1: \documentclass[aps,twocolumn,amsmath,amssymb,prl,showpacs]{revtex4}
2: \usepackage{graphicx}
3: \usepackage[dvips,colorlinks=true,linkcolor=blue]{hyperref}
4:
5: \begin{document}
6: %\draft
7: \title{Understanding Anomalous Transport in Intermittent Maps:\\
8: From Continuous Time Random Walks to Fractals}
9: \author{N. Korabel$^{1}$, A.V. Chechkin$^{2}$, R. Klages$^{1}$,
10: I.M. Sokolov$^{3}$, V.Yu. Gonchar$^{2}$}
11: \affiliation{$^1$Max-Planck-Institut f\"ur Physik komplexer Systeme,
12: N\"othnitzer Str.\ 38, D-01187 Dresden, Germany\\
13: $^2$Institute for Theoretical Physics NSC KIPT, Akademicheskaya st.\ 1, 61108 Kharkov, Ukraine\\
14: $^3$Institut f\"ur Physik, Humboldt-Universit\"at zu Berlin, Newtonstr.\ 15,
15: D-12489 Berlin, Germany}
16: \date{\today}
17: \begin{abstract}
18: We show that the generalized diffusion coefficient of a subdiffusive
19: intermittent map is a fractal function of control parameters. A modified
20: continuous time random walk theory yields its coarse functional form and
21: correctly describes a dynamical phase transition from normal to anomalous
22: diffusion marked by strong suppression of diffusion. Similarly, the
23: probability density of moving particles is governed by a time-fractional
24: diffusion equation on coarse scales while exhibiting a specific fine
25: structure. Approximations beyond stochastic theory are derived from a
26: generalized Taylor-Green-Kubo formula.
27: \end{abstract}
28: \pacs{05.45.Ac, 05.60.-k, 05.40.Fb}
29:
30: \maketitle
31:
32: The notion of {\it anomalous diffusion} derives from the fact that the mean
33: squared displacement (MSD) must not follow the law of normal diffusion, $<x^2>
34: \sim t^\gamma$ with $\gamma = 1$, but may be superdiffusive with $\gamma > 1$
35: or subdiffusive with $\gamma < 1$. Such anomalous dynamics has been
36: investigated theoretically and observed experimentally not only in amorphous
37: semiconductors, surface diffusion, turbulence, polymers and plasmas but also
38: in chemical, biological and economical problems
39: \cite{Bou90,Met00}. For this variety of applications it is desirable to have
40: a class of test systems at hand which are yet easy to handle. Here
41: periodically continued deterministic maps provide an important basis for
42: analytical and numerical investigations
43: \cite{Gei84,Gei85,Zum93,Zum93a,Sto95,Bar03}. Both the sub-
44: and the superdiffusive dynamics of such models was successfully described by
45: continuous time random walk (CTRW) approaches
46: \cite{Gei84,Zum93,Zum93a,Bar03}. This stochastic theory yields
47: non-Gaussian probability density functions (PDF) for anomalous diffusive
48: processes exhibiting stretched exponential decay in case of sub- and L{\'e}vy
49: power law decay in case of superdiffusion \cite{Zum93}.
50:
51: On the other hand, a {\em microscopic theory} of anomalous deterministic
52: transport explaining the origin of this behavior in terms of the theory of
53: dynamical systems is just beginning to evolve \cite{Sto95}. Surprises along
54: these lines were already encountered in normal diffusive maps, where the
55: diffusion coefficient was found to be a fractal function of control parameters
56: \cite{Kla95,Kla96}. This behavior was also reported for more complex systems
57: like the climbing sine map \cite{Kor02}, bouncing ball billiards \cite{Har01}
58: and coupled Josephson junctions \cite{Tan02}. The fractality can be understood
59: as a signature of long-range dynamical correlations that, due to topological
60: instabilities, change in a complicated manner under parameter variation
61: \cite{Kla02}. However, no such fractal structures were yet identified in
62: anomalous dynamics as modeled by the maps of Refs.\
63: \cite{Gei84,Gei85,Zum93,Zum93a,Sto95,Bar03}.
64:
65: In this Letter we focus on a subdiffusive map whose functional form on the
66: unit interval was introduced by Pomeau and Manneville for describing
67: intermittency \cite{Pom80},
68: \begin{equation}
69: \label{map}
70: x_{n+1}\equiv M_{a,z}(x_n) = x_n +a x_n^z,\;\; 0 \le x_n <\frac 12 .
71: \end{equation}
72: For $\frac 12\le x_n<1$ the map is defined according to
73: $M_{a,z}(-x)=-M_{a,z}(x)$. The translation $M_{a,z}(x + m) = M_{a,z} (x) + m$,
74: $m \in \mathbb{Z}$, completes the definition on the real line, see the inset
75: of Fig.\ \ref{Fig1} (a). The degree of nonlinearity $z\ge 1$ and $a
76: \ge 1$ in Eq.\ (\ref{map}) hold for the two control parameters. For $1 \le z
77: < 2$ this model leads to normal diffusion while for $z \ge 2$ its behavior is
78: anomalous \cite{Gei84,Zum93}. The anomalous regime results from the existence
79: of marginal fixed points located at all integer values of $x$. Thus, a typical
80: trajectory of the map consists of long laminar phases interrupted by chaotic
81: bursts. In contrast to Refs.\ \cite{Gei84,Zum93}, which focused on the time
82: dependence of the MSD for the particular parameter value of $a=2^z$, here we
83: study the behavior of the generalized diffusion coefficient (GDC) \cite{Met00}
84: \begin{equation}
85: \label{GDC_Def}
86: K := \lim_{n \rightarrow \infty} \frac{\left< x^2 \right>}{n^{\gamma}}\:
87: \end{equation}
88: ($<\ldots>$ denotes an ensemble average) for this map under variation of both
89: control parameters. We first focus on numerical simulations \cite{sim} of the
90: map Eq.\ (\ref{map}) and on the interpretation of the results within the CTRW
91: approach. CTRW theory models diffusion processes by sequences of jumps
92: interrupted by periods of waiting. Let the PDFs of waiting times and of jump
93: lengths be defined by $\phi(t)$ and $\lambda(x)$, respectively. Choosing these
94: functions appropriately, CTRW theory predicts that $\gamma=1$ for $1\le z<2$
95: and $\gamma=1/(z-1)$ for $2<z$ irrespective of the parameter $a$ of the map
96: \cite{Gei84,Zum93}. Indeed, for all $a$ we find excellent agreement between
97: these solutions and the values for $\gamma$ obtained from
98: simulations. Consequently, the analytical expressions for $\gamma$ are used
99: throughout our work.
100:
101: Simulation results for $K$ as a function of $a$ for fixed $z$ are presented in
102: Fig.\ \ref{Fig1}. Magnifications of Fig.\ \ref{Fig1} (a) shown in parts (b)
103: and (c) reveal self similar-like irregularities indicating a fractal parameter
104: dependence of $K$. Note particularly the structure marked by triangles, which
105: is repeated on finer and finer scales. The parameter values for these symbols
106: correspond to specific series of Markov partitions \cite{Kla95,Kla96}. In
107: Fig.\ \ref{Fig2} $K$ is depicted as a function of $z$ for different values of
108: $a$, again displaying highly non-monotonic parameter dependences.
109:
110: %----------------------------------------------------------------------
111: % Fig1
112: %----------------------------------------------------------------------
113: \begin{figure}[b]
114: \includegraphics[width=8.5cm]{Fig1_fctrw.eps}
115: \caption{The generalized diffusion coefficient $K$, Eq.\ (\ref{GDC_Def}),
116: as a function of $a$ for $z=3$. The curve in (a) consists of $1200$ points,
117: the dashed-dotted line displays the CTRW result $K_1$, Eqs.\
118: (\ref{jump_length}), (\ref{GDC_CTRW}). (b) (600 points) and (c) (200 points)
119: show magnifications of (a) close to the onset of diffusion. The dotted line in
120: (b) is the CTRW approximation $K_2$, Eqs.\ (\ref{int_jump_length}),
121: (\ref{GDC_CTRW}). The dashed line represents the first term of the TGK formula
122: Eq.\ (\ref{GDC_GK}). The triangles mark a specific structure appearing on
123: finer and finer scales. The inset depicts the model Eq.\ (\ref{map}). All
124: quantities here and in the following figures are dimensionless.}
125: \label{Fig1}
126: \end{figure}
127: %----------------------------------------------------------------------
128:
129: In a first step for explaining these curves we apply standard CTRW theory
130: \cite{Gei84,Zum93} by modifying this approach at
131: three points: Firstly, the waiting time PDF must be calculated according to
132: the grid of elementary cells indicated in Fig.\
133: \ref{Fig1} \cite{Kla96,Kla97} yielding
134: \begin{equation}
135: \phi (t)=a\left( 1+a(z-1)t\right) ^{-\frac z{z-1}}\:.
136: \label{waiting_time_pdf}
137: \end{equation}
138: Secondly, as a jump probability we use
139: \begin{equation}
140: \label{real_jump_PDF}
141: \lambda (x) = \frac{p}{2} \delta (\left| x \right| - l) + (1 - p) \delta (x)\:.
142: \end{equation}
143: The second term accounts for the fact that particles may stay in a box with a
144: probability of $(1 - p)$. This quantity is roughly determined by the size of
145: the escape region $p = ( 1 - 2 x_c )$ with $x_c$ as a solution of the equation
146: $x_c + a x^{z}_{c} = 1$. Thirdly, we introduce two definitions of a typical
147: jump length $l_i\:,i\in\{1,2\}$. Thus,
148: \begin{equation}
149: \label{jump_length}
150: l_1 = \left\{ | M_{a,z}(x) - x | \right\} \:
151: \end{equation}
152: corresponds to the actual mean displacement while
153:
154: %----------------------------------------------------------------------
155: % Fig2
156: %----------------------------------------------------------------------
157: \begin{figure}[t]
158: \includegraphics[width=8.5cm]{Fig2_fctrw.eps}
159: \caption{$K$ as a function of $z$ for $a=3.14, 4, 5$ consisting of $200$,
160: $250$ and $300$ points (bottom to top). Bold lines represent simulation
161: results for $n=10^4$ iterations, the dashed line is the CTRW solution $K_2$,
162: Eqs.\ (\ref{int_jump_length}), (\ref{GDC_CTRW}), for $n\to\infty$. The thin
163: line displays the first term of the TGK formula Eq.\ (\ref{GDC_GK}). At $z=2$
164: there is a transition from normal to anomalous diffusion. The inset describes
165: the time-dependence of $<x^2>$ around this value with $a=3.14$. Here dashed
166: lines correspond to $z=1.5, 1.9, 1.95$ (bottom to top), dashed-dotted lines to
167: $2.5, 2.1, 2.05$ (top to bottom) while the bold line depicts $z=2$. The dotted
168: line is proportional to $\ln(n)$.}
169: \label{Fig2}
170: \end{figure}
171: %----------------------------------------------------------------------
172: \begin{equation}
173: \label{int_jump_length}
174: l_{2} = \left\{ | [M_{a,z}(x)] | \right\} \:
175: \end{equation}
176: gives the coarse-grained displacement in units of elementary cells, as it is
177: often assumed in CTRW approaches. In these definitions $\left\{\ldots\right\}$
178: denotes both a time and ensemble average over particles leaving a box. The
179: modified CTRW approximation for the GDC resulting from these three changes
180: reads
181: \begin{equation}
182: K_i = \begin{cases}
183: p l_i^2 a^{\gamma} \sin (\pi \gamma) / \pi \gamma^{1+\gamma}, & 0 < \gamma < 1
184: \cr
185: p l_i^2 a (1-1/\gamma), & 1 \le \gamma < \infty \: .
186: \end{cases}
187: \label{GDC_CTRW}
188: \end{equation}
189: Fig.\ \ref{Fig1} (a) shows that $K_1$ well describes the coarse functional
190: form of $K$ for large $a$. $K_2$ is depicted in Fig.\ \ref{Fig1} (b) by the
191: dotted line and is asymptotically exact in the limit of $a\to0$. Hence, the
192: GDC exhibits a dynamical crossover analogous to the one found for normal
193: diffusion \cite{Kla96,Kla97,Kor02}. Let us now focus on the GDC at
194: $a=12,20,28,...$, which corresponds to integer values of the height
195: $h=[M_{a,z}(1/2)]$ of the map. Simulations reproduce, within numerical
196: accuracy, the results for $K_2$ by indicating that $K$ is discontinuous at
197: these parameter values. Due to the self similar-like structure of the GDC we
198: arrive at the conjecture that $K$ exhibits infinitely many discontinuities on
199: fine scales as a function of $a$, in contrast to the continuity of the CTRW
200: solution.
201:
202: We now switch back to the $z$-dependence of $K$, where CTRW theory predicts a
203: dynamical phase transition from normal to anomalous diffusion at $z=2$
204: \cite{Gei84,Zum93}. At the transition point one has $<x^2>\sim
205: n/\log(n)$. According to Eq.\ (\ref{GDC_Def}), Eq.\ (\ref{GDC_CTRW}) yields a
206: continuous transition in $z$, see $K_2$ in Fig.\ \ref{Fig2}. However, this
207: coarse functional form is obscured not only by fractal irregularities but also
208: by an ultra-slow convergence of the MSD in our simulations. Using CTRW theory,
209: we verified that {\em around} $z=2$ the MSD is determined by a series of
210: logarithmic corrections in time, where the lowest-order term is proportional
211: to $n^{\gamma}/\log n$ and dominates at time scales controlled by
212: $n\ll\tilde{n}\sim \exp(1/(1-\gamma))$. Hence, for $\gamma\neq 1$ and
213: $n\gg\tilde{n}$ simple power laws are recovered while for $\gamma=1$ this
214: logarithmic term survives in the infinite time limit. The inset of Fig.\
215: \ref{Fig2} exemplifies this behavior explaining deviations between CTRW theory
216: and the simulation results in the main figure. In other words, around $z=2$
217: both normal and anomalous diffusion are suppressed due to logarithmic
218: corrections in time leading to a vanishing $K$. We strongly suspect that such
219: a behavior of the GDC is typical for dynamical phase transitions in anomalous
220: dynamics altogether \cite{Zum93a}.
221:
222: The stochastic CTRW theory presented above describes only the coarse parameter
223: dependence of the GDC. This motivates us to develop an approach that incorporates
224: dynamical correlations. For this purpose we generalize the
225: Taylor-Green-Kubo formula (TGK) \cite{Remark2} for maps to anomalous
226: diffusion. We start by expressing Eq.\ (\ref{GDC_Def}) via sums over the
227: integer velocities $v_m=\left[x_{m+1}\right]-\left[x_m\right]$
228: \cite{Kla02}. However, in contrast to normal diffusion the anomalous dynamics
229: generated by Eq.\ (\ref{map}) is not stationary for $z \ge 2$
230: \cite{Gas88,Zum93a}. This is reminiscent in the non-existence of an invariant
231: probability density for this map, or mathematically, in the existence of
232: infinite invariant measures \cite{Aar}. Consequently, the resulting expression
233: cannot be simplified using time-translational invariance, and we get
234: \begin{equation}
235: K = \lim_{n
236: \rightarrow \infty} \frac{1}{n^{\gamma}} \left[ <\sum_{k=0}^{n-1} v_k^2> + 2
237: <\sum_{k=0}^{n-1} \sum_{l=1}^{n-1} v_k v_{k+l}> \right] \:.
238: \label{GDC_GK}
239: \end{equation}
240: Focusing on the first term only, simulations confirm that it is proportional
241: to $n^{\gamma}$. For $z<2$ the dynamics is ergodic, hence this term is equal
242: to $n <v_0^2>$, and we recover the random walk formula for normal diffusion
243: \cite{Kla96,Kla97}. For $z\ge 2$ only generalized ergodic theorems may hold
244: \cite{Aar}, and whether the CTRW result Eq.\ (\ref{GDC_CTRW}) can be derived
245: from the first term is a non-trivial question. Considering this term as an
246: approximation of $K$ for all $z$, the numerical results are depicted in Figs.\
247: \ref{Fig1} (b) and \ref{Fig2}. The comparison of this data with the simulation
248: values based on Eq.\ (\ref{GDC_Def}) indeed shows that this term already
249: provides a first step beyond the modified CTRW model: In the $a$ dependence it
250: reproduces the major irregularities of $K$ and even appears to follow the
251: discontinuities discussed above. However, we cannot say yet whether it yields
252: exact values for $K$ at integer heights.
253:
254: Eq.\ (\ref{GDC_GK}) thus provides a suitable starting point for a systematic
255: understanding of the fractal GDC beyond CTRW theory. We remark that already
256: the first term can be related to de Rham-type fractal functions, which
257: explains why it features irregularities \cite{Unp}. Since the series expansion
258: in Eq.\ (\ref{GDC_GK}) is exact, working out further terms one must recover
259: more and more structure in the GDC \cite{Kla96,Kla02,Kor02}. Note that the
260: correlation function in the second term depends on two times, which enables to
261: express $K$ in terms of {\em aged} \cite{Bar03} de Rham-type fractal functions
262: summing up to the exact value.
263: %----------------------------------------------------------------------
264: % Fig3
265: %----------------------------------------------------------------------
266: \begin{figure}[t]
267: \includegraphics[width=8cm]{Fig3_fctrw.eps}
268: \caption{Comparison of the PDF obtained from simulations of the map Eq.\
269: (\ref{map}) (oscillatory structure) with the analytical solution Eq.\
270: (\ref{Sol}) of the fractional diffusion equation Eq.\ (\ref{Frac_Dif_Eq})
271: (continuous line in the middle) for $z=3$ and $a=8$. The PDF was computed from
272: $10^7$ particles after $n=10^3$ iterations. For the GDC in Eq.\ (\ref{Sol})
273: the simulation result was used. The crosses (x) represent the numerical
274: results coarse grained over unit intervals. The upper and the lower curves
275: correspond to fits with a stretched exponential and a Gaussian distribution,
276: respectively. The inset depicts the PDF for the map on the unit interval with
277: periodic boundaries.}
278: \label{Fig3}
279: \end{figure}
280: %----------------------------------------------------------------------
281:
282: We now turn to the PDFs generated by the map Eq.\ (\ref{map}). The great
283: success of CTRW theory derives from the fact that it correctly predicts both
284: the power $\gamma$ and the form of the coarse grained PDF $P(x,t)$ of
285: displacements for a large class of models \cite{Zum93}. Correspondingly, the
286: diffusion process generated by Eq.\ (\ref{map}) is not described by an
287: ordinary diffusion equation but by a fractional generalization of it. Starting
288: from the CTRW model for the map Eq.\ (\ref{map}) discussed above, one can
289: derive the time-fractional diffusion equation
290: \begin{equation}
291: \frac{\partial ^\gamma P(x,t)}{\partial t^\gamma }\equiv
292: \int_0^tdt^{^{\prime }}(t-t^{^{\prime }})^{-\gamma }\frac{\partial P}{%
293: \partial t^{^{\prime }}}=D\frac{\partial ^2P}{\partial x^2}
294: \label{Frac_Dif_Eq}
295: \end{equation}
296: describing the long-time limit of the PDF of the coarse grained dynamics with
297: initial condition $P(x,0)=\delta(x)$ and $D=K\Gamma (1 +
298: \gamma)/2$, $0<\gamma <1$. The fractional derivative is understood in
299: the Caputo sense \cite{Mai97}. Time-fractional equations of such a form have
300: already been extensively studied by mathematicians \cite{Gor00}. We remark
301: that the two other fractional diffusion equations proposed in Refs.\
302: \cite{Bar03,Met00}, which are based on a Riemann-Liouville fractional
303: derivative, are equivalent to Eq.\ (\ref{Frac_Dif_Eq}) under rather weak
304: assumptions \cite{ACunp}. The solution of Eq.\ (\ref{Frac_Dif_Eq}) expressed
305: in terms of an M-function of Wright type
306: \cite{Mai97,Gor00} reads
307: \begin{equation}
308: P(x,t)=\frac 1{2\sqrt{D}t^{\gamma /2}}M\left( \xi,\frac \gamma 2\right)
309: \label{Sol}
310: \end{equation}
311: giving exactly the same asymptotics that was obtained in Ref.\ \cite{Zum93}
312: for small and large values of $\xi =|x|/\sqrt{D}t^{\gamma/2}$. By using a
313: series representation of $M$ it can be demonstrated \cite{ACunp} that this
314: form is equivalent to those expressed via H-functions \cite{Met00} or
315: one-sided extremal L\'evy stable distributions \cite{Bar03}. Fig.\ \ref{Fig3}
316: demonstrates an excellent agreement between the analytical solution Eq.\
317: (\ref{Sol}) and the PDF obtained from simulations for the map Eq.\ (\ref{map})
318: if the PDF is coarse grained over unit intervals. However, it also shows that
319: the coarse graining eliminates a periodic fine structure that is not captured
320: by Eq.\ (\ref{Sol}) \cite{Kla96}. This fine structure derives from the
321: `microscopic' PDF of an elementary cell (with periodic boundaries) as
322: represented in the inset of Fig.\ \ref{Fig3}. The singularities are due to the
323: marginal fixed points of the map, where particles are trapped for long
324: times. Remarkably, that way the microscopic origin of the intermittent
325: dynamics is reflected in the shape of the PDF on the whole real line: From
326: Fig.\ \ref{Fig3} it is seen that the oscillations in the PDF are bounded by
327: two functions, the upper curve being of a stretched exponential type while the
328: lower is Gaussian. These two envelopes correspond to the laminar and chaotic
329: parts of the motion, respectively \cite{Remark3}.
330:
331: In conclusion, we have shown that the anomalous dynamics generated by a
332: paradigmatic subdiffusive one-dimensional map exhibits a fractal GDC under
333: variation of control parameters. The coarse dependence of this GDC and a
334: non-trivial phase transition from normal to anomalous diffusion are captured
335: by a modified CTRW theory. Near the phase transition point the GDC is strongly
336: suppressed by logarithmic corrections in time. A more detailed understanding
337: of the GDC is provided by an anomalous TGK formula suggesting intimate
338: relations to aging, fractal functions and ergodic theory. The coarse-grained
339: PDF of this anomalous dynamics is in excellent agreement with the solution of
340: a suitable fractional diffusion equation while on fine scales it reflects the
341: microscopic details of the intermittent dynamics. Here we have only treated a
342: subdiffusive map, however, we expect these findings to be typical for
343: spatially extended, low-dimensional, anomalous deterministic dynamical systems
344: altogether. Further studies will focus on a more detailed analysis of the
345: anomalous TGK-formula, on a spectral analysis of this map and on treating
346: superdiffusive dynamics along the same lines.
347:
348: The authors acknowledge helpful discussions with J. Klafter, R. Metzler and
349: A. Pikovsky. They thank the MPIPKS Dresden for hospitality and financial
350: support. IMS acknowledges partial financial support by the Fonds der
351: Chemischen Industrie.
352: %--------------------------------------------------------------------------------------------------------------------
353:
354: \begin{thebibliography}{Bou90}
355:
356: \bibitem{Bou90} J.-P. Bouchaud, A. Georges, Phys. Rep. {\bf 195}, 127 (1990);
357: M.F. Shlesinger, G.M. Zaslavsky, and J. Klafter, Nature (London) {\bf 263}, 31
358: (1993); J. Klafter, M.F. Shlesinger, G. Zumofen, Phys. Today {\bf 49}, 32
359: (1996); I.M. Sokolov, J. Klafter, A. Blumen, Phys. Today {\bf 55}, 48 (2002).
360:
361: \bibitem{Met00} R. Metzler, J. Klafter, Phys. Rep. {\bf 339}, 1 (2000).
362:
363: \bibitem{Gei84} T. Geisel, S. Thomae, Phys. Rev. Lett. {\bf 52}, 1936 (1984).
364:
365: \bibitem{Gei85} T. Geisel, J. Nierwetberg, A. Zacherl, Phys. Rev. Lett. {\bf 54}, 616 (1985).
366:
367: \bibitem{Zum93} G. Zumofen, J. Klafter, Phys. Rev. E {\bf 47}, 851 (1993).
368:
369: \bibitem{Zum93a} G. Zumofen, J. Klafter, Physica D {\bf 69}, 436 (1993).
370:
371: \bibitem{Sto95} R. Stoop, Phys. Rev. E {\bf 52}, 2216 (1995);
372: C. Dettmann, P. Cvitanovic, Phys. Rev. E {\bf 56}, 6687 (1997); Tasaki,
373: P. Gaspard, Physica D {\bf 187}, 51 (2004); R. Artuso, G. Cristadoro,
374: Phys. Rev. Lett. {\bf 90}, 244101 (2003).
375:
376: \bibitem{Bar03} E. Barkai, Phys. Rev. Lett. {\bf 90}, 104101 (2003).
377:
378: \bibitem{Kla95} R. Klages, J.R. Dorfman, Phys. Rev. Lett. {\bf 74}, 387
379: (1995); Phys. Rev. E {\bf 59}, 5361 (1999).
380:
381: \bibitem{Kla96} R. Klages, {\it Deterministic Diffusion in One-Dimensional
382: Chaotic Dynamical Systems} (Wissenschaft \& Technik Verlag, Berlin, 1996).
383:
384: \bibitem{Kor02} N. Korabel, R. Klages, Phys. Rev. Lett. {\bf 89}, 214102
385: (2002); Physica D {\bf 187}, 66 (2004).
386:
387: \bibitem{Har01} T. Harayama, P. Gaspard, Phys. Rev. E {\bf 64}, 036215
388: (2001); L. Matyas, R. Klages, Physica D {\bf 187}, 165 (2004).
389:
390: \bibitem{Tan02} K.I. Tanimoto, T. Kato, and K. Nakamura, Phys. Rev. B {\bf 66}, 012507 (2002).
391:
392: \bibitem{Kla02} R. Klages N. Korabel, J. Phys. A: Math. Gen. {\bf 35}, 4823 (2002).
393:
394: \bibitem{Pom80} Y. Pomeau, P. Manneville, Commun. Math. Phys. {\bf 74}, 189
395: (1980); P. Manneville, J. Physique {\bf 41}, 1235 (1980)
396:
397: \bibitem{sim} All simulations were performed starting from a uniform,
398: random distribution of $10^6$ initial conditions on the unit interval by
399: iterating for $n=10^4$ time steps.
400:
401: \bibitem{Kla97} R. Klages, J.R. Dorfman, Phys. Rev. E {\bf 55}, R1247 (1997).
402:
403: \bibitem{Remark2} The so-called Green-Kubo formula for diffusion was already
404: derived in G.I. Taylor, Proc. Lond. Math Soc. {\bf 20}, 196 (1921).
405:
406: \bibitem{Gas88} P. Gaspard, X.-J. Wang, Proc. Natl. Acad. Sci. {\bf 85},
407: 4591 (1988).
408:
409: \bibitem{Aar} J. Aaronson, LMS lecture notes {\bf 277}, 1 (2000); H. Hu,
410: L.-S. Young, Erg. Th. Dyn. Syst. {\bf 15}, 67 (1995).
411:
412: \bibitem{Unp} N. Korabel et al., in preparation.
413:
414: \bibitem{Mai97} F. Mainardi, {\em Fractional Calculus: Some Basic Problems in
415: Continuum and Statistical Mechanics.} in: A. Carpinteri and F. Mainardi
416: (Eds.), {\it Fractals and Fractional Calculus in Continuum Mechanics}, p.291
417: (Springer, New York, 1997).
418:
419: \bibitem{Gor00} R. Gorenflo, Y, Luchko, F. Mainardi, J. Comp. Appl. Math. {\bf 118}, 175 (2000).
420:
421: \bibitem{ACunp} I.M. Sokolov, A. Chechkin and J. Klafter, Acta Physica
422: Polonica B, in press.
423:
424: \bibitem{Remark3} The two envelopes shown in Fig.\ \ref{Fig3} represent
425: fits with the Gaussian $a_0 \exp (- x^2/a_1)$ and with the M-function $b_0 \;
426: M(\left| x \right|/ b_1, \frac{\gamma}{2})$, where $a_0=0.0036$, $a_1=55.0183$
427: and $b_0 = 3.05$, $b_1 = 0.37$.
428:
429: \end{thebibliography}
430:
431: \end{document}
432:
433: