1{\leavevmode \setbox0\hbox{#1}

35:      \kern-0,2pt \copy0 \kern-\wd0 \kern0,4pt \copy0 \kern-\wd0

36:      \kern-0,2pt \raise0,3pt \box0 }

37: \def\pt{\partial_t} \def\px{\partial_x}

38: 

39: \newcommand{\curlv} {\nabla \times {\bf v}}

40: \newcommand{\ba}{\mathbf{a}} \newcommand{\bb}{\mathbf{b}}

41: \newcommand{\bA}{\mathbf{A}} \newcommand{\bB}{\mathbf{B}}

42: \newcommand{\Asz}{A_{s_z}}

43: \newcommand{\alp}{\alpha} \newcommand{\alpm}{\alpha^{-1}} 

44: \newcommand{\alpmm}{\alpha^{-1}_m} \newcommand{\alpmv}{\alpha^{-1}_v}

45: \newcommand{\bc}{\mathbf{c}} \newcommand{\bd}{\mathbf{d}}

46: \newcommand{\bj}{\mathbf{j}} \newcommand{\bk}{\mathbf{k}}

47: \newcommand{\bom}{\mbox{\boldmath $\omega$}}

48: \newcommand{\bomp}{\mbox{\boldmath $\omega^+$} }

49: \newcommand{\bomm}{\mbox{\boldmath $\omega^-$} }

50: \newcommand{\bompm}{\mbox{\boldmath $\omega^{\pm}$} }

51: % bold face

52: \newcommand{\br}{\mathbf{r}}

53: \newcommand{\bu}{\mathbf{u}}

54: \newcommand{\bv}{\mathbf{v}} \newcommand{\bw}{\mathbf{w}}

55: \newcommand{\bAs}{\mathbf{A_s}}  \newcommand{\bjs}{\mathbf{j_s}}

56: \newcommand{\bus}{\mathbf{u_s}}  \newcommand{\bBs}{\mathbf{B_s}}

57: \newcommand{\boms}{\mbox{\boldmath $\omega_s$}}

58: \newcommand{\bx}{\mathbf{x}} \newcommand{\bxp}{\mathbf{x^{\prime}}}

59: \newcommand{\bzp}{\mathbf{z^{+}}} \newcommand{\bzm}{\mathbf{z^{-}}}

60: \newcommand{\bzpm}{\mathbf{z^{\pm}}} \newcommand{\bzmp}{\mathbf{z^{\mp}}}

61: \newcommand{\ca}{{\rm a}} \newcommand{\caa}{{\rm aa}}

62: \newcommand{\cab}{{\rm a} b} \newcommand{\vba}{vb{\rm a}}

63: \newcommand{\K}{{\cal K}} \newcommand{\ud}{{\langle{u}^2 \rangle}}

64: \newcommand{\li}{\ell_{I}} \newcommand{\Rla}{R_{\lambda}}

65: \newcommand{\up}{{{\bf u}({\bf x})}} \newcommand{\R}{{\cal R}}

66: \newcommand{\dr}{{\partial_r}} \newcommand{\dt}{{\partial_t}}

67: \newcommand{\vg}{{{\bf v(x)} \cdot \nabla}}

68: %

69: %\topmargin -3pt

70: \begin{document}

71: \title{Small scale structures in three-dimensional magnetohydrodynamic turbulence}

72: 

73: \author{P.D. Mininni$^1$, A.G. Pouquet$^1$ and D.C. Montgomery$^2$}

74: \affiliation{$^1$ NCAR, P.O. Box 3000, Boulder, Colorado 80307 \\

75:  $^2$ Dept. of Physics and Astronomy, Dartmouth College, Hanover, 

76: NH 03755}

77: \date{\today}

78: \begin{abstract}

79: We investigate using direct numerical simulations with grids up to 

80: $1536^3$ points,  the rate at which small scales develop in a decaying 

81: three-dimensional MHD flow both for deterministic and random initial conditions.

82:  % at a unit magnetic Prandtl number. 

83: Parallel current and vorticity sheets form at the same spatial locations, 

84: and further destabilize and fold or roll-up after an initial exponential 

85: phase. At 

86: high Reynolds numbers, a self-similar evolution of the current and vorticity maxima  is found, in which they grow as a cubic power of time; the flow then reaches a finite 

87: dissipation rate independent of Reynolds number.

88: \end{abstract}

89: \maketitle

90: 

91: Magnetic fields are ubiquitous in the cosmos and play an 

92: important dynamical role, as in the solar wind,

93: stars or the interstellar medium.  Such flows have large Reynolds 

94: numbers and thus nonlinear mode coupling  leads to the formation of  %small-scale 

95: strong intermittent structures. 

96: It has been observed that such extreme events in magnetohydrodynamics (MHD) 

97: are more intense than for fluids; for example, wings of Probability 

98: Distribution Functions of field gradients are wider and one observes 

99: a stronger 

100: departure from purely self-similar linear scaling with the order of the 

101: anomalous exponents of structure functions \cite{carbone}.  

102: Since Reynolds numbers are high but 

103: finite, viscosity and magnetic resistivity play a role, 

104: tearing mode instabilities develop and reconnection takes place. 

105: The question then becomes at what rate does dissipation occur, as 

106: the Reynolds number increases? What is the origin of these structures, 

107: and how fast are they formed?

108: 

109: This is a long-standing problem in astrophysics, \eg in the context of reconnection events in the magnetopause, or of heating of solar and stellar corona. 

110: In such fluids, many other phenomena may have to be taken into account, 

111: such as finite compressibility and ionization, leading to a more complex Ohm's law 

112: with e.g. a Hall current %or ambipolar diffusion, or an anisotropic pressure gradient, 

113: or radiative or gravitational processes to name a few.

114: Many aspects of the two-dimensional (2D) case are understood, but the three-dimensional (3D) turbulent case 

115: remains more obscure.

116: Pioneering works \cite{green}

117: show that the topology of the reconnecting region,

118: more complex than in 2D, can lead to varied behavior.

119: 

120: The criterion for discriminating between a singular and a regular behavior 

121: in the absence of magnetic fields follows the seminal work by Beale, Kato 

122: and Majda (hereafter BKM) \cite{BKM} where, for a singularity to develop in the 

123: Euler case, the time integral of the supremum of the vorticity must grow as $(t-t_*)^{-\alp}$ with

124: $\alp \ge 1$ and $t_*$ the singularity time. 

125: In MHD \cite{BKM_MHD}, one deals 

126: with the Els\"asser fields $\bzpm=\bv\pm \bb$ and

127: $\bompm=\bom \pm \bj=\nabla \times (\bv \pm \bb)$, with $\bom$ the 

128: vorticity, $\bv$ the velocity, $\bj$ the 

129: current density and $\bb=\nabla \times \bA$ the induction in dimensionless Alfvenic units, $\bA$ being the vector potential. Intense current sheets are known 

130: to form at either magnetic nulls ($\bb\equiv 0$) or when one or two (but 

131: not all) components of the magnetic field go to zero or have strong 

132: gradients. In two dimensional configurations, a vortex quadrupole is 

133: also associated with these structures. The occurrence of singularities 

134: in MHD has 

135: been examined in direct numerical simulations (DNS), with either 

136: regular \cite{PPS95,kerrB1} or adaptive grids \cite{grauer}, and with 

137: different initial configurations with no clear-cut conclusions in view of 

138: the necessity for resolving a large range of scales (see \cite{kerr05} and 

139: references therein for the Euler case).  Laboratory experiments and DNS 

140: have also studied the ensuing acceleration of particles in the 

141: reconnection region (see e.g. \cite{gekelman}).

142: 

143: The early development of small scales in such flows is exponential \cite{syro}

144:  (in the context of turbulent flows, see \eg \cite{sanmin}),

145: because of the large-scale gradients of the velocity 

146: and magnetic fields, assumed given, stretching the vorticity and current. The phase beyond the linear stage,

147: though, is still unclear. In 2D, numerical simulations with periodic 

148: boundary conditions show that the late-time evolution of non-dissipative 

149: MHD flows remains at most exponential \cite{fpsm}, 

150: a point latter confirmed theoretically \cite{klapper1} by examining the 

151: structure around hyperbolic nulls, although finite dissipation seems to 

152: set in \cite{PPS89}.

153: 

154: In 3D, most initial conditions develop

155: sheets that may render the problem quasi two-dimensional locally;

156: 3D MHD flows display a growth of small scales of an exponential nature, although at later times 

157: a singular behavior may emerge \cite{kerrB1}. In this light, we address in this paper

158: the early development of structures in 3D and the ensuing evolution in the presence of dissipation.

159: \vskip0.03truein

160: 

161: The incompressible MHD equations read:

162: \begin{eqnarray}

163: &&{ \partial {\bf v} \over \partial t}+ {\bf v} \cdot \nabla {\bf v} =-{1 \over

164: \rho \sb{0}}\nabla {\cal P} + {\bf j} \times {\bf b} + \nu \nabla \sp{2} {\bf v}  

165: \nonumber\\

166: &&{ \partial {\bf b} \over \partial t}= \nabla \times ( {\bf v}\times

167: {\bf b})

168: +\eta \nabla \sp{2} {\bf b} 

169: %+\eta \nabla \sp{2} {\bf b}  \nonumber\\

170: %&&{\bf \nabla} \cdot {\bf v} =0 \ \ ,\ \  \nabla \cdot {\bf b} =0 \ \ ,

171: \label{MHDt}\end{eqnarray}

172: together with ${\bf \nabla} \cdot {\bf v} =0 ,  \nabla \cdot {\bf b} =0$;  ${\cal P}$ is the pressure, 

173: %$\bb = \bB /\sqrt {\mu_0 \rho_0}$ is the induction in Alfv\'enic units, $\mu_0$ the permeability, 

174: $\rho_0=1$ is the constant density, 

175: and $\nu$ and $\eta$ are  the kinematic viscosity and magnetic diffusivity.

176: With $\nu=0,\ \eta=0$, the energy 

177: $E=\left<v^2+b^2\right>/2$ and cross helicity 

178: $H_C=\left<{\bf v} \cdot {\bf b}\right>/2$, 

179: are conserved \cite{notez}, with the magnetic helicity 

180: $H_M=\left<{\bf A} \cdot {\bf b}\right>$ in 3D.

181: Defining 

182: $D_{\pm}/Dt=\partial _t + \bzpm\cdot\nabla$, one can symmetrize Eqs. (\ref{MHDt}) and obtain

183:  \cite{sanmin}:

184: \begin{eqnarray}

185: &&{D_{\mp}\bzpm\over Dt}\  = - \nabla {\cal P} \ \ ,\label{zpm}\\

186: &&{D_{\mp}\bompm\over Dt} = \bompm \cdot \nabla \bzmp

187: + \sum _m \nabla z^{\pm}_m \times \nabla z^{\mp}_m\ \ , \label{opm1}

188: \end{eqnarray}

189: omitting dissipation. Note that the first term on the \rhs of (\ref{opm1}) is equal to zero 

190: in 2D; the second term is absent in the Navier-Stokes case and may account for extra growth 

191: of the generalized vorticities for conducting 

192: fluids unless the Els\"asser field gradients are parallel.

193: 

194: To study the development of structures in MHD turbulence, we solve 

195: numerically Eqs. (\ref{MHDt}) using a pseudospectral method in a three 

196: dimensional box of side $2\pi$ with periodic boundary conditions. 

197: All computations are de-aliased, 

198: using the standard $2/3$ rule. With a minimum wavenumber of $k_{min}=1$ 

199: corresponding to $L_0=2\pi$, a 

200: linear resolution of $N$ grid points has a maximum wavenumber $k_{max}=N/3$. 

201: At all times we have $k_D/k_{max} < 1$, 

202: where $k_D$ is the dissipation wavenumber evaluated using the Kolmogorov 

203: spectrum (at early times the resolution condition is less stringent). 

204: Two different initial conditions are used; Table \ref{table:runs} 

205: summarizes all the runs.

206: 

207: As the system is evolved, we monitor the small scale development by following 

208: the dynamical evolution of the extrema 

209: of the generalized vorticities or of their individual components \cite{note1} 

210: in the spirit of the BKM criterion.

211: 

212: \begin{table}

213: \caption{\label{table:runs} Runs with an

214: Orszag-Tang vortex (OT1-4), or with large-scale ABC flows  

215: and small-scale random noise with a $k^{-3}$ spectrum (RND1-5);

216: $N$ is the linear resolution.

217: % $\nu$ the kinematic viscosity, and $\eta$ the magnetic diffusivity.

218: }

219: \begin{ruledtabular}

220: \begin{tabular}{ccc}

221: Run         & $N^3$      &$\nu=\eta$ \\

222: \hline

223: OT1 - OT4   & $64^3$ -- $512^3$  &$1\times 10^{-2}$ -- $7.5\times 10^{-4}$    \\

224: RND1 - RND4 & $64^3$ -- $512^3$  &$8\times 10^{-3}$ -- $6\times 10^{-4}$      \\

225: RND5        & $1536^3$     &$2\times 10^{-4}$     \\

226: \end{tabular} \end{ruledtabular} \end{table}

227: 

228: \begin{figure}

229: \centerline{\includegraphics[width=8.4cm]{fig1}}

230: \caption{(a) Evolution of the supremum of current for the 

231: OT runs in log-log. The inset shows the evolution at early times in 

232: lin-log units; a slope of 

233: $t^3$ is also indicated. The exponential phase ends at $t\sim 0.6$.  (b) Total dissipation as a function of time 

234: for the same runs in lin-lin. $R_e = 570$ (solid), $R_e = 1040$ (dot), 

235: $R_e = 3040$ 

236: (dash), and $R_e = 5600$ (dash-dot).}

237: \label{fig_OTdissrate}

238: \end{figure}

239: 

240: We start discussing the results for the Orszag-Tang vortex (OT hereafter); in two dimensions \cite{OT2d}, it has become a prototype flow for the study of MHD turbulence, including in the compressible case \cite{dahl1}. 

241: In 3D, the velocity and the magnetic field are taken to be:

242: \begin{eqnarray}

243: %$$

244: && \bv_{0}=[-2 \sin y,2 \sin x, 0] \ \ ,  \nonumber \\

245: %&& \bv_{O}=[-2\ \sin y,\ 2\ \sin x,\ 0] \\

246: %$$

247: %and

248: %$$

249: && \bb_{0}=\beta[-2 \sin 2y+ \sin z, 2\sin x + \sin z,\sin x +\sin y] .

250: %&& \bb_{O}=\beta[-2\ \sin 2y\ + \sin z, 2\sin x\ +\ \sin z,\ \sin x \ +\ \sin y] .

251: %$$

252: \nonumber \end{eqnarray}

253: 

254: The OT flow in 2D has a stagnation point in the $(x,y)$ plane 

255: and an hyperbolic X-point for the magnetic field; a 

256: 3D perturbation is added in the $z$ direction, 

257: resulting in a flow that has nulls for the magnetic field (three components 

258: equal to zero) of different types \cite{PF00}

259: corresponding to the signs of the eigenvalues of the $\partial_i b_j$ matrix \cite{PPS95}; 

260: initially, the kinetic and magnetic 

261: energy $E_V$ and $E_M$ are equal to 2 with $\beta =0.8$, 

262:  the normalized %velocity - magnetic field 

263: correlation 

264: $\tilde \rho^c=2\left<\bv \cdot \bb\right>/(\left<v^2+b^2\right>) \sim 0.41$, and $H_M=0$.

265: %and there is no magnetic helicity.

266: 

267: Four runs were done for spatial resolutions up to 

268: $512^3$. The Reynolds number $R_e=UL/\nu$ (where $U=\left<v^2\right>$ is 

269: the rms velocity, $L=2\pi \int E_V(k)k^{-1} dk/\int E_V(k)dk$ is the integral 

270: scale, and $E_V(k)$ is the kinetic energy spectrum) ranges from $570$ to 

271: $5600$ at the time of maximum dissipation of energy 

272: $\epsilon = -\nu \left< \omega^2 \right> -\eta \left< j^2 \right>$.

273: Figure \ref{fig_OTdissrate}(a) shows the temporal evolution of the 

274: maximum of the current $\max\{j\}$ (the vorticities $\bom$ and $\bompm$ 

275: behave in a similar fashion). 

276: After an initial exponential phase up to $t\sim 0.6$ and corresponding 

277: to the linear development of current (and vorticity) sheets through 

278: stretching by velocity gradients, a faster increase develops with, at 

279: high resolution, a self-similar $\sim t^3$ law. Note that the 

280: growth of $\max\{j\}$ during the early exponential phase seems to be 

281: independent of the value of $R_e$.

282: 

283: The first temporal maximum of $\max\{j\}$ is reached at slightly later times 

284: as $R_e$ increases; similarly [see 

285: Fig. \ref{fig_OTdissrate}(b)], 

286: the total energy dissipation 

287: $\epsilon$ shows a delay in the 

288: onset of the development of small scales as $R_e$ grows, 

289: reminiscent of 2D behavior \cite{PPS89},

290: with a slower global growth rate after the initial exponential 

291: phase; this delay, however, does not preclude reaching a quasi-constant 

292: maximum of $\epsilon$ in time as $R_e$ grows. Whereas in the 2D case, the 

293: constancy of $\epsilon$ only obtains at later times when reconnection 

294: sets in, with a multiplicity of current sheets, in the 3D case more 

295: instabilities of current and vorticity structures are possible 

296: and the flow becomes complex as soon as the linear phase has ended. 

297: %We verified that t

298: The dependence on $R_e$ of the time at which the first 

299: maximum of $\max\{j\}$ is reached  is slow ($\sim R_e^{0.08}$), and similarly for the time the maximum of 

300: $\epsilon$ is reached.

301: Computations at higher $R_e$ should be performed

302: to confirm these results.

303: 

304: \begin{figure}

305: \centerline{\includegraphics[width=8.4cm]{fig2}}

306: \caption{(a) Evolution of the supremum of the current density in 

307: log-log for runs RND1 to RND5, with $R_e=690$ (solid), $R_e=1300$ (dot), 

308: $R_e=2300$ (dash), $R_e=4200$ (dot-dash), and $R_e=10100$ (long dash). 

309: At high $R_e$, a power law consistent with $t^3$ is recovered. (b) Magnetic 

310: energy spectra at early times in run RND5. The lines (from below) 

311: correspond to $t=0.6$ up to $t=1.6$ with temporal increments of $0.2$. Slopes 

312: of $k^{-3}$ and $k^{-5/3}$ are indicated as a reference.}

313: \label{fig_RNDllogmaxc}

314: \end{figure}

315: 

316: The sharp transition around $t=0.6$ can be interpreted in terms of the 

317: non-locality of interactions in MHD turbulence \cite{alex_mhd} with transfer of energy involving widely 

318: separated scales. 

319: Thus, as the flow exits the linear phase, all scales can interact 

320: simultaneously; this may be a reason why, in a computation using OT and 

321: adaptive mesh refinement \cite{grauer}, it was found to be difficult 

322: to go beyond $t=0.6$ since small scales were developing abruptly in many 

323: places in the flow through prevailing nonlinearities. The energy spectra 

324: in this early phase are

325: steep, with a power law $\sim k^{-3}$ (not shown). A shallower $\sim k^{-1.70}$ spectrum develops

326: at later times, as found in earlier works.

327: 

328: In view of similarities between the behavior observed on the 

329: 3D OT vortex and its 2D counter-part, it is worth asking whether such 

330: a development is not due to the high degree of symmetry of the flow. In 

331: that light, we now examine the temporal development of a Beltrami flow 

332: on which small scale random fluctuations are added. 

333: The initial velocity and magnetic 

334: field spectra  are taken $\sim k^{-3} e^{-2{(k/k_0)}^2}$; the shells 

335: with $k\in [1,3]$ have a superposition of three ABC flows

336: \cite{ABC}, and the rest of the spectrum is loaded with Fourier 

337: modes with Gaussian random realizations for their phases chosen so that initially,

338:  $E_V=E_M=0.5$, ${\tilde {\rho}}^c\sim 10^{-4}$ and $H_M\sim 0.45$. 

339:  %Since random noise is added to the initial conditions, u

340: Unlike the OT runs, there are no 2D null 

341: points or exact zeros in the magnetic field. As for the OT case, four 

342: runs (RND1-RND4) were done with resolutions ranging from $64^3$ to 

343: $512^3$ grid points; RND5 on a grid of

344: $1536^3$ points is run until saturation of growth of the maximum current.

345: 

346: Both the exponential and the self-similar phases are noisier (see Fig. 

347: \ref{fig_RNDllogmaxc}a), as can be expected with several structures 

348: competing for the small scale development of maxima. At low $R_e$, 

349: self-similar evolution seems to occur at a slower pace, with laws 

350: $\sim t^2$, as in fact also found in 2D at comparable

351: resolutions. However, the two runs with highest 

352: resolution (RND4 and RND5) indicate a steeper development compatible 

353: with a $t^3$ law.

354: 

355: Figure \ref{fig_RNDllogmaxc}(b) shows the evolution of the magnetic 

356: energy spectrum $E_M(k)$ at early times, during the self-similar growth of the current 

357: density [the evolution of $E_V(k)$ is similar]. Before 

358: $t \sim 0.6$, the largest wavenumbers have amplitudes 

359:  of the order of the truncation error. For $t \ge 0.6$, as all scales are nonlinearly

360: excited, a self-similar growth sets in and the 

361: energy spectra are compatible with a $k^{-3}$ law. After $\max\{j\}$ %$j_{max}$ 

362: saturates, the slope of $E_M(k)$ increases slowly towards a $k^{-1.70}$ 

363: law. The same behavior is observed in the OT run, in which no 

364: $k^{-3}$ power law is imposed in the initial conditions.

365: 

366: \begin{figure}

367: \centerline{\includegraphics[width=8.5cm]{fig3}}

368: \caption{(Color online) Regions of strong current density, and magnetic 

369: field lines in their vicinity for run RND5 at $t=1.6$. The region at left has $450^2 \times 250$ points, and that at right has $260 \times 160 \times 200$. 

370: The sheets are thin and elongated (up to 1/3 the size of the box); the magnetic field lines are parallel to the sheet and quasi-orthogonal to each other on each side of it, and they depart from the sheet transversally. Both folding (left) and rolling (right) occurs at this $R_e$. Vortex sheets (not shown) are 

371: co-located and parallel to the current sheets.}

372: \label{fig_current2.jpg}

373: \end{figure}

374: 

375: The structures that develop appear to be accumulations of current sheets (similarly 

376: for the vorticity, not shown), as was already found in \cite{PPS95}. Figure 

377: \ref{fig_current2.jpg} shows a zoom on two such structures, with the magnetic 

378: field lines indicated as well. It appears clear from such figures that 

379: only one component of the magnetic field reverses sign in most of these 

380: configurations, reminiscent of magnetospheric observations.

381: Both terms appearing in Eq. (\ref{opm1}) for the dynamical evolution of 

382: $\bompm$ are substantial and comparable in magnitude although they may be quite weak elsewhere in the flow. 

383: Kelvin-Helmoltz instabilities with rolling up of such 

384: sheets are also present in the flow but only at the highest Reynolds number (run RND5); at lower $R_e$ the sheets are thicker, the instability is too slow and only folding of such sheets occur.

385: Magnetic field lines are parallel to 

386: the roll, in such a way that magnetic tension does not prevent the occurrence 

387: of the instability. Note that folding of magnetic 

388: structures has been advocated in the context of 

389: MHD at large magnetic Prandtl number \cite{cowley}. Alfvenization of the flow 

390: ($\bv = \pm \bb$) is rather strong in the vicinity of the sheets, with $0.7\le |{\tilde{\rho}}^c| \le 1$, although globally the flow remains uncorrelated 

391: (${\tilde{\rho}}^c\sim 4 \times 10^{-4}$);  this local Alfvenization gives stability to such structures since the nonlinear terms are weakened, in much the same way vortex tubes in Navier-Stokes flows are long-lived because of (partial) Beltramization ($\bv \sim \pm \bom$). Moreover, within the sheet ${\tilde{\rho}}^c$ is positive, and it is negative outside, with a slight predominance of $\bb$. All this indicates that a double velocity-magnetic field shear plays an important role in the development of small scales in MHD.

392: 

393: There is an elementary, analytically-soluble, one-dimensional model that 

394: illustrates sharply the role that velocity shear can play in enhancing 

395: current intensity, \eg during early dynamo activity \cite{montgo}. 

396: This consists of two semi-infinite slabs of rigid metal with equal 

397: conductivities, at rest against each other at the plane $y=0$, say. A 

398: uniform dc magnetic field ${\bf b}_0$ is perpendicular to the interface 

399: and penetrates both slabs. At time $t=0$, the slabs are impulsively 

400: forced to slide past each other in the $x$-direction with equal and 

401: opposite velocities (${\bf v}_0$, say). The developing (quasi-magnetostatic) 

402: field, which acquires an $x$-component, is a function of $y$ and $t$ only, 

403: and is governed by diffusion equations above and below the plane $y=0$. 

404: Matching tangential components of the electric field immediately above 

405: and below the interface reduces the pair to a soluble diffusion equation 

406: with a fixed $y$-derivative at $y=0$. The resulting magnetic field is 

407: expressible in terms of complementary error functions and grows without 

408: bound, as does the total Ohmic dissipation. The introduction of 

409: a time dependence in ${\bf v}_0$ may allow also for solutions 

410: in which the maximum of the current grows as a power law in time.

411: 

412: In conclusion, high resolution simulations of the early stages in the development of 

413: MHD turbulence allowed us to study the creation and evolution of small scale 

414: structures in three-dimensional flows. Roll up of current and vortex sheets, 

415: and a self-similar growth of current and vorticity maxima was found, features that 

416: to the best of our knowledge were not observed in previous simulations at 

417: smaller Reynolds numbers. Also, a convergence of the maximum dissipation 

418: rate to a value independent of $R_e$ was found. More analysis will be carried out

419: to understand how structures are formed, the relevance of the development of 

420: alignment between the fields and the creation and role of local exact solutions to the 

421: MHD equations (such as Alfv\'en waves).

422: 

423: %, and the relation between the roll-up observed at high $Re$ with hydrodynamic models (see e.g. \cite{gilbert}).

424: \vskip0.15truein

425: 

426: %\acknowledgments

427: {\it 

428: NSF grants CMG-0327888 and ATM-0327533 are acknowledged. Computer 

429: time provided by NCAR. Three-dimensional visualizations were done using VAPoR.

430: \cite{vapor}.}

431: 

432: \begin{thebibliography}{1}

433: \bibitem{carbone}

434: H. Politano, A. Pouquet, and V. Carbone, EuroPhys. Lett. {\bf 43}, 516 (1988).

435: 

436: \bibitem{green}

437: J. Greene, J. Geophys. Res. {\bf 93}, 8583 (1988).

438: 

439: \bibitem{BKM}

440: J. Beale, T. Kato, and A. Majda, Comm. Math. Phys. {\bf 94}, 61 (1984).

441: 

442: \bibitem{BKM_MHD}

443: R. Catflisch, I. Klapper, and G. Steele, Comm. Math. Phys. {\bf 184}, 443 

444: (1997).

445: 

446: \bibitem{PPS95}

447: H. Politano, A. Pouquet, and P.L. Sulem, Phys. Plasmas {\bf 2}, 2931 (1995).

448: 

449: \bibitem{kerrB1}

450: R. Kerr and A. Brandenburg, Phys. Rev. Lett. {\bf 83}, 1155 (1999); see also arXiv:physics/0001016 (2000).

451: 

452: \bibitem{grauer}

453: R. Grauer and C. Marliani, Phys. Rev. Lett. {\bf 84}, 4850 (2000).

454: 

455: \bibitem{kerr05}

456: R. Kerr, Phys. Fluids {\bf A17}, 075103 (2005).

457: 

458: \bibitem{gekelman}

459: N. Wild, W. Gekelman, and R. Stenzel, Phys. Rev. Lett., {\bf 46}, 339 (1981);

460: A.C. Ting, W.H. Matthaeus, and D. Montgomery, Phys. Fluids {\bf 29},

461: 3261 (1983);

462: Y. Ono, M. Yamada, T. Akao, T. Tajima and R. Matsumoto, Phys. Rev. Lett. 

463: {\bf 76}, 3328 (1996); 

464: D. Knoll and J. Brackbill, Phys. Plasmas {\bf 9}, 3775 (2002).

465: 

466: \bibitem{syro}

467: I. Syrovatskii, Sov. Phys. JETP {\bf 33}, 933 (1971).

468: 

469: \bibitem{sanmin}

470: A. Pouquet, in {\it $V^{th}$ European School in Astrophysics},

471: C. Chiuderi and G. Einaudi Eds, Springer-Verlag, Lecture Notes in Physics

472: {\bf 468}, 163 (1996).

473: 

474: \bibitem{fpsm}

475: U. Frisch, A. Pouquet, P.L. Sulem, and M. Meneguzzi,

476: J. M\'ecanique Th\'eor. Appl., {\bf 2}, 191 (1983).

477: 

478: \bibitem{klapper1}

479: I. Klapper, A. Rado and M. Tabor, Phys. Plasmas {\bf 3}, 4281 (1996).

480: %C. Bardos, in {\it Nonlinear Problems: Present and Future}, A. Bishop, D. Campbell and B. Nicolaenko (Eds.), North-Holland (1982).

481: 

482: \bibitem{PPS89}

483: D. Biskamp and H. Welter, Phys. Fluids {\bf B1}, 1964 (1989);

484: H. Politano, A. Pouquet, and P.L. Sulem, Phys. Fluids {\bf B1}, 2330 (1989).

485: 

486: \bibitem{notez}

487: Or their combinations in terms of the pseudo-energies of the \els $E^{\pm}=\left<{z^{\pm}}^2\right>/2$.

488: 

489: %\bibitem{kerrB2} R. Kerr and A. Brandenburg, arXiv:physics/0001016.

490: 

491: \bibitem{note1}

492: One may also monitor the temporal development of the symmetrized velocity 

493: and magnetic field gradient matrices, as well as that of the total 

494: enstrophy production \cite{kerrB1}.

495: 

496: \bibitem{OT2d}

497: S. Orszag, and C.-M. Tang, J. Fluid Mech. {\bf 90}, 129 (1979).

498: 

499: \bibitem{dahl1}

500: J. Picone, and R. Dahlburg, Phys. Fluids {\bf B3}, 29 (1991).

501: 

502:  \bibitem{PF00}

503: E. Priest and T. Forbes, ``Magnetic reconnection: MHD Theory and 

504: Applications'', Cambridge U. Press (2000).

505: 

506: \bibitem{alex_mhd}

507: A. Alexakis, P.D. Mininni, and A. Pouquet, 

508: Phys. Rev. E {\bf 72}, 046301; P.D. Mininni, A. Alexakis, and A. Pouquet,

509: Phys. Rev. E {\bf 72}, 046302.

510:  

511: \bibitem{ABC}

512: V.I. Arnol'd, C. R. Acad. Sci. Paris {\bf 261}, 17 (1965); D. Galloway and 

513: U. Frisch, Geophys. Astrophys. Fluid Dyn {\bf 36}, 58 (1986).

514: 

515: \bibitem{cowley}

516: A. Schekochihin {\it et al.},

517: %, J. Maron, S. Cowley and J. McWilliams

518: Astrophys. J. {\bf 576} 806 (2002).

519:  

520: %\bibitem{gilbert} A. Gilbert, Phys. Fluids {\bf A5}, 2831 (1993).

521: 

522: \bibitem{montgo}

523: D.C. Montgomery, P.D. Mininni, and A. Pouquet, Bull. Am. Phys. Soc., Ser. 

524: II, {\bf 50}(8), 177 (2005).   

525: 

526: \bibitem{vapor}

527: J. Clyne and M. Rast, in {\it Visualization and data analysis 2005}, 

528: R.F. Erbacher {\it et al.} %, J.C. Roberts, M.T. Grohn, and K. Borner 

529: (Eds.), SPIE, 

530: Bellingham, Wash. (2005), 284; http://www.vapor.ucar.edu.

531: 

532: \end{thebibliography}

533: \end{document}

534: