1: %\documentstyle[aps,epsfig,preprint]{revtex}

2: %\documentstyle[aps,twocolumn]{revtex}

3: %\documentstyle[aps]{revtex}

4: \documentstyle[aps,epsf,epsfig,preprint]{revtex}

5: %\input{psfig.sty}

6: \def\prl#1#2#3{{ Phys. Rev. Lett.} {\bf #1}, #2 (#3)}

7: \def\ibid#1#2#3{{\it ibid.}  {\bf #1}, #2 (#3)}

8: \def\pla#1#2#3{Phys. Lett. A {\bf #1}, #2 (#3)}

9: \def\pre#1#2#3{Phys. Rev. E {\bf #1}, #2 (#3)}

10: \def\pra#1#2#3{Phys. Rev. A {\bf #1}, #2 (#3)}

11: \def\pr#1#2#3{Phys. Rev. {\bf #1}, #2 (#3)}

12: \def\jpa#1#2#3{J. Phys. A {\bf #1}, #2 (#3)}

13: \def\jcp#1#2#3{J. Chem. Phys.  {\bf #1}, #2 (#3)}

14: \def\jsp#1#2#3{J. Stat. Phys.  {\bf #1}, #2 (#3)}

15: \def\jpc#1#2#3{J. Phys. Chem. {\bf #1}, #2 (#3)}

16: \def\physd#1#2#3{Physica D {\bf #1}, #2 (#3)}

17: \def\pram#1#2#3{Pramana J. Phys. {\bf #1}, #2 (#3)}

18: \def\etc{ etc.}

19: \def\ie{i.e.}

20: \def\X{\bf X}

21: \def\F{\bf F}

22: \def\e{\bf e}

23: \def\eg{e. g.}

24: \def\etl{$et ~al.$~}

25: \def\MLE{$\Lambda$}

26: \def\epsilonp{\epsilon^{\prime}}

27: \def\MLEN{\MLE$_N$}

28: \def\sna{SNA }

29: \def\q{\qquad}

30: \def\i{\indent}

31: \def\noi{\noident}

32: \def\beqr{\begin{eqnarray}}

33: \def\eqnr{\end{eqnarray}}

34: \def\beq{\begin{equation}}

35: \def\bc{\begin{center}}

36: \def\ec{\end{center}}

37: \def\eqn{\end{equation}}

38: \topmargin=-1.5cm

39: \textheight=23cm

40: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

41: \begin{document}

42: \title{Bifurcations and transitions in the quasiperiodically 

43: driven logistic map}

44: \author{Surendra Singh Negi, Awadhesh Prasad and Ramakrishna Ramaswamy} 

45: \address{School of Physical Sciences\\ Jawaharlal Nehru

46: University, New Delhi 110 067, INDIA}

47: \date{\today}

48: \maketitle

49: \begin{abstract}

50: We discuss several bifurcation phenomena that occur in the 

51: quasiperiodically driven logistic map. This system can

52: have strange nonchaotic attractors (SNAs) in addition to 

53: chaotic and regular attractors; on SNAs the dynamics is aperiodic,

54: but the largest Lyapunov exponent is nonpositive. There are a

55: number of different transitions that occur here, from periodic

56: attractors to SNAs, from SNAs to chaotic attractors, etc. We describe

57: some of these transitions by examining the behavior of the largest Lyapunov

58: exponent, distributions of finite time Lyapunov exponents and

59: the invariant densities in the phase space.

60: \end{abstract}

61: 

62: \newpage

63: \section{INTRODUCTION}

64: 

65: Quasiperiodically driven mappings where the dynamics can be strange

66: (namely on fractal attractors) but nonchaotic (namely having a lack

67: of sensitivity to initial conditions) are of considerable current

68: interest \cite{rmp,general}.  Strange nonchaotic attractors (SNAs)

69: were first described by Grebogi \etl in 1984 \cite{gopy}, and are now

70: known to be generic in quasiperiodically driven systems. Such

71: dynamics is paradoxical in some ways. The motion on SNAs is

72: aperiodic, but over long times, nearby trajectories will coincide.

73: The dynamics is almost always characterized by intermittency, which

74: is indicative of the fact that such attractors are highly nonuniform

75: and have a complicated interweaving of (locally) stable and unstable

76: regions.

77: 

78: External forcing allows for a additional means of probing nonlinear

79: dynamical systems.  If the forcing is periodic, then the motion of

80: the system becomes either periodic or chaotic, but for quasiperiodic

81: forcing (for example when a system is modulated with two frequencies

82: which are incommensurate with each other) SNAs become possible.

83: Strange nonchaotic dynamics usually occurs in the vicinity of strange

84: chaotic behavior and periodic or quasiperiodic (nonstrange,

85: nonchaotic) behavior. As a result, such systems can show transitions

86: between dynamical states and bifurcation phenomena which are similar

87: to those in analogous autonomous systems (for example the several

88: scenarios \cite{eckmann81} such as the period-doubling route to

89: chaos, intermittency, attractor crises, \etc) as those that are

90: distinct from the bifurcations of unforced systems

91: \cite{prl,pre,ott}.  These have been extensively studied in a

92: number of different contexts: questions of interest range from the

93: mechanisms through which SNAs are created

94: \cite{grebogi87,rboag,kpf,hh,k,frac,yl,blowout,lai}, how they

95: may be characterized \cite{pf,ditto},

96: experimental systems where these might occur

97: \cite{ditto,bulsara,newexp}, etc.

98: 

99: In this paper we examine the transitions between a number of

100: different types of attractors in the quasiperiodically driven

101: logistic map. In this prototypical driven system,

102: the attractors can be strange and nonchaotic, in addition to being 

103: strange and chaotic or nonchaotic and regular (torus attractors). 

104: Studies of the driven logistic map have played an important

105: role in the study of SNAs. Kaneko \cite{k} first observed ``torus

106: wrinkling'' in this system: this was eventually described

107: as the fractalization route to SNAs \cite{frac}. The creation of 

108: SNAs through the collision of stable period--doubled tori with 

109: their unstable parent torus was also first studied in this 

110: system \cite{hh}. One great advantage in studying the driven

111: logistic mapping (even though it is somewhat difficult

112: to obtain analytic results) is that the 

113: undriven logistic map has been extensively studied over the past

114: few decades. Many of the features of the driven system find

115: their parallels in the undriven system. At the same time, however,

116: since the

117: dynamics in the logistic map is generic of a wide class,

118: the behaviour that can be simply studied in the driven logistic

119: map is characteristic of most driven nonlinear systems. 

120: Some scenarios through which SNAs are formed in this and related systems

121: have been reviewed recently \cite{rmp}.

122: 

123: SNAs occur in several different parameter ranges, between

124: regions of periodic or torus attractors and regions of chaotic

125: attractors. There is a plethora of possible dynamical 

126: transitions, some of which have parallels in the undriven

127: system, such

128: as torus bifurcations, $$\ldots n~\hbox{T} \leftrightarrow

129: 2n~\hbox{T}\leftrightarrow

130: \ldots, $$ and others which do not, such as

131: transitions from tori to SNAs

132: $$\ldots n~\hbox{T} \leftrightarrow n~\hbox{ band SNAs}\leftrightarrow 

133: \ldots $$

134: $$\ldots 2^n~\hbox{T} \leftrightarrow 2^{n-1}~\hbox{ band

135: SNAs}\leftrightarrow \ldots .$$ SNAs merge in a manner similar to

136: the case of reverse bifurcations, SNAs widen, in a manner similar

137: to widening chaotic crises,  and can 

138: transform from one type to another or from SNAs to

139: chaotic attractors.

140: 

141: The main focus of this paper, in addition to characterizing 

142: the above bifurcations and transitions in this

143: system through the Lyapunov exponent and its fluctuations, is also to

144: examine the manner in which the invariant measure varies with the

145: parameters of the system, and the effect that this has on the

146: dynamics. 

147: In many instances, these bifurcations or transitions involve the

148: Lyapunov exponent going through zero, and can be analyzed in terms of

149: symmetries in the tangent--space dynamics. 

150: 

151: In Section II, we

152: introduce the model, and discuss the analysis in terms of global and

153: local Lyapunov exponents, and the return map for stretch exponents.

154: The the characteristic behavior of the Lyapunov exponents 

155: at the various transitions to different attractors are discussed

156: in Sec.~III. This is followed by a summary in Sec.~IV.

157: 

158: \section{Model}

159: 

160: We study the logistic map with quasiperiodic modulation of the

161: parameter $\alpha$,

162: \beqr

163: \label{logistic}

164:      x_{n+1} &=& \alpha [1 + \epsilon \cos 2\pi

165:      \theta_n] x_n (1 - x_n) \\ 

166: \theta_{n+1}  &=&  \theta_n + \omega \mbox{~mod~} 1, 

167: \eqnr

168: where $\omega$ is taken to be an irrational number (usually the

169: golden mean ratio, $ (\sqrt{5}-1)/2$. Successive iterates of

170: $\theta$ will densely and uniformly cover the unit interval in a

171: quasiperiodic manner, and the system therefore has {\it no} periodic

172: orbits.  As in our previous studies \cite{prl,pre}, we rescale the parameter

173: $$\epsilonp =\frac{\epsilon}{4/\alpha-1}$$ for convenience, and

174: study the system in the range $2 \le \alpha \le 4$ and $0 \le 

175: \epsilonp~\le~1$. 

176: 

177: 

178: The Lyapunov exponent corresponding to the $\theta$-rotation

179: is trivially zero; however the other exponent, which is of

180: most importance in determining the dynamics varies with the

181: parameters. It can be calculated by averaging the {\it stretch exponents}

182: tangential to the flow, namely 

183: \beq

184: \label{stretch}

185: y_n = \ln \vert \alpha [1 + \epsilon \cos 2\pi

186:      \theta_n] (1 - 2 x_n) \vert ,

187: \eqn

188: which is the local derivative of the mapping

189: along a trajectory.  The local or $N$--step Lyapunov exponent is

190: \beq

191: \label{nlyap}

192: \lambda_N  =  \frac{1}{N} \sum_{j=1}^N y_j

193: \eqn

194: from which, asymptotically, one gets the global Lyapunov exponent

195: \beq

196: \Lambda = \lim_{N\to \infty} \lambda_N.

197: \label{lyap}

198: \eqn

199: 

200: The stretch exponents are negative in the region bounded by the curves 

201: \beq

202: \label{xtheta}

203:  x_{\pm}(\theta) = {1\over 2} \left [1\pm {1\over{

204: \alpha(1+\epsilon \cos 2\pi\theta)}}\right ].

205: \eqn

206: Any change in the dynamics such that the invariant measure 

207: is increased in the region $[x_-(\theta), x_+(\theta)]$

208: will therefore lead to a decrease in the

209: Lyapunov exponent, and, conversely, depletion of 

210: measure in this region will

211: naturally lead to an increase in the Lyapunov exponent.  

212: 

213: Note, however, that the invariant measure, 

214: $\rho_{\alpha,\epsilon}(x,\theta)$, for 

215: this mapping is not known exactly (except at $\alpha=4, \epsilon = 0$).

216: It is therefore 

217: determined numerically by partitioning the phase space $(x,\theta)$

218: into bins and examining the itinerary of a long trajectory.

219: Shown in Fig.~1(a) is an example of a SNA; the corresponding invariant

220: measure is shown in Fig.~1(b).  The solid line is the locus of 

221: $x_{\pm}(\theta)$, (cf. Eqn.~(\ref{xtheta})), showing that the

222: SNA is largely located within the contracting regions in phase space, but

223: also has considerable support in the unstable regions.

224: 

225: Local LEs, $\lambda_N$, depend on initial conditions but (with

226: probability 1) $\Lambda$ does not \cite{multi}. In order to characterize

227: the non--uniformity of the attractor, it has proved instructive to 

228: examine the distribution of local Lyapunov exponents. The probability

229: density,

230: \beqr P(N,\lambda) d\lambda &=& {\rm probability~ }~{\rm that~} \lambda_N

231: {\rm ~lies~ between}~ \nonumber \\

232:      & &\lambda ~ {\rm and}~ \lambda + d\lambda,

233: \eqnr

234: has been seen to have characteristic limiting forms that depend

235: on the nature of the attractor \cite{pr,trichy}. Of course, as $N

236: \to \infty$, $P(N,\lambda) \to \delta(\lambda-\Lambda)$, but

237: the nature of the finite--size corrections and the approach to the

238: limit are distinctive for different dynamical states.  This analysis

239: is relevant for the study of nonuniform attractors, particularly for

240: different attractors across crisis points or at intermittency,

241: when stretched exponentials often occur \cite{pr}.

242: 

243: 

244: %newpage

245: \section{Results and Discussion}

246: 

247: In this Section we study the variation of the dynamics through

248: several transitions in the system as the parameters $\alpha$ and

249: $\epsilonp$ are varied. Phase diagrams for this system in different

250: parameter ranges have been obtained in a number of previous studies 

251: \cite{prl,pre,kpf,hh,trichy,witt}, and it is known that there are two distinct

252: regions of chaotic dynamics, corresponding, respectively to high driving

253: (large $\epsilonp$) and large nonlinearity (large $\alpha$). These are

254: shown in Fig.~2 as C$_2$ and C$_1$ \cite{prl}, and they 

255: separate a region of quasiperiodic (torus) dynamics. Strange nonchaotic motion 

256: occurs on the boundaries of these regions \cite{pre} as shown in Fig.~2.

257: 

258: In addition to the asymptotic nontrivial Lyapunov

259: exponent \MLE, we also examine the distribution of 

260: $N$--step Lyapunov exponents, and the variance

261: of this distribution. 

262: 

263: The variation in \MLE~with $\alpha$~for fixed $\epsilonp$

264: is shown in Fig.~3(a). Although there are several

265: bifurcations and transitions, these are not easily visible in

266: the behavior of the Lyapunov exponent. We therefore examine

267: an approximate or partial bifurcation 

268: diagram which can be obtained for this system by plotting the values 

269: of $x$ that obtain within a narrow window in $\theta$, namely

270: in the interval ($\theta, \theta + d\theta$). This

271: will depend upon the choice of $d\theta$ and also on

272: the particular value of $\theta$ chosen, but qualitative

273: features of the bifurcation diagram are not affected. 

274: 

275: The partial bifurcation diagram, Fig.~3(b), shows some of the

276: transitions clearly:

277: the torus--doubling, for instance, and the transition from

278: torus attractors to fractal attractors. Since singularities

279: are dense in $\theta$, a SNA or a chaotic attractor shows up

280: as a spread of points (see Fig.~3(b), for example), while

281: a torus attractor appears as a point or a finite set of points.

282: The distinction between 

283: a chaotic attractor or a SNA is not evident in the bifurcation

284: diagram, but examining Figs.~3(a) and (b), and the change in

285: the variance in \MLE, shown in Fig.~3(c) gives a complete

286: picture of the different attractors that are present in the system.

287: (Our calculations of the variance, $\sigma$, are from

288: $50$ samples, each of total length $10^6$ iterations.)

289: 

290: Quasiperiodic forcing converts the fixed points of the logistic map

291: into tori.  At the period--doubling bifurcation, the Lyapunov exponent 

292: is exactly zero, as in the unforced system, but 

293: with quasiperiodicity, the sequence of torus--doublings is interrupted

294: and leads to SNAs, either through the collision of stable and unstable

295: tori as discussed by Heagy and Hammel \cite{hh}, or through

296: fractalization \cite{frac,kk}. This latter scenario is the most

297: common route to SNA, and since there is,

298: apparently, no bifurcation involved, it is not clearly understood

299: as to how a torus gets increasingly wrinkled and transforms into 

300: a fractal attractor in the process. Other routes to SNA are 

301: known, some of which, like intermittency \cite{prl,venkat}, occur in this 

302: system and others, such as the blowout bifurcation route

303: \cite{yl}, which do not. 

304:     

305: \subsection{From SNAs to SNAs: Merging and widening crises}

306: \label{ss}

307: 

308: As has been remarked earlier, several 

309:  of the bifurcation phenomena of (unforced) chaotic dynamical

310: systems find their parallels in quasiperiodically forced systems.

311: For example, there are analogue of crisis phenomena. In the

312: present system SNAs have a certain number of

313: ``bands''.  $n$--band SNAs are formed from $n$-tori via

314: fractalization, or from 2$n$--tori through the Heagy--Hammel

315: mechanism.  Additionally, one--band SNAs can 

316: form from a 1--torus \cite{prl} or a 3--torus \cite{trichy,witt}

317: due to  saddle-node bifurcations.  

318: 

319: As parameters are further varied, SNAs themselves evolve and merge at

320: quasiperiodic analogues of band--merging crises or reverse

321: bifurcations: $n$--band SNAs transform to $n/2$--band SNAs.  Through

322: such a transition, when the dynamics remains nonchaotic and strange,

323: \MLE~ is a good order parameter.  Sosnovtseva \etl~\cite{sosno}, who

324: discovered an example of this transition in the driven H\'enon and

325: circle maps, demonstrated the merging by examining the phase

326: portrait.  Given the fairly narrow range over which SNAs exist in any

327: system, such transitions also occurs in a restricted range, and often

328: such crises can occur after the transition to chaotic attractors.

329: However, the variation of \MLE~does not follow a uniform pattern as

330: in the unforced case \cite{crutchfield,mr}. 

331: 

332: SNAs which are formed via fractalization may merge at negative \MLE.  The

333: Lyapunov exponent {\it {decreases}} with increasing nonlinearity;

334: shown in Fig.~4(a) is the variation of \MLE~at a band--merging

335: bifurcation (shown by the arrow) from a two--band SNA to a one--band

336: SNA for $\epsilon=0.05$. The critical value of the parameter is

337: $\alpha_c\approx 3.387 439$ and the corresponding `partial' 

338: bifurcation diagram of each alternate iterate of $x$

339: is shown in Fig.~4(b). At merging crises in unforced systems,  

340: the Lyapunov exponent usually does not show an increase (unlike at

341: widening crises) while with forcing, the exponent actually decreases.

342: 

343: This can be understood by examining the invariant density on the

344: attractor before and after merging. The two bands of the two--band SNA

345: straddle the region of phase space bounded by the curves $x_{\pm}(\theta)$,

346: Eq.~\ref{xtheta}.

347: After the merging transition, the density is enhanced mainly in this

348: region where the stretch exponents are all negative. As

349: a consequence, the Lyapunov exponent {\em must} decrease. 

350: (A plot of the difference in the invariant

351: density on the SNAs before and after merging is shown in Fig.~4(c).)

352: This appears to be the typical behaviour at the SNA

353: band--merging transition \cite{private}, and is in contrast

354: to the behaviour of the unforced system.

355: 

356: Widening crises also occur, where the SNA abruptly changes

357: size as a parameter is varied (near W$_1$ in Fig.~2). Shown in Fig.~5 is such an

358: example, which occurs on the line $\epsilonp=1$. Here, 

359: the post--crisis SNA, shown in Fig.~5(b), is clearly larger

360: than the pre--crisis SNA (Fig.~5(a)) which is created via 

361: fractalization. The signatures of the transition are evident 

362: in both \MLE~and 

363: the variance [Figs.~5(c-d)]. The sudden expansion of the attractor

364: seems to be due to collision with the unstable saddle (which also

365: gives rise to the intermittency transition to SNA \cite{prl}).

366: This saddle is difficult to locate, but in the three--dimensional

367: extension of this system, namely the quasiperiodically driven

368: H\'enon map, Osinga and Feudel have recently described similar

369: crises in detail \cite{osinga} and have obtained the saddle there

370: as well. In contrast to merging crises, the density is now 

371: enhanced in the unstable regions of phase space: \MLE~therefore increases. 

372: The distribution of finite---time LEs also has a stretched 

373: exponential tail which is characteristic of the

374: crisis--induced intermittency \cite{pr}

375: [see Fig.~5(e)]. Similar interior crises also take place

376: for lower forcing, when a fractalized SNA transforms to an

377: intermittent SNA. An example of this is shown in Fig.~6,  

378: near $\epsilonp=0.614...$ and $\alpha_c=3.13188...$ (W$_2$ in Fig.~2),

379: namely on the edge of region C$_2$. 

380: 

381: \subsection{From SNAs to Chaos}

382: 

383: On variation of a parameter, a SNA can be transformed into

384: a chaotic attractor if the largest Lyapunov exponent becomes

385: positive.  This transition, which is also not associated with any

386: bifurcation, is not as dramatic as from the torus to a SNA. Typically

387: there is a smooth variation in the Lyapunov exponent.

388: 

389: At the SNA$ \to $ chaos transition, all the Lyapunov exponents 

390: of the system are zero. Since \MLE~ [Eq.~(\ref{lyap})] is a 

391: global average, it is of interest to know how the individual terms in

392: Eq.~(\ref{stretch})  cancel out so as to give a value zero.

393: Consider the partial finite sums \cite{insa},

394: \beq

395: \label{sum1}

396: \lambda_N^+ =  {1 \over N_+} \sum_{i} y_i ,~~~~~~ y_i > 0,  

397: \eqn

398: and

399: \beq

400: \label{sum2}

401: \lambda_N^- =  {1 \over N_-} \sum_{i} y_i ,~~~~~~ y_i < 0,  

402: \eqn

403: namely the separate contributions to the local Lyapunov exponent.

404: These are obtained by partitioning a trajectory 

405: into $N_+$ points on expanding regions, where the 

406: stretch exponents are positive,  and $N_-$ points on 

407: contracting regions where the stretch exponents are negative with

408: $N=N_++N_-$. Clearly $\lambda_N = \frac{N_+}{N}\lambda^++\frac{N_-}{N} 

409: \lambda^-$, and with the limits

410: \beq

411: \lim_{N\to\infty} \lambda_N^{\pm} \to \lambda^{\pm}

412: \eqn

413: (cf. Eqs.~(\ref{nlyap}) and (\ref{lyap})) $\Lambda = \lambda^++\lambda^-$.

414: 

415: If there are symmetries in the system \cite{prss,newsymm},

416: then $\lambda^+$ and $\lambda^-$ can equal each other and thereby

417: yield $\Lambda = 0$. In such situations, there is a term--by--term

418: cancelation of positive and negative stretch exponents \cite{newsymm}. 

419: At torus bifurcations, for instance, or at the blowout bifurcation

420: transition to SNAs \cite{gopy,yl} this situation applies.

421: 

422: On the other hand, there can be an ``accidental''

423: equality, namely, since both $\lambda^+$ and $\lambda^-$ will

424: be functions of parameters $\alpha$ and $\epsilon$, it

425: may happen that they are both equal in magnitude for some

426: choice of parameters, and thereby lead to a zero value for the

427: Lyapunov exponent.

428: 

429: At the SNA to chaos transition, this latter situation obtains.

430: Lai \cite{lai} investigated the transition from SNAs to chaotic

431: attractors and found that this transition occurs only when the

432: contraction and expansion rates for infinitesimal vectors along a typical

433: trajectory on the attractor (namely  $\lambda^+$ and $\lambda^-$)

434: become equal; the LE passes through zero {\it linearly} as in Fig.~7(a).

435: 

436: There are, however, subtler effects that become apparent when

437: examining fluctuations in the local Lyapunov exponents. The variance

438: of the distribution shows a small increase across the

439: transition [Fig.~7(b)]. 

440: There is an increase in the fluctuations in the transition from

441: torus attractors to SNA as well (from essentially zero to some finite

442: value). At the SNA to chaos transition, the scale of fluctuations

443: doubles, and is therefore noticeable on this scale.

444: The SNA $\to$ Chaotic attractor transition is

445: not accompanied by any major change in the form or shape of the

446: attractor, and the Lyapunov exponent itself changes only from being

447: negative to positive. Yet the fluctuations on the chaotic attractor 

448: are always larger than those on the nonchaotic attractor.

449: 

450: This enhancement in fluctuations in \MLE~ can be analyzed via the

451: invariant density on the attractors. Shown in Fig.~8 is the 

452: {\it difference} in the invariant density on a SNA and a chaotic attractor

453: symmetrically placed about the SNA $\to$ Chaos transition. While it

454: is clear that the morphology of the attractor does not change

455: significantly, it can also be seen that on the SNA,

456: the invariant density is enhanced in the regions where stretch exponents 

457: are mainly negative. Since

458: these span the range from $-\infty$ to 0 within the region

459: $[x_-(\theta), x_+(\theta)]$, they contribute

460: significantly to the variance without greatly affecting the mean:

461: the Lyapunov exponent, thus, does not change significantly, but

462: the fluctuations show an increase.

463: 

464: Crises which occur {\em after} the transition to chaos appear to be

465: very similar to analogous phenomena in systems without forcing

466: \cite{grebogi87,mr};

467: $\Lambda$ has a power--law dependence on the parameter,

468: $\Lambda - \Lambda_c \propto  (\alpha-\alpha_c)^\beta$. For example

469: the  widening crisis due to saddle node bifurcation

470: along $\epsilon^{\prime}=0.02$, across the

471: transition point $\alpha_c=3.85226..$ is observed where the value of  $\beta$

472:  is $\approx 0.67$ which is larger than the unforced case \cite{mr}.

473: %newpage

474: \section{Summary}

475: 

476: In this paper we have studied different dynamical transitions

477: that occur in the quasiperiodically driven logistic map, the main

478: focus being on the quasiperiodic analogues of crisis phenomena. 

479: 

480: SNAs are formed via several different routes, and coexist with 

481: chaotic as well as other nonchaotic attractors.  As a result there

482: are interesting transformations of one type of attractor to another.

483: The torus to SNA transitions have been extensively described 

484: previously \cite{rmp,pre}.

485: Fractalization, which is the most common scenario for the formation 

486: of SNAs is a ``P2C2E'' \cite{p2c2e}: no explicit bifurcation mechanism

487: has been identified with this route. However, fractalized SNAs can

488: be transformed into intermittent SNAs upon collision with an unstable

489: torus in an analogue of the interior crisis, and also undergo merging

490: at the analogue of a merging crisis. 

491: 

492: In all these transitions, the

493: nontrivial Lyapunov exponent is a good order--parameter, and 

494: furthermore, its fluctuations the distributions of finite--time

495: exponents provide additional signatures for these transitions.

496: In order to understand the nature of the variations of these

497: quantities, however, it is necessary to study the

498: the invariant measure on SNAs. We show that they have

499: support even in regions which are locally unstable. 

500: 

501: At many of the

502: transitions that involve such attractors, there is no particular change

503: in form or morphology. However, by examining differences in the invariant

504: density at different parameter values, it becomes clear that at

505: these transitions, the manner in which the dynamics explores

506: the attractor can change drastically, and small

507: changes in the density can lead to significant effects in the 

508: fluctuation properties of such attractors.

509: 

510: Typical attractors in dynamical systems are nonuniform, both with 

511: respect to the natural invariant measure as well as in the rate 

512: at which nearby trajectories---locally---diverge or converge. 

513: If the measure on converging regions exceeds that on the diverging 

514: regions, then the attractor is asymptotically nonchaotic and will 

515: be characterized by a negative Lyapunov exponent. External 

516: modulation can be one method of

517: altering the invariant measure, and this method of 

518: creating nonchaotic dynamics \cite{insa,wong} therefore provides a new

519: means of synchronization and control. In this context,

520: therefore, studies of the bifurcations and transformations of 

521: SNAs give further insights into the interplay between global stability

522: and local instability.

523: 

524: %\vskip1cm

525: \centerline{\bf ACKNOWLEDGMENT} This research was supported

526: from the Department of Science and Technology, India.

527: 

528: 

529: 

530: \begin{thebibliography}{99}

531: \bibitem{rmp}  A. Prasad and R. Ramaswamy, to be published.

532: \bibitem{general} J. P. Eckmann and D. Ruelle, Rev. Mod. Phys. {\bf 57}, 

533: 617 (1985).

534: \bibitem{gopy} C. Grebogi, E. Ott, S. Pelikan, and J. A. Yorke,

535: \physd{13}{261}{1984}.

536: \bibitem{eckmann81} J. P. Eckmann, Rev. Mod. Phys. {\bf53}, 643 (1981).

537: \bibitem{prl} A. Prasad, V. Mehra, and R. Ramaswamy, \prl{79}{4127}{1997}.

538: \bibitem{pre} A. Prasad, V. Mehra, and R. Ramaswamy, \pre{57}{1576}{1998}.

539: \bibitem{ott} E. Ott, {\it Chaos in dynamical systems}, (Cambridge

540: University Press, Cambridge, England, 1994).

541: \bibitem{grebogi87} C. Grebogi, E. Ott, F. J. Romeiras, and J. A. Yorke, \pra{36}{5365}{1987}.

542: \bibitem{rboag} F. J. Romeiras, A. Bondeson, E. Ott, T. M. Antonsen,

543: and C. Grebogi, \physd{26}{277}{1987}.

544: \bibitem{kpf} S. P. Kuznetsov, A. S. Pikovsky, and U. Feudel, \pre{51}{R1629}{1995}. 

545: \bibitem{hh} J. F. Heagy and S. M. Hammel, \physd{70}{140}{1994}.

546: \bibitem{k} K. Kaneko, Prog. Theor. Phys., {\bf 71}, 1112 (1984).

547: \bibitem{frac} T. Nishikawa and K. Kaneko, \pre{54}{6114}{1996}.

548: \bibitem{yl} T. Yal\c{c}inkaya  and Y. C. Lai, \prl{77}{5039}{1996}.

549: \bibitem{blowout} E. Ott and J. C. Sommerer, \pla{188}{39}{1994}; P.

550: Ashwin, J. Buescu, and I. Stewart, \ibid{193}{126}{1994}; Y. C. Lai

551: and C. Grebogi, \pre{52}{R3313}{1995}. 

552: \bibitem{lai} Y. C. Lai, \pre{53}{57}{1996}; Y. C. Lai, U. Feudel, and C.

553: Grebogi, \ibid{54}{6070}{1996}.

554: \bibitem{pf} A. Pikovsky and U. Feudel, Chaos, {\bf 5}, 253 (1995).

555: \bibitem{ditto} W. L. Ditto, M. L. Spano, H. T. Savage, S. N. Rauseo,

556: J. Heagy, and E. Ott, \prl{65}{533}{1990}.

557: \bibitem{bulsara} T. Zhou, F. Moss, and A. Bulsara, \pra{45}{5394}{1992}.

558: \bibitem{newexp} W. X. Ding, H. Deutsch, A. Dinklage, and C. Wilke,

559: \pre{55}{3769}{1997}; Z. Liu and Z. Zhu, Int. J. Bifurcation and Chaos, {\bf 6}

560: 1383 (1996); Z. Zhu and Z. Liu, \ibid{7}{227}{1997}; A. J. Mandell and K. A.

561:  Selz, \jsp{70}{255}{1993}.

562: \bibitem{multi} Except for systems which have multiple coexisting

563: attractors, such as the double kicked--rotor, Duffing oscillator, \etc

564: \bibitem{pr} A. Prasad and R.  Ramaswamy, \pre{60}{2761}{1999}

565: and references therein.

566: \bibitem{trichy} A. Prasad and  R. Ramaswamy,  {\it Finite-time Lyapunov

567: Exponents of Strange Nonchaotic Attractors}, 

568: in {\it  Nonlinear Dynamics: Integrability and Chaos}

569: Eds. M Daniel, K Tamizhmani and R Sahadevan (Narosa, New Delhi,

570: 2000), pp. 227--34.

571: \bibitem{witt}  A. Witt U. Feudel, and A. S. Pikovsky, \physd{109}{180}{1997}.

572: \bibitem{kk}K. Kaneko, Prog. Theor. Phys. {\bf 72}, 202 (1984).

573: \bibitem{venkat} A. Venkatesan, M. Lakshmanan, A.  Prasad, and R.

574: Ramaswamy, Phys. Rev. E, in press 2000.

575: \bibitem{sosno} O. Sosnovtseva, U. Feudel, J. Kurths, and A. Pikovsky,

576: \pla{218}{255}{1996}. 

577: \bibitem{crutchfield} J. Crutchfield, D. Farmer, and B. Huberman, Phys. Rep.,

578: {\bf 92}, 45 (1982).

579: \bibitem{mr} V. Mehra and R. Ramaswamy, \pre{53}{3420}{1996}. 

580: \bibitem{private} We have studied the forced ring map and H\'enon maps and

581: find that at SNA mergings in these systems, the largest Lyapunov exponent

582: decreases in a manner very similar to that in the logistic system.

583: \bibitem{osinga} H. Osinga and U. Feudel, ``Boundary

584: crisis in quasiperiodically forced systems'', Physica D, in press 2000.

585: \bibitem{insa} A. Prasad and R. Ramaswamy, `` Can strange nonchaotic 

586: attractors be created through stochastic driving?'', Proc. Ind. 

587: Nat. Sci. Acad., in press 2000.

588: \bibitem{prss} A. Prasad, R. Ramaswamy, I. I.  Satija and N. Shah, 

589: \prl{83}{4530}{1999}.

590: \bibitem{newsymm} A. Prasad, S. S. Negi, S. Datta and R. Ramaswamy, 

591: in preparation. Note that the symmetries can be hidden as in Ref.~\cite{prss}

592: or complicated, as at tangent bifurcations. 

593: \bibitem{p2c2e} A ``Process too Complicated to Explain'' from 

594: S.~Rushdie, {\it Haroun and the Sea of Stories}, Penguin, 1990, pp.~57.

595: \bibitem{wong} J. W. Shuai and K. W. Wong, \pre{59}{5338}{1999}.

596: 

597: \end{thebibliography}

598: 

599: \newpage

600: \section*{Figure Captions}

601: 

602: \vskip0.5cm

603: 

604: \noindent {\bf Figure 1:} (a) A SNA in the driven logistic map

605: at $\epsilon^{\prime}=0.75$. and $\alpha= 3.26774$, and

606: (b) the corresponding invariant density (scaled $\times 10^{3}$) in

607: the phase space. The solid lines (in (a)) and the dashed

608: lines (in (b)) are the loci of

609: $x_{\pm}(\theta)$, Eqn.~(\ref{xtheta}). \\

610: 

611: \ \\

612: {\bf Figure 2:}Schematic phase diagram for a small region in parameter 

613: space, for the forced logistic map see details in Ref~.\cite{pre}

614: . T and C correspond to regions of torus and chaotic attractors.  

615: SNAs are mainly found in the shaded region along the boundary of 

616: T and C (marked S). W denotes the region where widening crises 

617: occur (see the txt). \\

618: 

619: \ \\

620: {\bf Figure 3:} (a) Variation of \MLE~ as a function of $\alpha$ for fixed

621: $\epsilon^{\prime} = 0.595$. Note the highly oscillatory structure

622: indicative of several transitions in the system. These are clearly

623: shown in (b) which is a `partial' bifurcation diagram  with

624: $\theta=0.5$ and $d\theta=0.0001$. The regions of torus attractors

625: as well as strange behavior can be easily seen.

626: (c) The variance in \MLE.  The symbol T and C correspond to torus and

627: chaotic attractors while S stands for SNA.  \\

628: 

629:  \ \\

630: {\bf Figure 4:}Variation of \MLE~at a band--merging bifurcation (shown by the arrow)

631: from a two--band (fractalized) SNA to a one--band 

632: (Heagy--Hammel) SNA along $\epsilon=0.05$ at $\alpha_c

633: \approx 3.387 439$.  (b) The corresponding `partial' bifurcation diagram

634: of the second iteration for $x$ in the interval

635: ($\theta,\theta+d\theta$), $\theta=0.3$ and $d\theta=0.01$, and 

636: (c) the difference in densities (scaled $\times 10^3$) after merging 

637: ($\alpha=3.389$) and SNA before merging ($\alpha=3.386$) 

638: along $\epsilon=0.05$. The subscripts 2bs and 1bs refer, respectively,

639: to 2--band and 1--band SNAs. \\

640:  \ \\

641: {\bf Figure 5:} (a) The SNA before widening along $\epsilon^{\prime}=1$ at

642: $\alpha=3.4449$ and (b) the SNA after the widening crisis at $\alpha=3.44498$;

643: (c) the variation of Lyapunov exponent, $\Lambda,$ across the transition 

644: point $\alpha_c=3.444955..$; (d) the variance

645: in \MLE, and (e) the distribution of finite-time Lyapunov exponents at

646: $\alpha=3.44498$, showing the stretched exponential tail which is

647: characteristic of intermittency.  \\

648: \ \\

649: {\bf Figure 6:}(a) The fractalized SNA (before widening) along 

650: $\epsilon^{\prime}= 0.614$ at $\alpha=3.13192$ and (b) the intermittent 

651: SNA after the widening crisis at $\alpha=3.13184$.  \\

652: \ \\

653: {\bf Figure 7:}The transition from SNA to a chaotic attractor along

654: $\epsilon^{\prime}=0.3$. (a) The Lyapunov exponent across the

655: transition, and (b) its fluctuations. \\

656: \ \\

657: {\bf Figure 8:}The difference in densities (scaled $\times 10^3$)

658: between a chaotic attractor ($\alpha=3.515$) and SNA

659: ($\alpha=3.508$) along $\epsilon^{\prime}=0.3$.

660: The plot of $x_{\pm}(\theta)$ (for $\alpha=3.508$)

661: of Eqn.~\ref{xtheta} is superimposed as a

662: dashed line.\\

663: 

664: 

665: \end{document}

666: