1: \documentclass[twocolumn,aps,prl,draft,showpacs,amssymb]{revtex4}
2: \usepackage{graphicx}
3: \usepackage{bm}
4: \begin{document}
5: \title{Multi-particle dispersion in fully developed turbulence}
6: \author{L.~Biferale$^1$, G.~Boffetta$^2$, A.~Celani$^3$, B.J. Devenish$^1$,
7: A.~Lanotte$^4$, and F.~Toschi$^5$} \affiliation{ $^1$ Dept. of Physics
8: and INFN,
9: University of ``Tor Vergata'', Via
10: della Ricerca Scientifica 1, 00133 Roma, Italy \\ $^2$ Dipartimento di
11: Fisica Generale and INFN, Universit\`a degli Studi di Torino, Via
12: Pietro Giuria 1, 10125, Torino, Italy \\
13: and CNR-ISAC, Sezione di Torino, Corso Fiume 4, 10133 Torino, Italy\\
14: $^3$ CNRS, INLN, 1361 Route
15: des Lucioles, 06560 Valbonne, France \\ $^4$ CNR-ISAC, Sezione di
16: Lecce, Str. Prov. Lecce-Monteroni km. 1.200, 73100 Lecce, Italy\\ $^5$
17: Istituto per le Applicazioni del Calcolo, CNR, Viale del Policlinico
18: 137, 00161 Roma, Italy} \date{\today}
19: \begin{abstract}
20: The statistical geometry of dispersing Lagrangian clusters of
21: four particles (tetrahedra) is studied by means of high-resolution direct numerical
22: simulations of three-dimensional homogeneous isotropic turbulence.
23: We give the first evidence of a self-similar regime of shape dynamics
24: characterized by almost two-dimensional, strongly elongated geometries.
25: The analysis of four-point velocity-difference statistics and
26: orientation shows that inertial-range eddies typically generate a
27: straining field
28: with a strong extensional component aligned with the elongation direction
29: and weak extensional/compressional components in the orthogonal plane.
30: \end{abstract}
31: \pacs{47.27-i}
32:
33: \maketitle
34: One of the most characteristic attributes of turbulence is the
35: efficient dispersion and mixing of advected Lagrangian particles \cite{MY75}.
36: Even though turbulent dispersion bears some similarities to
37: Brownian motion, especially at very large scales and for long times,
38: it has a much richer structure at small scales. This is already visible
39: at the level of single particle dispersion, which is
40: characterized by non-trivial time-correlations of the velocity
41: experienced by the particle along its trajectory (see e.g. \cite{OP}).
42: The statistics of pair dispersion display interesting properties as well
43: (see e.g. \cite{TP,BS02,BBCDLT05}), yet the complexity of Lagrangian turbulence
44: is particularly evident when looking at the dispersion of three or more particles.
45: This calls for the description of the geometrical properties of Lagrangian
46: dispersion -- the \lq \lq shape" of the particles' cloud as well as its
47: \lq \lq size". The geometrical characterization of dispersion proved
48: extremely important for the understanding of the problem of passive scalar advection
49: \cite{FGV01} and provides the basis for the
50: efficient modelling of the small-scale velocity dynamics itself
51: \cite{Pumiretal}. Previous studies dealt with
52: two-dimensional flows \cite{CV01,CP01}, synthetic flows \cite{FMV98,KPV03} or
53: three-dimensional turbulence at moderate Reynolds numbers
54: \cite{Pumiretal,BY98,YXBS02}.
55: %%%%%%%%%%%%%%%%%%%%% SIMULATION %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
56: In this Letter we study multi-particle Lagrangian statistics
57: by means of high resolution direct numerical simulations
58: of three-dimensional Navier-Stokes turbulence. Simulations were done
59: at resolutions of $1024^3$ corresponding to a Reynolds number
60: $R_\lambda \sim 280$ (see Ref.~\cite{BBCLT05}). The other parameters
61: of the numerical simulation are as follows: energy
62: dissipation $\varepsilon = 0.81 (8)$, viscosity $\nu = 8.8 \cdot 10^{-4}$, Kolmogorov length scale
63: $\eta= 5 \cdot 10^{-3}$, integral scale $L
64: = 3.14$, Lagrangian velocity autocorrelation time $T_L=1.2$, Kolmogorov time scale
65: $\tau_\eta = 3.3\cdot10^{-2}$.
66: % We have followed the trajectories of $3.8 \cdot 10^{5}$ particles
67: % arranged in $9.5\cdot 10^4$ regular tetraehdra.
68: % total
69: %integration time $T=4.4$ and the number of Lagrangian tracers $N_p=
70: %1.92 \cdot 10^6$.
71:
72: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
73: %\begin{table*}
74: %\begin{tabular}{cccccccccccc}
75: %$R_\lambda$ & $u_{rms}$ & $\varepsilon$ & $\nu$ & $\eta$ & $L$ & $T_L$
76: %& $\tau_\eta$ & $T$ & $\delta x$ & $N$ & $N_p$ \\ \hline 183 & 1.5 &
77: %0.886 & 0.00205 &0.01 & 3.14 & 2.1 & 0.048 & 5 & 0.012 & 512 &
78: %0.96$\cdot 10^6$ \\ 284 & 1.7 & 0.81 & 0.00088 &0.005& 3.14 & 1.8 &
79: %0.033 & 4.4 & 0.006 & 1024 & 1.92 $\cdot 10^6$ \\ \hline
80: %\end{tabular}
81: %\caption{{\bf check vs published table} Parameters of the numerical simulations. Microscale Reynolds
82: %number $R_\lambda$, root-mean-square velocity $u_{rms}$, energy
83: %dissipation $\varepsilon$, viscosity $\nu$, Kolmogorov length scale
84: %$\eta=(\nu^3/\varepsilon)^{1/4}$, integral scale $L$, large-eddy
85: %turnover time $T_L = L/u_{rms}$, Kolmogorov time scale
86: %$\tau_\eta=(\nu/\varepsilon)^{1/2}$, total integration time $T$, grid
87: %spacing $\delta x$, resolution $N$, and the384 number of Lagrangian
88: %tracers $N_p$.}
89: %\label{tab1}
90: %\end{table*}
91: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
92: With the present choice of parameters the dissipative
93: range of length scales is exceptionally well resolved.
94: Upon having reached a statistically stationary velocity field,
95: the Lagrangian tracers were seeded in
96: the flow. Their trajectories were integrated according to
97: $$
98: \frac{d {\bm x}}{dt} = {\bm u}({\bm x}(t),t)
99: $$
100: over a time lapse of the order of a few Lagrangian correlation times, $T_L$.
101: The velocity field, ${\bm u}$, results from the time-integration of
102: the three-dimensional Navier-Stokes equations (for further
103: details see Ref.~\cite{BBCLT05}).
104:
105: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% SHAPE CHARACTERIZATION %%%%%%%%%%%%%%%%%%
106: A set of $3.84 \cdot 10^{5}$ particles were initially seeded
107: in quadruplets forming $9.6\cdot 10^4$ regular tetrahedra of the
108: size of the Kolmogorov scale, with centers of mass
109: uniformly distributed over the domain.
110: The evolution of the separations between different particles in each
111: tetrahedron provides a way to quantify the shape evolution.
112: As particles move with the flow the size of the tetrahedra grows in
113: time and their shape deforms, generating
114: a variety of irregular objects. A description of this process
115: is then given in terms of the probability density functions (pdf) of sizes
116: and shapes. Within the inertial range of scales a self-similar evolution
117: of size according to Richardson's law and a stationary shape distribution are expected.
118: Figure~1 shows a sample of the tetrahedra evolving in the turbulent flow.
119: The presence of very different shapes, from almost regular to very flat and elongated
120: involving the interaction of diverse scales, is evident.
121:
122: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
123: \begin{figure}[hbt]
124: \includegraphics[draft=false,scale=0.44]{tetra.ps}
125: \caption{Snapshot at $t= 0.65T_L$ of $480$ tetrahedra evolving in the turbulent
126: flow starting from regular tetrahedra at the Kolmogorov scale.
127: }
128: \label{fig1}
129: \end{figure}
130: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
131: In order to characterize the shape dynamics quantitatively, it is useful to introduce
132: the following change of coordinates \cite{Pumiretal}:
133: ${\bm \rho}_0=({\bm x}_1+{\bm x}_2+{\bm x}_3+{\bm x}_4)/2$
134: ${\bm \rho}_1=({\bm x}_2-{\bm x}_1)/\sqrt{2}$,
135: ${\bm \rho}_2=(2{\bm x}_3-{\bm x}_2-{\bm x}_1)/\sqrt{6}$,
136: ${\bm \rho}_3=(3{\bm x}_4-{\bm x}_3-{\bm x}_2-{\bm x}_1)/\sqrt{12}$.
137: By virtue of the statistical homogeneity of the velocity field as well
138: as of the initial distribution of the centers of mass, the Lagrangian statistics
139: do not depend on $\rho_0$. The information about the particle separations can be
140: embodied in the square matrix ${\bm \rho}$ whose columns are the three vectors
141: $\rho_i$ with $i=1,2,3$. Denoting by $g_i$ ($g_1 \ge g_2 \ge g_3$) the eigenvalues
142: of the moment of inertia matrix, ${\bm I}={\bm \rho}{\bm \rho}^{T}$
143: (that is positive defined), we have that the size of the tetrahedron is
144: %\begin{equation}
145: $r \equiv \sqrt{tr({\bm I})}=\sqrt{g_1+g_2+g_3}=\sqrt{\frac{1}{8}\sum_{i,j}|{\bm x}_i-{\bm x}_j|^2}$,
146: %\label{trace}
147: %\end{equation}
148: whereas
149: the volume can be expressed as
150: %\begin{equation}
151: $V= \frac{1}{3} \det({\bm \rho})=\frac{1}{3}\sqrt{g_1 g_2 g_3}$.
152: %\end{equation}
153: A convenient characterization of shapes is given in terms of the
154: dimensionless quantities $I_i=g_i/r^2$ (where obviously $I_1+I_2+I_3=1$).
155: For a regular tetrahedron one has $I_1=I_2=I_3=1/3$. If the four points are
156: coplanar one has $I_3=0$ and for a collinear configuration $I_2=I_3=0$.
157:
158: Figure \ref{fig2} shows the temporal evolution of the mean
159: eigenvalues of ${\bm \rho}{\bm \rho}^{T}$ for the smallest
160: regular tetrahedra with $g_i(0)= \delta x^2/2$.
161: Two very different regimes are evident: at small times $t<\tau_{\eta}$ the
162: evolution of tetrahedra is governed by the dissipative range of turbulence.
163: Because of the smoothness and incompressibility of the velocity field
164: in this range, the volume of each tetrahedron is approximately
165: preserved and so is its average value which is shown in Fig.~\ref{fig2}.
166: In the viscous range the shape dynamics are essentially characterized
167: by the Lagrangian Lyapunov exponents \cite{FPV91}:
168: as a consequence the mean square separation $r^2$ grows
169: exponentially in time. From the average growth rate of the logarithms of the
170: separations, $R(t)=|{\bm \rho}_1|$, areas
171: $A(t)=\frac{\sqrt{3}}{2}|{\bm \rho}_1 \times {\bm \rho}_2|$ and volumes
172: $V(t)=\frac{1}{3}|{\bm \rho}_1 \times {\bm \rho}_2 \cdot {\bm \rho}_3|$
173: at small times, we can obtain an estimation of the Lagrangian Lyapunov spectrum
174: as shown in Fig.~\ref{fig2}.
175: We found two positive Lyapunov
176: exponents, with $\lambda_1 \tau_{\eta} \simeq 0.12$ and
177: $\lambda_2 \simeq \lambda_1/4$, in agreement with previous findings
178: at lower $R_\lambda$ \cite{GP90,PF04}. The sum of the
179: three Lyapunov exponents so obtained is close to zero for times
180: up to $3 \tau_\eta$.
181: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
182: \begin{figure}[hbt]
183: \includegraphics[draft=false,scale=0.68]{eigen.eps}
184: \caption{
185: Evolution of the mean eigenvalues $g_1$ ($+$), $g_2$ ($\times$) and $g_3$ ($*$) of the moment of
186: inertia matrix ${\bm I}={\bm \rho}{\bm \rho}^T$.
187: The line represents the dimensional scaling $t^3$.
188: In the inset, from top to bottom: evolution at small times
189: of $\langle \ln A(t) \rangle$ (surface), $\langle \ln R(t) \rangle$ (distance),
190: $\langle \ln V(t) \rangle$ (volume). The linear slopes of the three curves
191: in the range of times $\tau_{\eta} < t < 3 \tau_\eta$ yield $\lambda_1+\lambda_2$,
192: $\lambda_1$ and $\lambda_1+\lambda_2+\lambda_3$, respectively.
193: }
194: \label{fig2}
195: \end{figure}
196: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
197:
198: The exponential growth brings particle separations outside the
199: dissipative range, where the velocity field becomes rough
200: and the inertial range sets in.
201: According to the Kolmogorov-Richardson scaling,
202: eigenvalues should grow as $g_i \sim t^3$.
203: As previously reported \cite{Pumiretal}, it is hard to extract a
204: clear scaling regime for the shape dynamics shown in Fig.~\ref{fig2}.
205: The main reason for the lack of self-similarity is due to the
206: contamination of the inertial range by the dissipative range. Indeed,
207: because of the strong shape distortion taking place
208: at the crossover between the dissipative and inertial ranges
209: (as shown in Fig.~\ref{fig2} by the separation of the three eigenvalues),
210: a significant fraction of tetrahedra
211: has one side in the dissipative range even at times
212: much larger than $\tau_\eta$.
213: In order to overcome this problem we have utilized the technique
214: of doubling time statistics that has already been succesfully used to remove
215: contaminations in the statistics of pair dispersion
216: \cite{ABCCV97,BS02,BBCDLT05}.
217: Here, we focus on the doubling times of the eigenvalues $g_i$:
218: we compute the times, $T(g_i)$, taken by a tetrahedron to increase
219: its value of $g_i$ by a factor $a$. The result is shown in Fig.~\ref{fig3}.
220: The presence of a scaling range $T \sim g^{1/3}$ is more clear
221: and the self-similarity is made evident by superimposing the three curves
222: on top of each other by a simple multiplicative factor on the $g$-axis.
223: The ratio of the three eigenvalues in the scaling range is $g_1:g_2:g_3=
224: 40:8:1$, corresponding to shape indices $I_2 \approx 0.16$ and
225: $I_3 \approx 0.02$. The presence of a range where the
226: doubling times for different eigenvalues are the same is equivalent to
227: stating that the typical shape of the tetrahedron is preserved while
228: its size increases according to Richardson's law.
229:
230: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
231: \begin{figure}[hbt]
232: \includegraphics[draft=false,scale=0.68]{times_g.eps}
233: \caption{ Doubling times for the eigenvalues, $g_i$, of the moment of inertia
234: matrix, ${\bm \rho}{\bm \rho}^T$. In the inset: the same data rescaled on
235: the horizontal axis with the proportions $g_1:g_2:g_3 = 40:8:1$}
236: \label{fig3}
237: \end{figure}
238: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
239:
240: In view of the existence of a self-similar regime for
241: shape evolution, one would expect that the statistics of the
242: shape indices, $I_i$, should reach a time-independent distribution.
243: However, a direct inspection of the data does not support
244: this conclusion (not shown here, the results do not present
245: an appreciable scaling range in time in spite of the relatively
246: high $R_\lambda$ as compared with Ref.~\cite{Pumiretal}).
247: Once more this lack of a scaling range in the time domain can be
248: traced back to the contamination by the dissipative range dynamics.
249:
250: This difficulty can be overcome by selecting those tetrahedra
251: with eigenvalues in the ranges $ 5\cdot 10^2 \eta^2 < g_1 < 5\cdot 10^5
252: \eta^2$, $ 5\cdot 10^1 \eta^2 < g_2 < 5\cdot 10^4 \eta^2$,
253: $ 5\cdot \eta^2 < g_3 < 5\cdot 10^3 \eta^2$.
254: The thresholds are obtained by identifying the scaling ranges in
255: Fig.~\ref{fig3}. This procedure removes about 60\% of the initial
256: tetrahedra, mostly because $g_3$ falls below its lower threshold.
257: The probability density
258: functions of the shape indices $I_2$, $I_3$ after the selection
259: are shown in Fig.~\ref{fig4}.
260: The existence of an invariant regime appears now very clearly.
261: In this regime, the normalised probability density functions at different
262: times collapse, and the mean values hence display a
263: plateau in time: for the third index, the mean value
264: $\langle I_3 \rangle \simeq 0.011 \pm 0.001$
265: is not too far from the Gaussian value $0.03$, while
266: the second index is concentrated on values much smaller
267: ($\langle I_2 \rangle \simeq 0.135 \pm 0.003 $ as opposed to $0.22$).
268: Those values indicate a
269: relative abundance of flat and elongated configurations.
270: The tendency to form almost two-dimensional structures
271: has mostly an \lq \lq entropic" origin: indeed there is a large number
272: of pancake-like tetrahedra (very small $I_3$) already for Gaussian,
273: independent particle positions, as shown by the corresponding distribution
274: in Fig.~\ref{fig4}. However, it has to be remarked that
275: the pdf of $I_3$ is significantly more peaked at small values
276: than the Gaussian one.
277: The preference for elongated structures ($I_2 \ll I_1$) has a clear dynamical origin,
278: since it has no equivalent in the Gaussian ensemble.
279:
280: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
281: \begin{figure}[hbt]
282: \includegraphics[draft=false,scale=0.68]{pdficond.eps}
283: \caption{Probability density function of shape indices $I_2$
284: and $I_3$ (inset) at times $t=35 \tau_{\eta}$ ($+$) and
285: $t=63 \tau_{\eta}$ ($\times$). The full lines are the pdfs for independent,
286: Gaussian distributed particle positions.}
287: \label{fig4}
288: \end{figure}
289: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
290:
291: An interesting issue that we do not address here is connected with the
292: possibility of subleading, anomalous scaling in the
293: tetrahedra distribution. In the
294: simpler case of particles advected by a
295: Gaussian and white-in-time velocity field, it is known
296: that the asymptotic behaviour of the multi-particle pdf,
297: when the intial points are close, is governed by an expansion in zero
298: modes and slow modes of a given evolution operator \cite{FGV01}.
299: There, anomalous corrections emerge as sub-leading terms
300: to the Richardson scaling. These corrections are connected to the anomalous
301: scaling of the structure functions of a passive
302: scalar field advected by the flow. Here, in the presence of a real
303: turbulent flow, one can only argue that similar properties
304: may still hold \cite{CV01}. In order to check this, one should
305: perform a delicate
306: compensation between the evolution of the pdf with
307: different initial tetraehdra shapes, in order to cancel the leading scaling
308: terms and to highlight the sub-leading contributions.
309:
310: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% Velocity %%%%%%%%%%%%%%%%%%%%%%%%%%%%%
311: The dynamics of the shape evolution can be elucidated by analyzing
312: the local geometrical properties of Lagrangian velocities.
313: In analogy with the relative coordinates ${\bm \rho}$,
314: we introduce the relative velocity matrix ${\bm W}$:
315: ${\bm W}_1=({\bm u}_2-{\bm u}_1)/\sqrt{2}$,
316: ${\bm W}_2=(2 {\bm u}_3-{\bm u}_2-{\bm u}_1)/\sqrt{6}$,
317: ${\bm W}_3=(3 {\bm u}_4-{\bm u}_3-{\bm u}_2-{\bm u}_1)/\sqrt{12}$.
318: Obviously, $\dot{\bm \rho}={\bm W}$.
319: The geometrical aspects of Lagrangian velocity evolution
320: can be described by the tetrahedron \lq \lq turbulent diffusion'' tensor
321: \begin{equation}
322: {\bm K} \equiv \frac{1}{2}\frac{d}{dt} {\bm \rho} {\bm \rho}^T =
323: \frac{1}{2}({\bm W}{\bm \rho}^T + {\bm \rho}{\bm W}^T).
324: \label{diff}
325: \end{equation}
326: The trace $tr({\bm K})= \frac{1}{8}\sum_{i,j} ({\bm u}_i - {\bm u}_j)
327: \cdot({\bm x}_i-{\bm x}_j)$
328: is proportional to the longitudinal velocity difference multiplied by the separation
329: averaged over all pairs within the tetrahedron.
330: The geometrical information about the Lagrangian velocity
331: fluctuations may be obtained from the eigenvalues $\kappa_1 \ge \kappa_2 \ge \kappa_3$ of ${\bm K}$
332: which are shown in Fig.~\ref{fig5}.
333: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
334: \begin{figure}[ht]
335: \includegraphics[draft=false,scale=0.68]{diff.eps}
336: \caption{
337: Evolution of the mean eigenvalues of the \lq \lq turbulent diffusion
338: tensor'', ${\bm K}$, as a function of the tetrahedron size,
339: $r$. In the inset, the eigenvalues as a function of time.}
340: \label{fig5}
341: \end{figure}
342: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
343: On dimensional grounds these should grow in time
344: as $t^2$ or, equivalently, with the tetrahedron size, $r$,
345: as $r^{4/3}$: this is satisfied to a good accuracy
346: for all three eigenvalues, especially as a function of size.
347: The third eigenvalue, $\kappa_3$, is negative
348: (notice that, strictly speaking, this makes abusive the definition of
349: ${\bm K}$ as a diffusion tensor): geometrically this means that
350: the local velocity field experienced by the tetrahedron has
351: two extensional components, a strong one and a weak one,
352: $\kappa_1 \gg \kappa_2$, with the latter
353: smaller by a factor of ten than the former, and a weak compressional component
354: $|\kappa_3| \approx \kappa_2$.
355: It is also interesting to study the relative orientation
356: of the eigenvectors of the matrix ${\bm I}={\bm \rho}{\bm \rho}^T$,
357: i.e. the principal axes of inertia, and the eigenvectors of the matrix
358: ${\bm K}$. We found that the directions of the eigenvectors associated
359: with $g_1$ and $\kappa_1$ are preferentially aligned.
360: About $45\%$ of the
361: tetrahedra show a relative angle smaller than $\pi/6$ (for a uniform
362: distribution on a unit sphere one would have $13\%$).
363: This agrees with the intuitive idea
364: that strongly extensional velocity differences result in intense
365: elongations approximately in the same direction.
366: In the plane orthogonal to the
367: first principal axis of inertia,
368: the eigenvectors of ${\bm I}$
369: and ${\bm K}$ associated with the smaller eigenvalues
370: are also aligned albeit to a lesser degree (about $25\%$ of relative angles
371: below $\pi/6$).
372:
373: The overall geometrical picture that emerges
374: is the following: tetrahedra tend to be elongated, almost coplanar objects,
375: subject to a straining velocity field that has a strong extensional
376: part in the direction of elongation and relatively
377: weak compressive and extensional contributions in the orthogonal
378: plane of approximately equal magnitude.
379: The recent advances in experimental techniques
380: for particle tracking should soon allow
381: precise measurements of shape dynamics in real turbulent flows.
382: The joint effort on the numerical and experimental side can
383: shed further light on the geometrical statistics of Lagrangian
384: turbulence. This, in turn, will lead to the development of
385: new, more effective parameterizations of small-scale turbulence,
386: a problem of paramount importance for geophysical and industrial
387: applications.
388:
389: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
390: \begin{acknowledgments}
391: We are grateful to A. Pumir for discussions and useful suggestions.
392: The simulations were performed on the IBM-SP4 of the Cineca (Bologna,
393: Italy). We are grateful to C.~Cavazzoni and G.~Erbacci for resource
394: allocation and precious technical assistance.
395: \end{acknowledgments}
396:
397: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
398: \begin{thebibliography}{99}
399:
400: \bibitem{MY75} A. S. Monin and A. M. Yaglom, {\it Statistical Fluid Mechanics\/}
401: (MIT Press, Cambridge, MA 1975), vol 2.
402:
403: \bibitem{OP} N. Mordant {\em et al.}, Phys. Rev. Lett. {\bf 87}, 214501 (2001).
404: L. Chevillard {\em et al.}, Phys. Rev. Lett. {\bf 91}, 214502 (2003).
405: L. Biferale {\em et al.}, Phys. Rev. Lett.{\bf 93}, 064502 (2004).
406:
407: \bibitem{TP} P.K. Yeung, Phys. Fluids {\bf 6}, 3416 (1994).
408: S. Ott and J. Mann, J. Fluid Mech. {\bf 422}, 207 (2000).
409: T. Ishihara and Y. Kaneda, Phys. Fluids {\bf 14}, L69 (2002).
410: G. Boffetta and I.M. Sokolov, Phys Fluids {\bf 14}, 3224 (2002).
411: P.K. Yeung and M.S. Borgas, J. Fluid Mech. {\bf 503}, 93 (2004).
412:
413: \bibitem{BS02} G. Boffetta and I.M. Sokolov, Phys. Rev. Lett. {\bf 88}, 094501 (2002).
414:
415: \bibitem{BBCDLT05} L. Biferale {\em et al.}, submitted to Phys. Fluids, arxiv.org/0501054.
416:
417: \bibitem{FGV01} G. Falkovich, K. Gaw\c{e}dzki, and M. Vergassola, Rev. Mod. Phys. {\bf 73}, 913 (2001).
418:
419: %\bibitem{PSC99} M. Chertkov, A. Pumir and B.I. Shraiman, Phys. Fluids {\bf 11}, 2394 (1999) %\bibitem{PSC00} A. Pumir, B.I. Shraiman and M. Chertkov, Phys. Rev. Lett. {\bf 85}, 5324 (2000). %\bibitem{PSC01} A. Pumir, B.I. Shraiman and M. Chertkov, Europhys. Lett. {\bf 56}, 379 (2001) %\bibitem{PS03} A. Pumir and B. Shraiman, J. Stat. Phys. {\bf 113}, 693 (2003)
420:
421: \bibitem{Pumiretal}
422: M. Chertkov, A. Pumir and B.I. Shraiman, Phys. Fluids {\bf 11}, 2394 (1999),
423: A. Pumir, B.I. Shraiman and M. Chertkov, Phys. Rev. Lett. {\bf 85}, 5324 (2000),
424: A. Pumir, B.I. Shraiman and M. Chertkov, Europhys. Lett. {\bf 56}, 379 (2001),
425: A. Pumir and B. Shraiman, J. Stat. Phys. {\bf 113}, 693 (2003).
426:
427: \bibitem{CV01} A. Celani and M. Vergassola, Phys. Rev. Lett. {\bf 86},
428: 424 (2001).
429:
430: \bibitem{CP01} P. Castiglione and A. Pumir, Phys. Rev. E {\bf 64}, 056303 (2001).
431:
432:
433: \bibitem{FMV98} U. Frisch, A. Mazzino and M. Vergassola,
434: Phys. Rev. Lett. {\bf 80}, 5532 (1998).
435:
436: \bibitem{KPV03} K.A.I. Khan, A. Pumir and J.C. Vassilicos, Phys. Rev. E {\bf 68}, 026313 (2003).
437:
438: \bibitem{BY98} M.S. Borgas, Proc. 13th Australian Fluid Mech. Conf., Melbourne, (1998).
439:
440: \bibitem{YXBS02} P.K. Yeung {\em et al.}, IUTAM Symp. Reynolds
441: Number Scaling in Turb. Flow, Princeton, (2002).
442:
443: \bibitem{BBCLT05} L. Biferale {\em et al.}, Phys. Fluids {\bf 17}, 021701 (2005).
444:
445: \bibitem{FPV91} M. Falcioni, G. Paladin and A. Vulpiani, Riv. Nuovo Cimento {\bf 14}, 1 (1991).
446:
447: \bibitem{GP90} S.S. Girimaji and S. B. Pope, J. Fluid. Mech. {\bf 220}, 427 (1990).
448:
449: \bibitem{PF04} G. Falkovich and A. Pumir, Phys. Fluids {\bf 16}, L47 (2004).
450:
451: \bibitem{ABCCV97} V. Artale, G. Boffetta, A. Celani, M. Cencini and A. Vulpiani,
452: Phys. Fluids {\bf 9} 3162 (1997).
453:
454:
455: \end{thebibliography}
456: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
457: \end{document}
458:
459: