1: \documentclass[aps,preprint,showpacs]{revtex4}
2:
3: \usepackage{amsmath}
4: \usepackage{amsfonts}
5: \usepackage{amssymb}
6: \usepackage{graphicx}
7:
8: \setcounter{MaxMatrixCols}{10}
9:
10: %\input{tcilatex}
11:
12: \begin{document}
13:
14: \title{\textbf{Rotating Accelerator-Mode Islands}}
15: \author{\textbf{Oded Barash and Itzhack Dana}}
16: \affiliation{Minerva Center and Department of Physics, Bar-Ilan University,
17: Ramat-Gan 52900, Israel}
18:
19: \begin{abstract}
20: The existence of rotating accelerator-mode islands (RAIs), performing
21: quasiregular motion in rotational resonances of order $m>1$ of the standard
22: map, is firmly established by an accurate numerical analysis of all the
23: known data. It is found that many accelerator-mode islands for relatively
24: small nonintegrability parameter $K$ are RAIs visiting resonances of
25: different orders $m\leq 3$. For sufficiently large $K$, one finds also
26: ``pure'' RAIs visiting only resonances of the {\em same} order, $m=2$ or
27: $m=3$. RAIs, even quite small ones, are shown to exhibit sufficient
28: stickiness to produce an anomalous chaotic transport. The RAIs are basically
29: different in nature from accelerator-mode islands in resonances of the
30: ``forced'' standard map which was extensively studied recently in the
31: context of quantum accelerator modes.\newline
32: \end{abstract}
33:
34: \pacs{05.45.Ac, 45.05.+x, 05.45.Mt}
35: \maketitle
36:
37: \begin{center}
38: {\bf I. INTRODUCTION}\\[0pt]
39: \end{center}
40:
41: During the last three decades, the classical concept of ``accelerator mode''
42: (AM) has become of central importance in the theory of chaotic transport in
43: Hamiltonian systems \cite
44: {bvc,cm,krw,ikh,d,km,dh,ihkm,zen,bkwz,wbkz,ze,sd,shi,d1,kz,bd}. Recently,
45: this concept has also provided an illuminating explanation \cite
46: {fgr,qam,qam1} for a purely quantum acceleration of kicked atoms falling
47: under gravity, observed in atom-optics experiments \cite{ao}. AMs are
48: generalized periodic orbits (POs) of Hamiltonian maps having a translational
49: symmetry in phase space. A paradigmatic example is the standard map \cite
50: {bvc}
51: \begin{equation}
52: \ M:\ \ p_{t+1}=p_{t}+K\sin (x_{t}),\ \ \ \ x_{t+1}=x_{t}+p_{t+1}\ \text{mod}
53: (2\pi ), \label{map}
54: \end{equation}
55: where $p$ is angular momentum, $x$ is angle, $K$ is a nonintegrability
56: parameter, and $t$ is the ``integer'' time. The map (\ref{map}) and its
57: orbit structure are translationally invariant in $p$ with period $2\pi $ on
58: the cylindrical phase space $-\infty <p<\infty $, $0\leq x<2\pi $. This
59: allows to define consistently a PO of (minimal) period $n$ of (\ref{map}) in
60: a generalized fashion:
61: \begin{equation}
62: p_{t+n}=p_{t}+2\pi w,\ \ \ \ x_{t+n}=x_{t},\ \ \ \label{am}
63: \end{equation}
64: where $w$ is an integer, the ``jumping index''. For $w=0$, (\ref{am})
65: corresponds to a usual (closed) PO on the cylinder while for $w\neq 0$ the
66: PO is an AM with average acceleration $2\pi w/n$ per map iteration. If the
67: AM is stable, its $n$ points are surrounded by islands. Stickiness to the
68: boundaries of AM islands can lead to a superdiffusion of the chaotic motion,
69: $\left\langle p_{t}^{2}\right\rangle \propto t^{\mu }$, $1<\mu <2$ \cite
70: {ihkm,zen,bkwz,wbkz,ze,sd}, where $\left\langle \ \right\rangle $ denotes
71: ensemble average in the chaotic region.\newline
72:
73: AMs can arise only for sufficiently large $K$, $K>K_{{\rm c}}\approx 0.9716$,
74: when no rotational tori exist \cite{jg} and unbounded motion in the $p$
75: direction becomes then possible. Thus, AM islands are basically different in
76: nature from the well-known rotational-resonance islands which exist for
77: arbitrarily small $K$. The latter islands form the most basic component of
78: ordered and stable motion in the twist map (\ref{map}). They are associated
79: with the closed ($w=0$) ``Poincar\'{e}-Birkhoff'' or {\em ``ordered''} POs
80: \cite{ll,jdm} which are dynamically equivalent to pure rotations, i.e., the $
81: K=0$ POs, and emerge from them as $K$ is ``switched on'' \cite{ll}. Despite
82: the difference above, however, one may expect from the following general
83: arguments the existence of an interesting kind of AM islands, resembling
84: rotational-resonance islands in some aspects.\newline
85:
86: It is known \cite{mmp} that rotational resonances for the standard map (and
87: similar maps \cite{saw,bd1}) can be constructed in a well-defined way for
88: {\em all} $K$; a resonance of order $m$ is a chain of $m$ ``zones'' built on
89: a hyperbolic ordered PO of period $m$\ (see more details in Sec. II). Strong
90: numerical evidence \cite{mmp} and exact results \cite{saw,qc} indicate that
91: for $K>K_{{\rm c}}$ the resonances so constructed give a {\em partition} of
92: phase space. This implies that an arbitrary orbit of (\ref{map}) consists of
93: {\em quasiregular} segments within resonances, where each segment is a piece
94: of the orbit performing a number of rotations in one resonance \cite{d2,d3}.
95: Now, a general stability island must lie entirely in some resonance zone
96: \cite{bd} and will thus perform a similar quasiregular motion within
97: resonances. Clearly, the rotational quasiregularity is evident {\em only}
98: when the island visits resonances of order $m>1$. AM islands visiting $m>1$
99: resonances are most interesting objects since they exhibit a ``hybrid''
100: nature: In some time intervals, they rotate like $m>1$ resonance islands in
101: a near-integrable regime ($K\ll 1$) and at other times they accelerate due
102: to particular transitions between resonances occurring only for $K>K_{{\rm c}
103: }$. We thus call these islands, if they exist, ``rotating accelerator-mode
104: islands'' (RAIs). The RAIs should have a distinct impact on Hamiltonian
105: transport by generating a new kind of chaotic flight, featuring a
106: quasiregular steplike structure due to the ``horizontal'' rotation within
107: resonances. General ideas in Ref. \cite{bd} were illustrated only for the
108: most well-known AM islands of the standard map, those with central period $
109: n=1$ which emerge for $K>2\pi $ \cite{bvc,krw,ikh}. These islands lie within
110: $m=1$ resonances \cite{bd}. The question of the actual existence of RAIs was
111: not addressed in Ref. \cite{bd}.\newline
112:
113: In this paper, the existence of RAIs in the standard map is firmly
114: established by an accurate numerical analysis, examining also all the known
115: data on AM islands of which we are aware. In this analysis, the sequence of
116: resonances visited by an orbit is determined by using the efficient method
117: introduced in work \cite{d3}. A large fraction of the AM islands for $K<2\pi
118: $, listed in Ref. \cite{ikh}, are found to be RAIs visiting resonances of
119: different orders $m\leq 3$. Some of the significant peaks in the
120: chaotic-diffusion coefficient observed in Ref. \cite{ikh} for $K<2\pi $ are
121: due to RAIs. Among all the period-2 AM islands for $2\pi <K<20$, listed in
122: Ref. \cite{krw}, we have found RAIs visiting resonances of the {\em same}
123: order $m=2$ (``pure'' $m=2$ RAIs). We discover at $K\approx 8.916$ an
124: apparently new AM island, a pure $m=3$ RAI. It is shown that even quite
125: small RAIs exhibit sufficient stickiness to produce an anomalous chaotic
126: transport. Due to limitations in our available computational resources, we
127: were not able to find RAIs visiting resonances of order $m>3$. The paper is
128: organized as follows. In Sec. II, we briefly summarize the notion of
129: rotational quasiregularity within resonances. In Sec. III, the existence of
130: RAIs is established by an accurate determination of the quasiregularity
131: characteristics of many AM islands. In Sec. IV, we briefly study some of the
132: effects of RAIs on Hamiltonian chaotic transport. A discussion and
133: conclusions are presented in Sec. V, where we also consider the basic
134: difference between RAIs and AM\ islands visiting resonances of the
135: ``forced'' standard map [map (\ref{map}) with the addition of a constant
136: force], which has attracted much attention recently in the context of
137: ``quantum AMs'' \cite{fgr,qam,qam1,ao}.\newline
138:
139: \begin{center}
140: {\bf II. ROTATIONAL RESONANCES AND QUASIREGULARITY}\\[0pt]
141: \end{center}
142:
143: We briefly summarize here the definition of rotational resonances for the
144: standard map \cite{mmp} and the notion of quasiregularity within these
145: resonances \cite{bd,d2,d3}. Let us first recall the concept of rotationally
146: {\em ordered} POs \cite{jdm}. In the pure-rotation case of $K=0$, with
147: constant $p_{t}=p_{0}$, the sequence of orbit angles $x_{t}$ is given by $
148: x_{t}=x_{0}+p_{0}t$ mod($2\pi $). For rational winding number $\nu
149: =p_{0}/(2\pi )=l/m$, where $(l,m)$ are coprime integers, the orbit must be a
150: PO with period $m$; the $m$ PO points are uniformly distributed on the
151: circle $[0,\ 2\pi )$, i.e., the ``gap'' ${\bf G}_{t}$ between $x_{t}$ and a
152: neighboring point has the constant width $2\pi /m$, independent of $t$. The
153: rotational motion on the circle is expressed by the fact that ${\bf G}_{t+1}$
154: is also a gap, always separated from ${\bf G}_{t}$ by $|l|-1$ gaps. Now, for
155: $K\neq 0$, the winding number $\nu $ for a closed ($w=0$) PO is the average
156: value of $p_{t}/(2\pi )$ and $\nu $ is again rational. A gap is a pair of PO
157: points having neighboring values of $x_{t}$ in the circle. Then, a
158: rotationally ordered PO is a closed PO having the two main characteristics
159: of a $K=0$ PO: (a) $\nu =l/m$, where $m$ is the PO period and $(l,m)$ are
160: coprime. (b) If ${\bf G}_{t}$ is a gap, ${\bf G}_{t+1}$ is also a gap,
161: always separated from ${\bf G}_{t}$ by $|l|-1$ gaps. Unlike the case of $K=0$,
162: however, the gap width generally depends on $t$.
163: \newline
164:
165: An ordered hyperbolic PO with arbitrary winding number $\nu =l/m$ exists for
166: all $K$ \cite{jdm}. One gap of this PO, say ${\bf G}_{0}$, appears to be
167: always symmetrically positioned around the ``dominant'' symmetry line $x=\pi
168: $, i.e., $\pi -x_{L}=x_{R}-\pi $, where $L$ and $R$ denote the left and
169: right point, respectively, of ${\bf G}_{0}$. The $l/m$ resonance is now
170: defined, briefly, as follows (see more details in Refs. \cite{bd,mmp} and
171: refer to the examples in Fig. 1). One constructs in ${\bf G}_{0}$ a closed
172: region ${\cal Z}^{(0)}(l/m)$ bounded by four curved segments, which are
173: suitably chosen pieces of the stable and unstable manifolds of $L$ and $R$
174: under the map $M^{m}$. The $l/m$ resonance is then the chain of $m$ zones $
175: {\cal Z}^{(t)}(l/m)=M^{-t}{\cal Z}^{(0)}(l/m)$, $t=0,...,m-1$; see, e.g.,
176: resonances $0/1$ and $1/2$ in Fig. 1. Clearly, the zone ${\cal Z}
177: ^{(m)}(l/m)=M^{-m}{\cal Z}^{(0)}(l/m)$ lies again in ${\bf G}_{0}$ and
178: differs from the ``principal'' zone ${\cal Z}^{(0)}(l/m)$ by two {\em
179: turnstiles} created by homoclinic oscillations under $M^{-m}$ (the dashed
180: lines in Fig. 1). Each turnstile consists of two lobes of equal area. By
181: construction, the lobes outside (inside) ${\cal Z}^{(0)}(l/m)$ form the
182: region entering (exiting) resonance $l/m$ in one iteration of $M$.\newline
183:
184: Strong numerical evidence \cite{mmp,d3} and exact results \cite{saw,qc}
185: indicate that for $K>K_{{\rm c}}\approx 0.9716$ the resonances constructed
186: as above give, for all $l/m$, a complete partition of phase space. This
187: implies that a generic orbit must have all its points within resonances and
188: must therefore perform a quasiregular motion as follows. An initial orbit
189: point in, say, resonance $l/m$ will ``rotate'', jumping from zone ${\cal Z}
190: ^{(t)}(l/m)$ in gap ${\bf G}_{t}$ to zone ${\cal Z}^{(t+1)}(l/m)$ in gap $
191: {\bf G}_{t+1}$, until it will arrive to ${\cal Z}^{(0)}(l/m)$. If it does
192: not lie in an exiting turnstile lobe, it will rotate again, returning to $
193: {\cal Z}^{(0)}(l/m)$ after $m$ iterations. If, on the other hand, it lies in
194: an exiting turnstile lobe, more precisely in the {\em overlap} of this lobe
195: with an entering turnstile lobe of resonance $l^{\prime }/m^{\prime }$ (such
196: overlaps are the shaded regions in Fig. 1), it will escape to zone ${\cal Z}
197: ^{(m^{\prime }-1)}(l^{\prime }/m^{\prime })$ of $l^{\prime }/m^{\prime }$;
198: it will then perform at least a finite number of rotations (of $m^{\prime }$
199: iterations each) in $l^{\prime }/m^{\prime }$ before escaping to another
200: resonance. Thus, the orbit is a sequence of quasiregular segments, each
201: lying in some resonance $l_{r}/m_{r}$, $-\infty <r<\infty $, and having a
202: length of $q_{r}m_{r}$ iterations, where $q_{r}$ is the number of rotations
203: performed in $l_{r}/m_{r}$. We then say that the orbit is of quasiregularity
204: {\em type} $\tau =\dots ,(l_{r}/m_{r})_{q_{r}},(l_{r+1}/m_{r+1})_{q_{r+1}}\dots$
205: \cite{d2,d3}. As an example, Fig. 1 shows five orbit points, labeled by $
206: t=1,\dots ,5$, in two consecutive quasiregular segments $(0/1)_{3},(1/2)_{1}$.
207: \newline
208:
209: In the case that the orbit is a PO, i.e., it satisfies (\ref{am}), its type
210: can be written in a more compact form \cite{bd,d2}. Clearly, a PO can visit
211: only a finite number ($d$)\ of resonances on the torus $0\leq x,\ p<2\pi $
212: [by taking also $p_{t}$ modulo $2\pi $ in (\ref{map})]. Thus, on the
213: cylinder, it will generally visit a set of $d$ resonances $\left\{
214: l_{r}/m_{r}\right\} _{r=1}^{d}$ and all the translates $\left\{
215: l_{r}/m_{r}+bw_{\tau }\right\} _{r=1}^{d}$ of this set in the $p$-direction,
216: where $b$ takes all the integer values and $w_{\tau }$ is some integer
217: related to $w$, see below. The type $\tau $ of the PO must be then
218: essentially the repetition of a ``block'' $\Gamma $, $\tau =\dots ,\Gamma
219: (-w_{\tau }),\Gamma (0),\Gamma (w_{\tau }),\Gamma (2w_{\tau }),\dots $,
220: where $\Gamma (bw_{\tau })=(l_{1}/m_{1}+bw_{\tau })_{q_{1}},\dots ,
221: (l_{d}/m_{d}+bw_{\tau })_{q_{d}}$. If the periodic cycle is completed
222: exactly after visiting one block, one has $w=w_{\tau }$ and the period $
223: n=n_{\tau }=\sum_{r=1}^{d}q_{r}m_{r}$. Generally, however, the periodic
224: cycle is completed only after visiting more than one block, say $c$ blocks.
225: Then, $w=cw_{\tau }$ and $n=cn_{\tau }$. The type $\tau $ of the PO will be
226: thus specified by $(\Gamma ;w_{\tau },c)$, where $\Gamma $ stands, e.g., for
227: $\Gamma (0)$. The average acceleration per iteration is $2\pi w/n=2\pi
228: w_{\tau }/n_{\tau }$.\newline
229:
230: \begin{center}
231: {\bf III. ROTATING ACCELERATOR-MODE ISLANDS (RAIs)}\\[0pt]
232: \end{center}
233:
234: If a period-$n$ PO is stable, each of its $n$ points is the ``center'' of an
235: island in a chain of $n$ islands. As shown in Ref. \cite{bd}, an island must
236: lie entirely within the zone of some resonance $l/m$. If this zone is the
237: principal one, ${\cal Z}^{(0)}(l/m)$, the island will be either outside the
238: turnstiles or completely within the turnstile overlap (TO) of $l/m$ with
239: another resonance $l^{\prime }/m^{\prime }$ \cite{bd}. Thus, the island will
240: always lie entirely in the basic region [resonance zone (outside the
241: turnstiles) or TO] where its center lies, so that one can characterize the
242: island chain by the type $(\Gamma ;w_{\tau },c)$ of its central PO. For
243: example, the well-known period-1 AM islands arising for\ $K>$ $2\pi |w|$
244: \cite{bvc,krw,ikh} must all lie within first-order ($m=1$) resonances and
245: their type is $\tau =((0/1)_{1};w,1)$; this is because $n=cn_{\tau
246: }=c\sum_{r=1}^{d}q_{r}m_{r}$ implies, for $n=1$, that $c=d=q_{1}=m_{1}=1$.
247: Since the fraction of phase space occupied by the $m=1$ resonances
248: approaches $100\%$ as $K$ increases \cite{mmp}, and is already significant
249: for $K>K_{{\rm c}}$, it is natural to ask about the existence of RAIs, i.e.,
250: AM islands visiting resonances of order $m>1$.\newline
251:
252: To answer this question, we have carefully examined all the data on
253: standard-map AM islands of which we are aware. Most of this data appears in
254: Refs. \cite{krw,ikh,shi}; apparently new AM islands have been also
255: considered. The type of an island chain or of its central PO was accurately
256: determined by using the efficient method introduced in Ref. \cite{d3}.
257: Briefly, this method is based on the fact, proven in Ref. \cite{d3}, that
258: one can always find a sawtooth map $M_{{\rm s}}$ [i.e., the map (\ref{map})
259: with $\sin (x)$ replaced by a sawtooth function and with $K$ replaced by a
260: properly chosen parameter $K_{{\rm s}}$] such that for each orbit ${\cal O}$
261: of $M$ there exists an orbit ${\cal O}_{{\rm s}}$ of $M_{{\rm s}}$ visiting
262: the {\em same} resonances as those visited by ${\cal O}$. Since the
263: boundaries of the resonances of $M_{{\rm s}}$ are given by simple analytic
264: expressions \cite{saw}, this allows one to determine the type of ${\cal O}$
265: without calculating the complicated resonance boundaries of $M$.\newline
266:
267: Our results are presented in Tables I and II. In Table I, we give the type $
268: \tau $ of many AM island chains for $K_{{\rm c}}<K<2\pi $. These chains,
269: most of which appear in Table I in Ref. \cite{ikh}, were chosen in a
270: well-defined and natural way, i.e., they have at least one island lying on $
271: p=0$ (see also next section), except for chains no. 7, 8 which are given for
272: future reference; initial conditions $(x_{0},p_{0})$ within the islands are
273: specified, as well as the values of $K$, $n$, and $w$. We see that more than
274: half of these island chains (no. 1-12, 15, 16), mostly at the smaller values
275: of $K$, are RAIs visiting resonances of order $m=2$ and/or $m=3$. The
276: central PO for RAI no. 6, with $n=5$ and $w=1$, consists of the five points
277: shown in Fig. 1. As another example, we show in Fig. 2 the central PO for
278: RAI no. 1, with $n=11$ and $w=1$, together with the four resonances visited,
279: $l/m=0/1,1/3,1/2,2/3$. RAI no. 5, with $n=10$, emerges by period-doubling
280: bifurcation from RAI no. 4 with $n=5$. This is reflected in the fact that
281: the types of RAIs no. 4 and 5 have the same basic block $\Gamma $ but $c=w=2$
282: for RAI no. 5, in contrast with $c=w=1$ for RAI no. 4. Similarly, RAI no. 8,
283: with $n=8$, emerges by period-doubling bifurcation from RAI no. 7 with $n=4$.
284: \newline
285:
286: In Table II, we give the type $\tau $\ of AM\ islands with $n=1,2,3$ for $
287: 2\pi <K<20$. Most of these islands, those with $n=1,2$, were selected from
288: Table I in Ref. \cite{krw}. As in that work, we give in Table II the
289: stability interval $(K_{1},K_{2})$ of each island and corresponding initial
290: conditions $(x_{0},p_{0})$ for $K=K_{1,2}$; the values of $n$ and $w$ re
291: also shown. Islands no. 2, 6, and 9 emerge by period-doubling bifurcation
292: from islands no. 1, 5, and 8, respectively. All these islands lie in $m=1$
293: resonances. The only AM islands that we were able to identify as RAIs in
294: this $K$ interval are no. 3, 4, and 7. These RAIs are ``pure'', i.e., they
295: visit resonances of the same order $m=2$ (RAIs no. 3 and 7) or $m=3$ (RAI
296: no. 4). The latter RAI is apparently a new AM island. Figs. 3-5 show the
297: central PO for each of these pure RAIs, together with the resonances visited
298: and the turnstiles responsible for the transition from resonance $1/m$ to
299: resonance $1/m+w$; this transition causes the acceleration.\newline
300:
301: RAIs can be very small islands. As examples, we show in Fig. 6 one island of
302: each of the RAI chains to which we refer in Figs. 2-5.\newline
303:
304: \begin{center}
305: {\bf IV. CHAOTIC TRANSPORT IN THE PRESENCE OF RAIs}\\[0pt]
306: \end{center}
307:
308: We now briefly study the effect of RAIs on chaotic transport. Given an
309: ensemble ${\cal E}$ of initial conditions $(x_{0},p_{0})$ in phase space for
310: $K>K_{{\rm c}}$, the transport of ${\cal E}$ is usually measured by the time
311: evolution of $\left\langle (p_{t}-p_{0})^{2}\right\rangle _{{\cal E}}$,
312: where $\left\langle \ \right\rangle _{{\cal E}}$ denotes average over ${\cal
313: E}$. In the absence of AM islands, with ${\cal E}$ lying entirely within the
314: connected chaotic region, $\left\langle (p_{t}-p_{0})^{2}\right\rangle _{
315: {\cal E}}\approx 2Dt$ for large $t$, where $D$ is the chaotic-diffusion
316: coefficient \cite{bvc}. In Ref. \cite{ikh}, ${\cal E}$ was naturally chosen
317: as a physical ensemble of well-defined angular momentum $p=p_{0}$, $0\leq
318: x_{0}<2\pi $, and the quantity $D_{{\cal E},t}(K)=\left\langle
319: (p_{t}-p_{0})^{2}\right\rangle _{{\cal E}}/(2t)$ was calculated at fixed
320: large $t$ as a function of $K$, for $K_{{\rm c}}<K<2\pi $. Whenever an AM
321: island crosses the line $p=p_{0}$ as $K$ is varied, $D_{{\cal E},t}(K)$
322: exhibits ``ballistic'' peaks, see Fig. 7 for $p_{0}=0$ and Ref. \cite{ikh}.
323: The AM islands on $p=0$ in Table I were determined in this way. Some of the
324: most significant peaks in Fig. 7 are due to RAIs, e.g., RAIs no. 6-8, 10,
325: 15, 16. In particular, the peak due to RAI no. 16 is relatively broad in $K$.
326: \newline
327:
328: If ${\cal E}$ is an ensemble lying entirely within the connected chaotic
329: region and there exist AM islands exhibiting sufficient stickiness, one
330: observes an anomalous, superdiffusive chaotic transport, $\left\langle
331: (p_{t}-p_{0})^{2}\right\rangle _{{\cal E}}\propto t^{\mu }$ with anomalous
332: exponent $\mu $, $1<\mu <2$ \cite{ihkm,zen,bkwz,wbkz,ze,sd}. As far as we
333: are aware, this anomalous transport in the standard map was observed only
334: for AM islands whose central PO has period $n=1$ (such as islands no. 1, 5,
335: and 8 in Table II) or satellites of these islands. All these islands lie in $
336: m=1$ resonances and are not RAIs.\newline
337:
338: To show that RAIs affect chaotic transport, we consider, as a first example,
339: the case of RAI no. 7 in Table I ($K=2.975$), see Fig. 8. The ensemble $
340: {\cal E}$ consists of the points $(x_{0},p_{0})$ with $p_{0}=0$ and $x_{0}$
341: taking $10^{5}$ values uniformly distributed in $[0,2\pi )$. This ensemble
342: is chaotic with the exception of $\sim 20\%$ of it lying in ordinary
343: (non-AM) islands within the $0/1$ resonance. The only source of anomalous
344: transport can be stickiness to the boundary of the RAI above, since no other
345: AM\ islands seem to exist for $K=2.975$. To verify that this RAI boundary is
346: indeed sticky, we have iterated the ensemble $t=10^{4}$ times and plotted
347: only the points $(x_{t^{\prime }},p_{t^{\prime }}$ mod$(2\pi ))$, for all $
348: t^{\prime }\leq t$ and with $p_{t}/(2\pi )>100$; since the RAI has central
349: period $n=4$, $p_{t}/(2\pi )$ cannot be larger than $t/4=2500$. The results
350: are shown in the inset of Fig. 8, and one can see a strong stickiness to the
351: RAI\ boundary. As a consequence, we observe a clearly superdiffusive chaotic
352: transport with anomalous exponent $\mu \approx 1.28$, see Fig. 9.\newline
353:
354: As other examples, we have considered the much smaller RAIs shown in Figs.
355: 6(b), 6(c), and 6(d). To verify the stickiness to the boundary of these
356: RAIs, we first chose ${\cal E}$ as a chaotic ensemble in the close
357: neighborhood of the RAI, see details in the captions of Figs. 6 and 10. The
358: time evolution of $\left\langle p_{t}-p_{0}\right\rangle _{{\cal E}}/(2\pi )$
359: was then calculated for sufficiently large $t$; the results are shown in
360: Fig. 10. We see that in a significant time interval (e.g., $t\leq 100$ in
361: Fig. 10(b)), $\left\langle p_{t}-p_{0}\right\rangle _{{\cal E}}/(2\pi )$
362: evolves essentially as if ${\cal E}$ were concentrated inside the RAI, i.e.,
363: it exhibits the steplike structure due to rotation in $m=2$ resonances (see
364: inset of Figs. 10(a) and 10(c)) or in $m=3$ resonances (see inset of Figs.
365: 10(b)) and its initial average slope is $\sim w/n$. This is clear evidence
366: for stickiness to the RAI boundary, leading to chaotic flights. In the
367: course of time, more and more points of the ensemble leave the RAI boundary
368: and enter the chaotic region. Then, $\left\langle p_{t}-p_{0}\right\rangle _{
369: {\cal E}}/(2\pi )$ starts to saturate around some constant value which
370: should correspond to the center of a Gaussian distribution describing normal
371: chaotic diffusion. The saturation value of $\left\langle
372: p_{t}-p_{0}\right\rangle _{{\cal E}}/(2\pi )$ in Fig. 10(c) is much larger
373: than that in Figs. 10(a) and 10(b) due to the relatively large value of $
374: w/n=1$ and to the much stronger stickiness, as one can see by comparing Fig.
375: 6(d) with Figs. 6(b) and 6(c).\newline
376:
377: \begin{center}
378: {\bf V. DISCUSSION AND CONCLUSIONS}\\[0pt]
379: \end{center}
380:
381: In this paper, we have established the existence of a most interesting
382: kind of stability island in the standard map, the RAI, exhibiting two
383: diametrically opposite dynamical behaviors: The rotational motion,
384: characteristic of the integrable ($K=0$) case, and the acceleration which
385: emerges only in the global-chaos regime of $K>K_{{\rm c}}\approx 0.9716$. As
386: indicated by Table I, RAIs appear to be abundant for sufficiently small $
387: K>K_{{\rm c}}$ but they also exist in strong-chaos regimes (Table II), where
388: $m>1$ resonances are quite small and essentially all phase space is occupied
389: by $m=1$ resonances. For large $K>2\pi $, it is possible to have turnstile
390: overlap between resonances of the same order, e.g., resonances $1/m$ and $
391: 1/m+w$, and pure RAIs can then arise, see Figs. 3-5. If the map (\ref{map})
392: is restricted to the torus $T^{2}:0\leq x,\ p<2\pi $, by taking also $p_{t}$
393: modulo $2\pi $, pure RAIs look precisely like rotational-resonance islands
394: in a near-integrable regime ($K\ll 1$). Stickiness to the boundary of RAIs,
395: especially of pure RAIs, leads to chaotic flights featuring a quasiregular
396: steplike structure due to the rotational motion within resonances, see Fig.
397: 10. Such a quasiregular structure was observed recently \cite{d1} in the
398: weak-chaos regime of a perturbed pseudochaotic map. We have shown here that
399: it also occurs in strong-chaos regimes. It would be most interesting if one
400: could establish the existence of pure RAIs of arbitrarily large order $m$.
401: Such RAIs may give rise to chaotic flights with arbitrarily long
402: quasiregular steps; these flights were shown to be possible for the system
403: studied in Ref. \cite{d1}.\newline
404:
405: There has been much interest recently \cite{fgr,qam,qam1} in the forced
406: standard map
407: \begin{equation}
408: M_{\Omega }:\ \ p_{t+1}=p_{t}+K\sin (x_{t})+2\pi \Omega ,\ \ \ \
409: x_{t+1}=x_{t}+p_{t+1}\ \text{mod}(2\pi ), \label{mapf}
410: \end{equation}
411: where $\Omega $ is a constant ``force''. It was shown \cite{fgr} that $
412: M_{\Omega }$ provides an approximate description of the vicinity of quantum
413: resonance for a periodically kicked particle in the presence of gravity. The
414: quantum resonant behavior of this system manifests itself in the so-called
415: quantum AMs \cite{ao} which correspond to classical AM islands of $M_{\Omega }$.
416: The map (\ref{mapf}) arises and was studied also in other contexts \cite{mapj,mapj1}.
417: It is instructive to compare some AM islands of (\ref{mapf}) with the RAIs
418: in the standard map ($\Omega =0$). Consider, for simplicity, the case of
419: integer $\Omega \neq 0$ (the arguments and conclusions below can be easily
420: extended to the case of general rational $\Omega $). In this case, $
421: M_{\Omega }$ and the standard map obviously coincide if both maps are
422: restricted to the basic torus $T^{2}$. Thus, for sufficiently small $K$,
423: there exist rotational-resonance islands in any resonance $l/m$ of $
424: M_{\Omega }$ on $T^{2}$. On the cylinder, such an island will correspond to
425: an AM island: In one iteration of (\ref{mapf}), the island in zone ${\cal Z}
426: ^{(t)}(l/m)$, $t=0,...,m-1$, will accelerate by jumping to zone ${\cal Z}
427: ^{(t+1)}(l/m+\Omega )$ of resonance $l/m+\Omega $. Clearly, this behavior is
428: basically different from that of, say a pure RAI, which {\em remains}
429: (rotates) in resonance $l/m$ for $m$ (or a multiple of $m$) iterations
430: before jumping to resonance $l/m+w$. The rotational motion of RAIs, combined
431: with acceleration, is {\em not} featured by the AM islands above of $
432: M_{\Omega }$. It would be therefore interesting to study the fingerprints of
433: RAIs in the quantized standard map.\newpage
434:
435: \begin{center}
436: {\bf ACKNOWLEDGMENTS}\\[0pt]
437: \end{center}
438:
439: This work was partially supported by the Israel Science Foundation (Grant
440: No. 118/05).
441:
442: \begin{references}
443: \bibitem{bvc} B.V. Chirikov, Phys. Rep. {\bf 52}, 263 (1979).
444:
445: \bibitem{cm} J.R. Cary and J.D. Meiss, Phys. Rev. A {\bf 24}, 2664 (1981).
446:
447: \bibitem{krw} C.F.F. Karney, A.B. Rechester, and R.B. White, Physica
448: (Amsterdam) {\bf 4D}, 425 (1982).
449:
450: \bibitem{ikh} Y.H. Ichikawa, T. Kamimura, and T. Hatori, Physica
451: (Amsterdam) {\bf 29D}, 247 (1987).
452:
453: \bibitem{d} I. Dana, Physica (Amsterdam) {\bf 39D}, 205 (1989).
454:
455: \bibitem{km} H.T. Kook and J.D. Meiss, Phys. Rev. A {\bf 41}, 4143 (1990).
456:
457: \bibitem{dh} I. Dana and T. Horesh, Lecture Notes in Physics {\bf 511}, 51
458: (1998).
459:
460: \bibitem{ihkm} R. Ishizaki, T. Horita, T. Kobayashi, and H. Mori, Prog.
461: Theor. Phys. {\bf 85}, 1013 (1991).
462:
463: \bibitem{zen} G.M. Zaslavsky, M. Edelman, and B.A. Niyazov, Chaos {\bf 7},
464: 159 (1997).
465:
466: \bibitem{bkwz} S. Benkadda, S. Kassibrakis, R. White, and G. Zaslavsky,
467: Phys. Rev. E {\bf 55}, 4909 (1997).
468:
469: \bibitem{wbkz} R.B. White, S. Benkadda, S. Kassibrakis, and G.M. Zaslavsky,
470: Chaos {\bf 8}, 757 (1998).
471:
472: \bibitem{ze} G.M. Zaslavsky and M. Edelman, Chaos {\bf 10}, 135 (2000).
473:
474: \bibitem{sd} G.M. Zaslavsky, Phys. Rep. {\bf 371}, 461 (2002), and
475: references therein.
476:
477: \bibitem{shi} S. Saito, K. Hirose, and Y.H. Ichikawa, Chaos, Solitons, and
478: Fractals {\bf 11}, 1433 (2000), and references therein.
479:
480: \bibitem{d1} I. Dana, Phys. Rev. E {\bf 69}, 016212 (2004).
481:
482: \bibitem{kz} V. Rom-Kedar and G. Zaslavsky, Chaos {\bf 9}, 697 (1999), and
483: references therein.
484:
485: \bibitem{bd} O. Barash and I. Dana, Phys. Rev. E {\bf 71}, 036222 (2005).
486:
487: \bibitem{fgr} S. Fishman, I. Guarneri, and L. Rebuzzini, Phys. Rev. Lett.
488: {\bf 89}, 084101 (2002); J. Stat. Phys. {\bf 110}, 911 (2003).
489:
490: \bibitem{qam} A. Buchleitner, M.B. d'Arcy, S. Fishman, S.A. Gardiner, I.
491: Guarneri, Y.-Z. Ma, L. Rebuzzini, and G.S. Summy, Phys. Rev. Lett. {\bf 96},
492: 164101 (2006).
493:
494: \bibitem{qam1} I. Guarneri, L. Rebuzzini, and S. Fishman, Nonlinearity {\bf
495: 19}, 1141 (2006).
496:
497: \bibitem{ao} R.M. Godun {\it et al.}, Phys. Rev. A {\bf 62}, 013411 (2000);
498: M.B. d'Arcy {\it et al.}, Phys. Rev. E {\bf 64}, 056233 (2001); S. Schlunk
499: {\it et al.}, Phys. Rev. Lett. {\bf 90}, 054101 (2003); {\it ibid.}, 124102
500: (2003); Z.-Y. Ma, M.B. d'Arcy, and S.A. Gardiner, Phys. Rev. Lett. {\bf 93},
501: 164101 (2004).
502:
503: \bibitem{jg} J.M. Greene, J. Math. Phys. {\bf 20}, 1183 (1979).
504:
505: \bibitem{ll} A.J. Lichtenberg and M.A. Lieberman, {\em Regular and Chaotic
506: Dynamics} (Springer-Verlag, Berlin, 1992).
507:
508: \bibitem{jdm} J.D. Meiss, Rev. Mod. Phys. {\bf 64}, 795 (1992), and
509: references therein.
510:
511: \bibitem{mmp} R.S. MacKay, J.D. Meiss, and I.C. Percival, Physica
512: (Amsterdam) {\bf 27D}, 1 (1987), and references therein.
513:
514: \bibitem{saw} Q. Chen, I. Dana, J.D. Meiss, N.W. Murray, and I.C. Percival,
515: Physica (Amsterdam) {\bf 46D}, 217 (1990).
516:
517: \bibitem{bd1} O. Barash and I. Dana, Phys. Rev. E {\bf 74}, 056202 (2006).
518:
519: \bibitem{qc} Q. Chen, Phys. Lett. A {\bf 123}, 444 (1987).
520:
521: \bibitem{d2} I. Dana, Phys. Rev. Lett. {\bf 64}, 2339 (1990).
522:
523: \bibitem{d3} I. Dana, Phys. Rev. Lett. {\bf 70}, 2387 (1993), and
524: references therein.
525:
526: \bibitem{mapj} A.T. Filippov and Yu. S. Gal'pern, Phys. Lett. A {\bf 172},
527: 471 (1993).
528:
529: \bibitem{mapj1} R. Ishizaki and H. Mori, Prog. Theor. Phys. {\bf 97}, 201
530: (1997); {\it ibid.}, {\bf 100}, 1131 (1998).\newpage
531: \end{references}
532:
533: Table I. Quasiregularity type $\tau =(\Gamma ;w_{\tau },c)$ of AM island
534: chains for $K_{{\rm c}}<K<2\pi $. The period $n$, jumping index $w$, and
535: initial conditions $(x_{0},p_{0})$ are also shown.
536: \[
537: \begin{tabular}{|r|l|r|r|c|c|}
538: \hline
539: & $\ \ \ \ K$ & $\tau =(\Gamma ;\ w_{\tau },\ c)$ & $n\ $ & $w$ & $(x_{0},\
540: p_{0})/(2\pi )$ \\ \hline
541: $1$ & $2.1834$ & $((0/1)_{3},(1/3)_{1},(1/2)_{1},(2/3)_{1};\ 1,\ 1)$ & $11$
542: & $1$ & $(0.097,\ 0.0)$ \\
543: $2$ & $2.374$ & $((0/1)_{3},(1/2)_{1};\ 1,\ 1)$ & $7$ & $1$ & $(0.1085,\
544: 0.0) $ \\
545: $3$ & $2.55097$ & $((0/1)_{4},(1/2)_{1},(1/1)_{3},(3/2)_{1};\ 2,\ 1)$ & $11$
546: & $2$ & $(0.102632,\ 0.0)$ \\
547: $4$ & $2.5875$ & $((0/1)_{3},(1/2)_{1};\ 1,\ 1)$ & $5$ & $1$ & $(0.098,\
548: 0.0) $ \\
549: $5$ & $2.58867$ & $((0/1)_{3},(1/2)_{1};\ 1,\ 2)$ & $10$ & $2$ & $(0.099,\
550: 0.00151)\ $ \\
551: $6$ & $2.63894$ & $((0/1)_{3},(1/2)_{1};\ 1,\ 1)$ & $5$ & $1$ & $(0.105,\
552: 0.0)$ \\
553: $7$ & $2.975$ & $((0/1)_{2},(1/2)_{1};\ 1,\ 1)$ & $4$ & $1$ & $(0.33445,\
554: 0.4)\ $ \\
555: $8$ & $2.9845$ & $((0/1)_{2},(1/2)_{1};\ 1,\ 2)$ & $8$ & $2$ & $(0.344,\
556: 0.412)$ \\
557: $9$ & $3.34579$ & $((0/1)_{9},(1/2)_{1};\ 1,\ 1)$ & $11$ & $1$ & $(0.3851,\
558: 0.0)$ \\
559: $10$ & $3.50287$ & $((0/1)_{5},(1/2)_{1};\ 1,\ 1)$ & $7$ & $1$ & $(0.64,\
560: 0.0)$ \\
561: $11$ & $3.61283$ & $((0/1)_{9},(1/3)_{1},(1/1)_{2},(5/3)_{1};\ 2,\ 1)$ & $17$
562: & $2$ & $(0.336347,\ 0.0)$ \\
563: $12$ & $3.76991$ & $((0/1)_{5},(1/2)_{1};\ 1,\ 1)$ & $7$ & $1$ & $(0.6195,\
564: 0.0)$ \\
565: $13$ & $3.78247$ & $((0/1)_{3},(1/1)_{2};\ 2,\ 1)$ & $5$ & $2$ & $(0.0667,\
566: 0.0)$ \\
567: $14$ & $3.80761$ & $((0/1)_{3},(1/1)_{2};\ 2,\ 1)$ & $5$ & $2$ & $(0.07,\
568: 0.0)$ \\
569: $15$ & $4.141$ & $((0/1)_{1},(1/2)_{1};\ 1,\ 1)$ & $3$ & $1$ & $(0.285,\
570: 0.0) $ \\
571: $16$ & $4.66526$ & $((0/1)_{1},(1/2)_{1};\ 1,\ 1)$ & $3$ & $1$ & $(0.34805,\
572: 0.0)$ \\
573: $17$ & $5.02654$ & $((0/1)_{3},(1/1)_{1},(2/1)_{2},(3/1)_{1};\ 4,\ 1)$ & $7$
574: & $4$ & $(0.04432,\ 0.0)$ \\
575: $18$ & $5.12079$ & $((0/1)_{2},(1/1)_{1};\ 2,\ 1)$ & $3$ & $2$ & $(0.277,\
576: 0.0)$ \\
577: $19$ & $5.32499$ & $((0/1)_{3},(1/1)_{2};\ 2,\ 1)$ & $4$ & $2$ & $(0.055,\
578: 0.0)$ \\
579: $20$ & $5.41924$ & $((0/1)_{2},(1/1)_{2},(2/1)_{1};\ 3,\ 1)$ & $5$ & $3$ & $
580: (0.338162,\ 0.0)$ \\
581: $21$ & $5.45066$ & $((0/1)_{2},(1/1)_{1};\ 2,\ 2)$ & $6$ & $4$ & $(0.22265,\
582: 0.0)$ \\
583: $22$ & $5.51663$ & $((0/1)_{2},(1/1)_{1};\ 2,\ 1)$ & $3$ & $2$ & $(0.334,\
584: 0.0)$ \\
585: $23$ & $5.79623$ & $((0/1)_{6},(1/1)_{1};\ 2,\ 1)$ & $7$ & $2$ & $(0.3255,\
586: 0.0)$ \\
587: $24$ & $6.173229$ & $((0/1)_{3},(1/1)_{1},(2/1)_{1},(3/1)_{1};\ 4,\ 1)$ & $6$
588: & $4$ & $(0.0286,\ 0.0)$ \\ \hline
589: \end{tabular}
590: \]
591:
592: \newpage Table II. Quasiregularity type $\tau =(\Gamma ;w_{\tau },c)$ of AM
593: island chains for $2\pi <K<20$. Also shown are the period $n$, the jumping
594: index $w$, and the stability interval $(K_{1},K_{2})$ of the island chain
595: with corresponding initial conditions $(x_{0},p_{0})$ for $K=K_{1,2}$.
596:
597: \[
598: \begin{tabular}{|c|r|c|c|l|c|l|c|}
599: \hline
600: & $\tau =(\Gamma ;\ w_{\tau },\ c)$ & $n$ & $w$ & $\ \ \ \ K_{1}$ & $
601: (x_{0},\ p_{0})/(2\pi ),\ K=K_{1}$ & $\ \ \ \ K_{2}$ & $(x_{0},\
602: p_{0})/(2\pi ),\ K=K_{2}$ \\ \hline
603: $1$ & $((0/1)_{1};\ 1,\ 1)$ & $1$ & $1$ & $6.28319$ & $(0.2500,\ 0.0)$ & $
604: 7.44840$ & $(0.34059,\ 0.0)$ \\
605: $2$ & $((0/1)_{1};\ 1,\ 2)$ & $2$ & $2$ & $7.44840$ & $(0.34059,\ 0.0)$ & $
606: 7.71340$ & $(0.3845,\ 0.0928)$ \\
607: $3$ & $((1/2)_{1};\ 1,\ 1)$ & $2$ & $1$ & $8.67893$ & $(0.27999,\ 0.321577)$ &
608: $8.68826$ & $(0.28696,\ 0.32717)$ \\
609: $4$ & $((1/3)_{1};\ 1,\ 1)$ & $3$ & $1$ & $8.91596$ & $(0.934594,\ 0.6522)$
610: & $8.91603$ & $(0.934641,\ 0.651819)$ \\
611: $5$ & $((0/1)_{1};\ 2,\ 1)$ & $1$ & $2$ & $12.56638$ & $(0.2500,\ 0.0)$ & $
612: 13.1876$ & $(0.2990,\ 0.0)$ \\
613: $6$ & $((0/1)_{1};\ 2,\ 2)$ & $2$ & $4$ & $13.1876$ & $(0.2990,\ 0.0)$ & $
614: 13.33848$ & $\ (0.3233,\ 0.0494)$ \\
615: $7$ & $((1/2)_{1};\ 2,\ 1)$ & $2$ & $2$ & $15.2394$ & $(0.268606,\ 0.295564)$ & $
616: 15.24126$ & $(0.270938,\ 0.29762)$ \\
617: $8$ & $((0/1)_{1};\ 3,\ 1)$ & $1$ & $3$ & $18.84956$ & $(0.2500,\ 0.0)$ & $
618: 19.26929$ & $(0.2832,\ 0.0)$ \\
619: $9$ & $((0/1)_{1};\ 3,\ 2)$ & $2$ & $6$ & $19.26929$ & $(0.2832,\ 0.0)$ & $
620: 19.3728$ & $(0.2998,\ 0.0333)$ \\ \hline
621: \end{tabular}
622: \]
623:
624: \newpage
625:
626: \begin{center}
627: {\bf FIGURE CAPTIONS}\\[0pt]
628: \end{center}
629:
630: Fig. 1. Solid lines: Resonances $0/1$ (region $LERF$) and $1/2$ (with principal
631: zone $L'E'R'F'$) for $K=2.63894$. The dashed-line segments $FBH$ and $GAE$ form,
632: together with the corresponding solid-line segments, the lower and upper turnstile,
633: respectively, of $0/1$; similarly for the turnstiles of $1/2$. The shaded regions are the
634: overlaps of the upper turnstile of $0/1$ with the lower turnstile of $1/2$. Also shown are
635: five orbit points labeled by the time index $t=1,...,5$. Since point 3 lies in $0/1$ within
636: a turnstile overlap, its iterate (point 4) lies in zone 1 of $1/2$.\newline
637:
638: Fig. 2. Regions bounded by solid lines: Resonances $0/1$, $1/3$, $1/2$, and
639: $2/3$ (in ascending order) for $K=2.1834$. The dashed lines define the resonance
640: turnstiles. The points (filled circles) give the central PO of the AM island chain no. 1 in
641: Table I; this is a RAI chain. The arrow indicates a point to which we shall refer in the
642: caption of Fig. 6.\newline
643:
644: Fig. 3. Regions bounded by solid lines: Resonances $1/2$ and $3/2$ for
645: $K=8.68$. The dashed lines define the resonance turnstiles. The four points, labeled by
646: the time index $t=1,...,4$, give the central PO and its iterate for the AM island chain no.
647: 3 in Table II; this is a pure RAI chain.\newline
648:
649: Fig. 4. Regions bounded by solid lines: Resonances $1/3$ and $4/3$ for
650: $K=8.916$. The dashed lines define the resonance turnstiles. The six points, labeled by
651: the time index $t=1,...,6$, give the central PO and its iterate for the AM island chain no.
652: 4 in Table II; this is a pure RAI chain.\newline
653:
654: Fig. 5. Regions bounded by solid lines: Resonances $1/2$, $3/2$, and $5/2$
655: for $K=15.24035$. The dashed lines define the resonance turnstiles. The four points,
656: labeled by the time index $t=1,...,4$, give the central PO and its iterate for the AM island
657: chain no. 7 in Table II; this is a pure RAI chain. Point 2 in $1/2$ is mapped into point 3
658: in $5/2$ by turnstile overlap, ``jumping over" $3/2$; this leads to $w=2$.\newpage
659:
660: Fig. 6. (a) RAI surrounding the point indicated by an arrow in Fig. 2
661: ($K=2.1834$); the region $(x_1,x_2)/(2\pi )\times (p_1,p_2)/(2\pi )$ covered by the
662: graph is $(0.09694, 0.097165)\times (-5.5\cdot 10^{-5},5.5\cdot 10^{-5})$. (b) RAI
663: surrounding point 1 in Fig. 3 ($K=8.68$), graph region is $(0.9582,0.9592)\times
664: (0.674,0.68)$. (c) RAI surrounding point 2 in Fig. 4 ($K=8.916$), graph region is
665: $(0.019943,0.0199537)\times (0.085314,0.085334)$. (d) RAI surrounding point 1 in
666: Fig. 5 ($K=15.24035$), graph region is $(0.97312,0.97326)\times (0.7019,0.704)$.\newline
667:
668: Fig. 7. Plot of $D_{{\cal E},t}(K)/D_{\rm ql}$ for $t=500$ and $K\in (1,2\pi )$
669: ($\Delta K=5\cdot 10^{-3}$), where $D_{\rm ql}=K^2/4$ is the quasilinear diffusion
670: coefficient. The ensemble ${\cal E}$ consists of the points $(x_{0},p_{0})$ with
671: $p_{0}=0$ and $x_{0}$ taking $20000$ values uniformly distributed in $[0,2\pi )$. The
672: peaks indicated by arrows correspond to the RAIs no. 1, 3, 9, and 12 in Table I and
673: were produced using $t=20000$ iterations and a finer grid in both $K$ and $x_0$.\newline
674:
675: Fig. 8. Solid lines: Resonances 0/1 and 1/2 for $K=2.975$. Shown is the RAI
676: chain no. 7 in Table I, labeled by the time index $t=1,...,4$. A magnification of island
677: no. 1 appears in the inset, showing strong stickiness of chaotic orbits to the RAI
678: boundary.\newline
679:
680: Fig. 9. Solid line: Log-log plot of $\langle p_t^2\rangle _{\cal E}$ for $K=2.975$
681: (see more details in the text). Dashed line: Linear fit to the solid line, with slope
682: $\mu\approx 1.28$.\newline
683:
684: Fig. 10. Time evolution of $\langle p_t-p_0 \rangle _{\cal E}/(2\pi )$, where
685: ${\cal E}$ is an ensemble of chaotic initial conditions (ICs) $(x_0,p_0)$ extracted from a
686: grid covering the graph region specified in the caption of: (a) Fig. 6(b) (19276 chaotic
687: ICs out of 26867), $K=8.68$; (b) Fig. 6(c) (15639 chaotic ICs out of 21600), $K=8.916$;
688: (c) Fig. 6(d) (10693 chaotic ICs out of 11788), $K=15.24035$. The inset in (a), (b), and
689: (c) shows a magnification of the first iterates with average slope $\sim w/n=1/2,\ 1/3$,
690: and $1$, respectively. The steplike structure due to the stickiness to the boundaries of
691: the pure RAIs is quite evident in all cases.\newline
692:
693: \end{document}
694: