nlin0701017/mmm.tex
1: \documentclass[twocolumn,aps,prl,showpacs,amsmath,amssymb,nobalancelastpage]{revtex4}
2: \usepackage[dvips]{graphicx}
3: \usepackage{bm,eucal}
4: \hyphenation{sta-ti-stics}
5: \hyphenation{blo-wing}
6: \hyphenation{sca-ling}
7: \hyphenation{ave-ra-ging}
8: \begin{document}
9: %\topmargin=0.01cm
10: %
11: % Special commands
12: %
13: \newcommand{\nn}{\nonumber\\}
14: \newcommand{\rbr}[1]{\left(#1\right)}
15: \newcommand{\sbr}[1]{\left[#1\right]}
16: \newcommand{\cbr}[1]{\left\{#1\right\}}
17: \newcommand{\av}[1]{\prec #1 \succ}
18: %%%
19: %%% Latex-commands for environments
20: %%%
21: \newcommand{\eq}[1]{(\ref{#1})}
22: %%%
23: %%% Latex-commands for mathematical operations
24: %%%
25: \newcommand{\fou}[1]{\check{#1}}                           % Fourier transform
26: \newcommand{\fvol}[2]{\frac{d^{#2} #1}{(2\,\pi)^{#2}}}    % Volume in Fourier space
27: \newcommand{\de}{\partial}                                         % Partial derivative
28: \newcommand{\der}[2]{\frac{d #1}{d #2}}               % Total derivative
29: \newcommand{\pder}[2]{\frac{\partial #1}{\partial #2}} % Partial derivative
30: \newcommand{\fde}[2]{\frac{\delta #1}{\delta #2}}      % Total derivative
31: \newcommand{\fset}[2]{\chi_{#1}\left(#2\right)}   % Characteristic function of a set
32: \newcommand{\dirac}[2]{\delta^{(#1)}(#2)}         % Delta Dirac
33: \newcommand{\flu}[1]{\delta #1}                   % Variation
34: \newcommand{\vv}[1]{\mathbf{#1}}                  % Boldface vector
35: \newcommand{\uv}[1]{\hat{\mathbf{#1}}}                  % Boldface vector
36: %%%
37: %%% Notation for fields
38: %%%
39: \newcommand{\fphi}{\check{\phi}}
40: \newcommand{\bj}{\bar{\jmath}}
41: \newcommand{\bv}{\bar{v}}
42: \newcommand{\fv}{\fou{v}}
43: \newcommand{\fbv}{\fou{\bar{v}}}
44: %%%
45: %%% Definition of variables
46: %%%
47: \newcommand{\eps}{\varepsilon}
48: \newcommand{\mf}{m_{F}}
49: \newcommand{\dd}[1]{\eta_{#1}}
50: %%%
51: \newcommand{\ac}{\mathcal{A}}
52: \newcommand{\partfun}{\mathcal{Z}}
53: \newcommand{\mes}{\mathcal{D}}
54: %%%
55: \newcommand{\stf}[1]{\mathcal{S}_{#1}}
56: \newcommand{\cf}{\mathcal{C}}
57: %%%
58: \title{On the scaling properties of $2d$ randomly stirred Navier--Stokes equation}
59: \author{Andrea Mazzino}
60: \email{mazzino@fisica.unige.it}
61: \affiliation{Department of Physics, University of Genova, and  INFN, CNISM, Genova Section, 
62: Via Dodecaneso 33, I--16146 Genova, Italy,}
63: \author{Paolo Muratore-Ginanneschi}
64: \email{paolo.muratore-ginanneschi@helsinki.fi}
65: \affiliation{Department of Mathematics and Statistics, University of Helsinki, 
66: P.O. Box 4, 00014 Helsinki, Finland,}
67: \author{Stefano Musacchio}
68: \email{stefano.musacchio@gmail.com}
69: \affiliation{Department of Physics of Complex Systems
70: Weizmann Institute of Science
71: Rehovot 76100, Israel
72: %Institut Non Lin\'eaire de Nice,  1361 route des Lucioles
73: %Sophia Antipolis F06560 Valbonne, FRANCE.
74: } 
75: %\draft
76: \date{\today}
77: \begin{abstract}
78: We inquire the scaling properties of the $2d$ Navier-Stokes equation
79: sustained by a forcing field with Gaussian statistics, white-noise in time and 
80: with power-law correlation in momentum space of degree $2-2\,\eps$. This is at variance 
81: with the setting usually assumed to derive Kraichan's classical theory. We contrast accurate 
82: numerical experiments with the different predictions provided  for the small $\eps$ regime 
83: by Kraichnan's double cascade theory and by renormalization group (RG) analysis. 
84: We give clear evidence that for all $\eps$ Kraichnan's theory is consistent with the observed 
85: phenomenology. Our results call for a revision in the RG analysis of ($2d$) fully 
86: developed turbulence.
87: %As $\eps$ increases from zero, the forcing spectrum turns 
88: %from ultra-violet-dominated to infra-red.
89: %We present clear numerical evidences of the existence of an energy inverse cascade compatible
90: %with a spectral slope $-5/3$ for all values of $\eps$ less than $2$. 
91: %For $2 \leq \eps \leq 3$ the energy
92: %spectrum is compatible with local balance between energy input and non-linear terms. 
93: %Finally, for
94: %$\eps\geq 3$ we observe the onset of a direct enstrophy cascade.  
95: %The results are at variance with
96: %existing predictions based on renormalization group ananlysis 
97: %but admit a direct interpretation in terms
98: %of Kraichnan's theory of $2d$ turbulence. 
99: %We provide a simple argument accounting for the discrepancy.
100: \end{abstract}
101: \pacs{PACS number(s)\,: 47.27.Te, 47.27.$-$i, 05.10.Cc}
102: \maketitle
103: 
104: %There are remarkable differences in the qualitative properties of the 
105: %Navier-Stokes equation in two and $3d$.  
106: In two dimensions, the joint conservation of energy and enstrophy 
107: has relevant consequences for the Navier--Stokes equation.
108: In $2d$ it is possible to prove that in the deterministic case the solution of 
109: the Cauchy problem 
110: exists and is unique \cite{Lad69} and, very recently, 
111: that in the stochastic case \cite{KS00, BLK00, EMS01, BLK01, BLK02, Mat02} 
112: the solution is a Markov process exponentially mixing in time 
113: and ergodic with a unique invariant (steady state) measure 
114: even when the forcing acts only on two Fourier modes \cite{HM04}. 
115: At large Reynolds number, the $2d$ scaling properties also differ from the $3d$-case. 
116: A long standing hypothesis \cite{Pol92},  
117: very recently corroborated by numerical experiments \cite{BeBoCeFa06,BeBoCeFa06b}, 
118: surmises the existence of a conformal invariance in $2d$. 
119: A phenomenological theory due to Kraichnan \cite{Kr67} and Batchelor \cite{Bat69} 
120: predicts the presence of a double cascade mechanism governing the transfer 
121: of energy and enstrophy in the limit of infinite inertial range. 
122: Accordingly, an \emph{inverse energy cascade} with spectrum 
123: characterized by a scaling exponent $d_{\mathcal{E}}=-5/3$ appears for values of the 
124: wave-number $k$ smaller than the typical scale $k_{F}$ of the forcing scale.  
125: For wave-numbers larger than $k_{F}$ a \emph{direct enstrophy cascade} should occur. 
126: The corresponding energy spectrum scales as  $d_{\mathcal{E}}=-3+\dots$ 
127: where the dots here stand for possible logarithmic corrections.
128: As emphasized in \cite{Be99,Be00} Kraichnan's theory is encoded in three hypothesis, (i) 
129: velocity correlations are smooth at finite viscosity and exist in the inviscid limit even at 
130: coinciding points, (ii) Galilean invariant functions and in particular structure
131: functions reach a steady state and (iii) no dissipative anomalies occur for the energy cascade.
132: Under these hypotheses, if  the forcing field is homogeneous and isotropic Gaussian 
133: and time $\delta$-correlated it is possible to derive asymptotic expressions 
134: of the three point structure functions of the velocity field 
135: consistent with Kraichnan's predictions \cite{Be99,Be00,Lindborg}.
136: Very strong laboratory (see \cite{KeGo02,Ta02} and reference therein) and numerical evidences 
137: (see e.g. \cite{Bo06} and references therein) support Kraichnan' s theory. 
138: However, a first principle derivation of the statistical properties 
139: of two dimensional turbulence is still missing.
140: An attempt in this direction has recently been undertaken \cite{Ho98,Ho02} 
141: (see also \cite{Olla91} and \cite{AdHoKoVa05}) by applying a renormalization group 
142: improved perturbation theory (RG) (\cite{Zinn} and \cite{AdAnVa} 
143: for review of applications to fluid turbulence) to the randomly stirred Navier--Stokes equation 
144: with power law forcing. However how it will be discussed in details in the sequel, RG analysis
145: leads to a scenario not a priori consistent with Kraichnan's theory. The goal of the present work
146: is to shed light on this issue. 
147: 
148: RG starting point is the randomly stirred Navier--Stokes equation
149: \begin{eqnarray}
150: &&(\de_{t}+v\cdot\de)v^{\alpha}-\nu_{0}\,\de^2 v^{\alpha}=-\de^{\alpha}P 
151: + f^{\alpha}-\xi_{0}\,v^{\alpha}
152: \label{NS}
153: \\
154: && \de\cdot v=0 \nonumber
155: \end{eqnarray}
156: where $\alpha=1,2$,  $P$ is the pressure enforcing incompressibility 
157: and $\xi_{0}$ is an Eckman type coupling providing for large scale dissipation. 
158: The forcing $f$ is a Gaussian field with zero average 
159: and correlation
160: \begin{eqnarray}
161: &&\av{f^{\alpha}(\vv{x},t)f^{\beta}(\vv{y},s)}=\delta(t-s)\,F^{\alpha\,\beta}(\vv{x}-\vv{y})
162: \\
163: &&F^{\alpha\,\beta}(\vv{x}):= \int\fvol{p}{d}\,e^{\imath \vv{p}\cdot \vv{x}}\,\fou{F}(p)\,
164: T^{\alpha\,\beta}(\uv{p})
165: \label{correlation}
166: \\
167: &&\fou{F}(p):=\frac{F_{0}\,\chi\,\rbr{\frac{p}{m},\frac{p}{M}}}{p^{d-4+2\,\eps}}
168: %+ \frac{\nu_{0}^{3}\,a_{0}\,\chi_{2}\rbr{\frac{p}{m},\frac{p}{M}}}{p^{d-4}}
169: \label{correlation:core}
170: %\\&&T^{\alpha\,\beta}(p):=\delta^{\alpha\,\beta}-\frac{p^{\alpha}\,p^{\beta}}{p^2}
171: \end{eqnarray}
172: $F_{0}$ is a constant specifying the amplitude of forcing fluctuations 
173: and $T^{\alpha\,\beta}(p)$ is the transversal projector.
174: The function $\chi$ in \eq{correlation:core}  is slowly varying 
175: for $m\ll p \ll M$ and set infra-red $m$ and ultra-violet cut-offs $M$ scales for the forcing. 
176: Its detailed shape does not affect the scaling predictions of the RG.
177: Irrespectively of the spatial dimension $d$,  
178: the cumulative spectrum $F^{\alpha}_{\,\,\,\alpha}(0)$ of the forcing 
179: (Einstein convection for repeated vector indexes) 
180: diverges  for $0\,\leq\, \eps\, <\, 2$ as the ultra-violet cut-off $M$ tends to infinity 
181: hence providing for stirring at small spatial scales. 
182: For $\eps\,>\,2$,  $F^{\alpha}_{\,\,\,\alpha}(0)$ is dominated by small wave-numbers 
183: and thus describes infra-red stirring.  
184: At $\eps=0$, the scaling dimensions of the material derivative, 
185: dissipation and forcing in \eq{NS} coincide 
186: for a scaling dimension of the velocity field $d_{v}=1$ in momentum units. 
187: Thus $\eps=0$ provides a marginal limit around which scaling dimensions 
188: can be perturbatively determined with the help of ultra-violet RG. 
189: The main result \cite{Ho98,Ho02,AdHoKoVa05} is the existence of a 
190: non-Gaussian infra-red stable fixed point of the RG flow yielding 
191: for the energy spectrum the prediction
192: \begin{eqnarray}
193: \mathcal{E}(k)= \eps^{1/3} \,F_{0}^{2/3}\,k^{1-\frac{4\,\eps}{3}}\, R\rbr{\eps, %m/k,k_{b}/k}
194: \frac{m}{k},\frac{k_{b}}{k}}
195: \label{RGspectrum}
196: \end{eqnarray}
197: The adimensional function $R$ depends upon infra-red scales $m$ 
198: and $k_{b}=(\eps\, \xi_{0}^{3}/ F_{0})^{1/(6-2\eps)}$ 
199: and admits a regular expansion in powers of $\eps$ at the RG fixed point \cite{Ho98}. 
200: The resummation leading to $k^{1-\frac{4\,\eps}{3}}$ in \eq{RGspectrum} is derived from
201: the solution of the Callan--Symanzik equation \cite{Zinn} which in the present case requires 
202: scaling at finite $\eps$ to stem from the balance of the two terms in the material derivative 
203: with the forcing i.e. from the requirement of Galilean invariance alone.   
204: According to \cite{AdAnVa,Ho98,Ho02} composite operators at small $\eps$ 
205: does not induce any self-similarity breaking by the infra-red scales $m$ and $k_{b}$.  
206: The conclusion is that the spectrum should scale for small $\eps$ 
207: with exponent $d_{\mathcal{E}}=1-4\eps/3$ as in the $3\,d$-case \cite{AdAnVa}.
208: Such conclusion seems at variance with Kraichnan's theory which instead 
209: suggests the occurrence of an inverse cascade at small $\eps$. 
210: Namely  under the assumptions (i),(ii),(iii) \cite{Be99,Be99b} the energy balance equation
211: \begin{eqnarray}
212: \lefteqn{\rbr{\de_{t}+\xi_{0}-\nu_{0}\de^{2}}\av{v^{\alpha}(\vv{x},t)v_{\alpha}(0,t)}}
213: \nonumber\\
214: &&-\frac{1}{2}\de_{\mu}\av{||\flu{v}||^{2}(\vv{x},t)\,\flu{v}^{\mu}(\vv{x},t)}=F(\vv{x})
215: \label{enbal}
216: \end{eqnarray}
217: with $\flu{\vv{v}}^{\alpha}(\vv{x},t)=v^{\alpha}(\vv{x},t)-v^{\alpha}(0,t)$  
218: yields for $\xi_{0}=0$, $\eps<2$ and $r^2:=||x||^2$ 
219: \begin{eqnarray}
220: S_{3}(r)\simeq c_{1}F_{0}M^{4-2\eps}r+\frac{c_{2}\,F_{0}}{r^{3-2\eps}}\,,\quad  
221: m^{-1}\gg\,r\,\gg M^{-1} 
222: \label{s3}
223: \end{eqnarray}
224: $S_3$ denotes the three point velocity longitudinal structure function and $c_{i}$ $i=1,2$ two 
225: adimensional coefficients irrelevant for the present argument. 
226: Power counting based on \eq{s3} then predicts a $d_{\mathcal{E}}=-5/3$ inverse cascade spectrum 
227: at small $\eps$ with \emph{at most sub-leading} corrections consistent with the RG  prediction. 
228: %Such a discrepancy deserves to be inquired since it may highlight shortcomings 
229: %of theories whose inner consistency is otherwise widely believed.
230: 
231: The currently available numerical resources permit to contrast the two apparently 
232: discordant predictions \eq{RGspectrum} and \eq{s3} with the actual 
233: Navier--Stokes phenomenology.  
234: %In the sequel we shortly illustrate the numerical procedure used to achieve this goal. 
235: %The turbulent velocity field is obtained by direct integration of the 
236: To attain this goal we integrated the Navier--Stokes equation (\ref{NS})
237: for the vorticity field ($\omega =\epsilon_{\alpha\,\beta}\de^{\beta} v^{\alpha} $) 
238: with a standard, fully-dealiased pseudospectral method in a doubly periodic square 
239: domain of resolution $1024^2$.  
240: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
241: \begin{figure} [h!]
242: \centerline{
243: \includegraphics[scale=0.6,draft=false]{spettri.eps}}
244: \caption{Energy spectrum for $\eps=0.5$ and standard molecular dissipation. 
245: An inertial range with inverse energy cascade sets in for sufficiently 
246: small viscosity $\nu=O(10^{-4})$ (solid line). At larger viscosity the inertial range 
247: is suppressed and dissipative spectrum $\mathcal{E}\sim k^{1-2\,\eps}$ 
248: is observed (dashed line).}
249: \label{fig:1}
250: \end{figure}
251: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
252: %To avoid energy accumulation on the gravest wavenumber we used, 
253: %as customary, a linear damping term acting on wavenumbers close to $k=2$. 
254: Time evolution was computed by means of a standard second-order
255: Runge--Kutta scheme. 
256: %The integration has been carried out for one hundred turn-over 
257: %times after the velocity fields have reached the stationary state.
258: %Our steady state statistical ensemble  comprises around one hundred almost 
259: %independent snapshots equally spaced by half turn-over time. 
260: %Finally, 
261: We repeated our numerical experiments for different values 
262: of the  hyperviscosity $(-1)^{p+1}\nu_{0}\de^{2\,p}v$ including  $p=1$, 
263: standard viscosity as in \eq{NS}, and $p=4,6$ obtaining qualitatively identical results.
264: The outcomes evince that for \emph{all} $0\leq \eps \leq 2$
265: and sufficiently small viscosity, 
266: an inverse energy cascade with exponent consistent with $d_{\mathcal{E}}=-5/3$  
267: is observed (see Fig.~\ref{fig:1}).  
268: If the viscosity increases,  finite resolution effects prevent 
269: the observation  of an inertial range as the Kolmogorov scale becomes 
270: of the order of the infra-red dissipation scale.  
271: In such a case a dissipative spectrum appears with scaling exponent 
272: consistent with the prediction $d_{\mathcal{E}}=1-2\,\eps$ dictated 
273: by the balance between forcing and dissipation (see again Fig.~\ref{fig:1}). 
274: In agreement with \eq{s3}, local balance scaling 
275: becomes dominant in the range $2<\eps<\,3$ (see Fig. \ref{fig:2}). 
276: There, energy spectra scale with an exponent compatible with $d_{\mathcal{E}}=1-4\eps/3$. 
277: Finally, for $\eps> 3$ a direct enstrophy cascade invades the whole inertial range. 
278: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
279: \begin{figure} [h!]
280: \centerline{
281: \includegraphics[scale=0.6,draft=false]{struttura3.eps}}
282: \caption{Third order structure function of 
283: longitudinal velocity increments $S_3(r)$ for 
284: $\epsilon=1$ (squares), 
285: $\epsilon=2.5$ (circles), 
286: $\epsilon=4$ (triangles). 
287: When $\epsilon<2$ the presence of an inverse energy cascade 
288: with constant flux results in the scaling $S_3(r)\sim r$ (solid line).
289: When $\epsilon>3$ the presence of a direct enstrophy cascade 
290: with constant flux results in the scaling $S_3(r)\sim r^3$ (dashed line).
291: For $2<\epsilon<3$ the scaling is compatible with the RG argument (dash-dotted line)
292: }
293: \label{fig:2}
294: \end{figure}
295: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
296: The phenomenology just described  is confirmed by the inspection 
297: of the energy and enstrophy fluxes in the inertial range (see Fig.~\ref{fig:3}). 
298: They are respectively defined by
299: %$$
300: %\Pi_{E}(k)\propto \int_{k_{o}}^{k}\fvol{q}{2} 
301: %\Re\av{\mathcal{F}\cbr{v}^{\alpha}(-q)\mathcal{F}\cbr{(v\cdot\de %v_{\alpha})}(q)}
302: %\label{ensp}
303: %$$
304: %and
305: %$$
306: %\Pi_{Z}(k)\propto \int_{k_{o}}^{k} \fvol{q}{2} 
307: %\Re\av{\mathcal{F}\cbr{\omega}(-q)\mathcal{F}\cbr{(v\cdot\de %\omega)}(q)}
308: %\label{essp}
309: %$$
310: $
311: \Pi_{E}(k)\propto \int_{k_{o}}^{k}\fvol{q}{2} 
312: \Re\av{\mathcal{F}\cbr{v}^{\alpha}(-q)\mathcal{F}\cbr{(v\cdot\de v_{\alpha})}(q)}
313: $
314: and
315: $
316: \Pi_{Z}(k)\propto \int_{k_{o}}^{k} \fvol{q}{2} 
317: \Re\av{\mathcal{F}\cbr{\omega}(-q)\mathcal{F}\cbr{(v\cdot\de \omega)}(q)}
318: \label{essp}
319: $ where $\mathcal{F}$ denotes the Fourier transform and $k_{o}$ 
320: is chosen such that $\Pi_{E}(k)$ and $\Pi_{Z}(k)$ are positive quantities in the inertial range.
321: As spectra and structure functions, fluxes highlight the existence 
322: versus $\eps$ of three distinct scaling regimes the origin 
323: of which can be understood by contrasting \eq{enbal} with the 
324: energy and enstrophy inputs 
325: $I_{E}(k)\propto\int_{k_{o}}^{k}dq\,q\,\fou{F}_{v}(q)$ and 
326: $I_{Z}(k)\propto\int_{k_{o}}^{k}dq\,q\,\fou{F}_{v}(q)$.
327: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
328: \begin{figure} [h!]
329: %\centerline{
330: \includegraphics[scale=0.5,draft=false]{flusso_e1.eps}
331: \includegraphics[scale=0.5,draft=false]{flusso_e2.5.eps}
332: \includegraphics[scale=0.5,draft=false]{flusso_e4.eps}
333: %}
334: \caption{
335: The energy flux $\Pi_{E}$ (symbols) and the energy input $I_{E}(k)$ 
336: (thin line) for different values of $\epsilon$.
337: (a) $\epsilon=1$; (b) $\epsilon=2.5$; (c) $\epsilon=4$.
338: The energy input is defined as
339: $
340: I_{E}(k)\propto\int_{0}^{k}dq\,q\, \fou{F}_{v}(q)
341: $  for $\epsilon< 2$ and as
342: $
343: I_{E}(k)\propto\int_{k}^{\infty}dq\,q\,\fou{F}_{v}(q)
344: $  for $\epsilon > 2$.
345: In (b) the enstrophy input, 
346: %$I_{Z}(k)\propto \int_{0}^{k}dq\,q\,\fou{F}_{\omega}(q)$ 
347: $I_{Z}(k)\propto \int_{0}^{k}dq\,q^3\,\fou{F}(q)$ 
348: with and the enstrophy flux $\Pi_{Z}(k)$ are also plotted 
349: in order to show their balance at each inertial range scale.}
350: \label{fig:3}
351: \end{figure}
352: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
353: For $0\leq \eps <\,2$ both $I_{E}$ and  $I_{Z}$ are
354: peaked in the ultra-violet. Since stirring occurs mainly at small spatial scales  
355: an inverse energy cascade sets in the whole inertial range.  
356: In the range $2\,<\,\eps\,<\,3$ the energy input becomes infra-red divergent 
357: whilst $I_{Z}$ is still ultra-violet divergent in the absence of cut-offs.  
358: In this situation, the steady state is attained when the energy and enstrophy fluxes 
359: balance scale by scale the corresponding inputs  (see Fig.~\ref{fig:3}). 
360: In this $\eps$-range the energy spectrum scales in agreement with the exponent which 
361: can be extrapolated but no longer fully justified using the RG.  
362: Finally, for $\eps>3$ both the energy and enstrophy inputs become peaked in the infra-red. 
363: In such a case \cite{Be99,Be99b} the right hand sides of Eq. \eq{enbal} 
364: and of the analogous equation for the vorticity correlation 
365: admit a regular Taylor expansion for $m x \ll 1$. 
366: The hypotheses (i), (ii), (iii) thus recover  $S_{3}(x)\propto r^3$ 
367: i.e. Kraichnan's scaling for the \emph{direct enstrophy cascade} 
368: in agreement with our numerical observations. 
369: 
370: As far as intermittency is concerned, our numerics support 
371: normal scaling of velocity structure functions (Fig.~\ref{fig:4}). 
372: Higher order vorticity structure function are compatible 
373: with a weakly anomalous scaling (Fig.~\ref{fig:5}). 
374: These results provide an indirect positive test for  
375: the occurrence of a vorticity dissipative anomaly \cite{Be99b}. 
376: 
377: In conclusion, our numerics fully support the validity of Kraichnan's theory 
378: for all values of the H\"older exponent of power law forcing in $2d$. 
379: RG scaling prediction is not observed in the range where it was supposed to appear. 
380: It seems to us that this in not trivially a consequence 
381: of low Reynolds number ($O(\eps^{1/2})$ for $F_{0}=O(\eps)$ 
382: see discussion in section 9.6.4 of \cite{Frisch}) entailed by perturbative expansion.
383: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
384: \begin{figure} [h!]
385: %\centerline{
386: \includegraphics[scale=0.6,draft=false]{ess_e1.eps}
387: \includegraphics[scale=0.6,draft=false]{ess_e2.5.eps}
388: \includegraphics[scale=0.6,draft=false]{ess_e4.eps}
389: %}
390: \caption{
391: Structure function of longitudinal velocity increments 
392: $S_4(r)$ (squares) and 
393: $S_6(r)$ (circles) vs. $S_2(r)$. 
394: The lines represent the non-intermittent scalings
395: $S_4(r) \sim S_2(r)^2$ (solid line) and $S_6(r) \sim S_2(r)^3$ (dashed line).
396: (a) $\epsilon=1$; (b) $\epsilon=2.5$; (c) $\epsilon=4$}
397: \label{fig:4}
398: \end{figure}
399: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
400: Instead scaling seems to be related to the convolution of the response field
401: with the forcing kernel in the Schwinger-Dyson equation \cite{AdAnVa,Zinn} giving rise 
402: to a non-local scaling field of dimension $d_{O}=-1/3$ independently of $\eps$.  
403: This point however deserves more theoretical inquiry.
404: 
405: We are pleased to thank J.~Honkonen, A.~Kupiainen and P.~Olla 
406: for numerous friendly and enlightening discussions. 
407: Numerical simulations have been performed on the ``Turbofarm'' 
408: cluster at the INFN computing center in Torino.  A.M. was supported by
409: COFIN 2006 N.2005027808 and by CINFAI consortium, P.M.-G. by
410: centre of excellence {\em ``Geometric Analysis and Mathematical Physics''} of the 
411: Academy of Finland and by FP5 EU network contract HPRN-CT-2002-00300. 
412: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
413: \begin{figure} [h!]
414: \centerline{
415: \includegraphics[scale=0.6,draft=false]{sfz_e4.eps}}
416: \caption{
417: Structure function of vorticity increments 
418: $S_n^{(\omega)}(r)$ for $n=1,2,3,4$ (square, circles, triangles, poligons). 
419: In the inset we show the scaling exponents $\zeta_n$. Here $\epsilon =4$.
420: }
421: \label{fig:5}
422: \end{figure}
423: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
424: 
425: 
426: \begin{thebibliography}{99}
427: 
428: \bibitem{AdAnVa}
429: L.Ts.~Adzhemyan,  A.V.~Antonov and A.N.~Vasil\'ev
430: {\em The field theoretic renormalization group in fully developed turbulence} 
431: Gordon and Breach, Amsterdam (1999). 
432: \bibitem{Bat69}
433: G.K. Batchelor, Phys. Fluids, {\bf Supp II 12} 233 (1969).
434: \bibitem{Be99}
435: D.~Bernard, Phys. Rev {\bf E 60} 6184 (1999) and chao-dyn/9902010.
436: \bibitem{Be99b}
437: D.~Bernard,  SPhT-00-039 and chao-dyn/9904034.
438: \bibitem{Be00}
439: D.~Bernard, SPhT-00-093 and cond-mat/0007106. 
440: \bibitem{BeBoCeFa06}
441:  D.~Bernard, G.~Boffetta, A.~Celani and G.~Falkovich, Nature Physics {\bf 2} 124 (2006)
442: \bibitem{BeBoCeFa06b}
443:  D.~Bernard, G.~Boffetta, A.~Celani and G.~Falkovich, nlin.CD/0609069
444: \bibitem{BLK00}
445: J.~Bricmont, A.~Kupiainen and R.~Lefevere,
446: J. Stat. Phys., {\bf 100} 743  (2000) and math-ph/9912018.
447: \bibitem{BLK01}
448: J.~Bricmont, A.~Kupiainen and R.~Lefevere,
449: Comm. Math. Phys. {\bf 224} 65 (2001) and \texttt{www.ma.utexas.edu/mp\_arc/}
450: \bibitem{BLK02}
451: J.~Bricmont, A.~Kupiainen and R.~Lefevere,
452: Comm. Math. Phys. {\bf 230} 87 (2002) and  math-ph/0010028.
453: \bibitem{Bo06} G. Boffetta, submitted to Phys. Fluids (2006).
454: \bibitem{EMS01}
455: W.~E,  J. C.~Mattingly, and Y. G.~Sinai. Comm. Math. Phys. {\bf 224} 83 
456: (2001).
457: \bibitem{Frisch}
458: U.~Frisch, {\em Turbulence}, Cambridge Univ. Press, Cambridge, (1995).
459: \bibitem{Ho98}
460: J. Honkonen, Phys. Rev. {\bf E 58} 4532 (1998).
461: \bibitem{Ho02}
462: J.~Honkonen, Yu.S.~Kabrits and M.V.~Kompaniets, nlin/0201025.
463: \bibitem{AdHoKoVa05}
464: L. Ts.~Adzhemyan, J.~Honkonen, M. V.~Kompaniets and A. N.~Vasil\'ev, 
465: Phys. Rev. {\bf E 71}, 036305 (2005)
466: and nlin.CD/0407067.
467: \bibitem{KeGo02}
468: H.~Kellay and W.I.~Goldburg, Rep. Prog. Phys. {\bf 65} 845 (2002).
469: \bibitem{Kr67}
470: R.~Kraichnan, Phys. Fluids {\bf 10} (1967) 1417.
471: \bibitem{KS00}
472: S.~Kuksin and A.~Shirikyan. Comm. Math. Phys. {\bf 213} 291 (2000).
473: \bibitem{Lad69}
474: O. A.~Ladyzhenskaya
475: \emph{The Mathematical Theory of Viscous Incompressible Flow}, Gordon and
476: Breach, New York, 1969.
477: \bibitem{Lindborg}
478: E.~Lindborg, Phys. Fluids {\bf 11} 510 (1999).   
479: \bibitem{HM04}
480: M.~Hairer and  J.C.~Mattingly, Annals of Mathematics, {\bf 164} (2006) 993 and math.PR/0406087.
481: \bibitem{Mat02}
482: J.C.~Mattingly,  Comm. Math. Phys. {\bf 230} 421 (2002).
483: \bibitem{Olla91}
484: P.~Olla, Phys. Rev. Lett. {\bf 67} 2465 (1991).
485: \bibitem{Pol92}
486: A.~Polyakov, Nucl.Phys. B {\bf 396} 367 (1993) and hep-th/9212145
487: \bibitem{Ta02}
488: P.~Tabeling, Physics Reports {\bf 362} 1 (2002) 
489: \bibitem{Zinn}
490: J.~Zinn-Justin, {\em Quantum field theory and critical phenomena}, 
491: 4th ed. Clarendon Press,  Oxford (2002).
492: \end{thebibliography}
493: 
494: \end{document}
495: