nlin0308030/noisy.tex
1: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%  
2: %  
3: %     Lifetimes of noisy repellors
4: %  
5: %     Holger Faisst and B. Eckhardt
6: %  
7: %     bruno.eckhardt@physik.uni-marburg.de  
8: %   
9: %     first version: 19.02.02
10: %     revisions:     15.03.02, 18.04.02
11: %			revisions March 2, 03
12: %                 
13: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%  
14: %BLA
15: %\documentstyle[aps,preprint,psfig]{revtex}   
16: %\documentstyle[aps,multicol,epsf,epsfig
17: %]{revtex}   
18: %\documentstyle[11pt,multicol]{article} 
19: \documentclass[showpacs,amsfonts,amsmath,twocolumn,floatfix,aps]{revtex4}
20: \usepackage{amsfonts,amsmath}
21: \usepackage{epsfig}
22: 
23: \newcommand{\CS}{{\cal S}}  
24: \newcommand{\beq}{\begin{equation}}  
25: \newcommand{\eeq}{\end{equation}}  
26: \newcommand{\bea}{\begin{eqnarray}}  
27: \newcommand{\eea}{\end{eqnarray}}  
28:   
29: \begin{document}     
30: %\draft     
31: %\widetext    
32: \title{Lifetimes of noisy repellors}
33: \author{
34: Holger Faisst and Bruno Eckhardt
35: }     
36: \address{Fachbereich Physik, Philipps Universit\"at    
37:          Marburg, D-35032 Marburg, Germany}    
38: %\date{$ $Id: noisy.tex,v 1.8 2003/06/17 14:51:28 faisst Exp $ $}     
39:   
40: \begin{abstract}
41: We study the effects of additive noise on the lifetimes of 
42: chaotic repellors. Using first order perturbation theory we argue
43: that noise will increase the lifetime if the escape holes lie
44: in regions where the unperturbed density is higher than in the
45: immediate vicinity and that it decreases if the density is lower.
46: Numerical experiments support the qualitative conclusions 
47: also beyond perturbation theory.
48: \end{abstract}     
49: \pacs{05.45.Ac, % Low-dimensional chaos
50:       05.40.Ca, % Noise
51:       05.45.Df % Fractals
52: % vom Editor gestrichen: zu allgemein: 05.45.-a %Nonlinear dynamics and nonlinear dynamical systems
53: }    
54: \maketitle
55: 
56: \section{Introduction}
57: Noise can affect the behaviour of dynamical systems in many ways.
58: It can induce transitions between otherwise disconnected regions
59: (Kramers' theory,~\cite{Haenggi90}), it changes the scaling near
60: bifurcations~\cite{Martin}, it gives rise to stochastic 
61: resonance~\cite{Haenggi98} and 
62: %in the form of multiplicative noise
63: it can even change repellors into attractors~\cite{arnold83}.
64: Some time ago Franaszek~\cite{Franaszek91} studied the effects
65: of additive noise on repellors~\cite{Tel} and found that in some
66: cases it stabilized the dynamics, i.e.~increased the life times.
67: He studied this behaviour near crises and bifurcations in the 
68: dynamics, but the reasons for the effects on the dynamics
69: remained unclear. We here want to approach the problem from
70: the side of the attractor which
71: then is perturbed to become a chaotic repellor. We will 
72: investigate the relation between the noise effects and 
73: a non-uniform density in the attractor. 
74: 
75: The hypothesis we want to test runs as follows:
76: opening up the attractor into a repellor is achieved
77: by punching holes into the support of the attractor. Additive noise 
78: can push trajectories that would barely miss the holes into escape,
79: but can also save trajectories that would escape in the unperturbed
80: situation. Whether the life time increases or decreases then
81: depends on which process is more likely: for a uniform density
82: noise will kick out as many trajectories as it saves, so one
83: cannot expect any effect. If the unperturbed density in the hole
84: region is higher than in the immediate vicinity, 
85: more points will be saved than kicked out, and the 
86: lifetime should increase. If the unperturbed density in the 
87: hole region is lower, more trajectories will escape and 
88: the lifetime should be reduced. 
89: 
90: In section~\ref{sec_pt} we present the perturbative arguments,
91: followed by numerical experiments in section~\ref{sec_plm} and some 
92: final remarks in section~\ref{sec_fr}.
93: 
94: \section{Perturbation theory\label{sec_pt}}
95: We start with a $d$-dimensional map 
96: \beq
97: {\bf x}_{n+1} = {\bf f}({\bf x}_n)
98: \label{kick}
99: \eeq
100: that has a chaotic attractor. The associated Frobenius-Perron equation
101: for the evolution of densities $\rho({\bf x})$ is
102: \bea
103: \rho_{n+1}({\bf x}) &=& \int d{\bf y} \, 
104: \delta({\bf x} - {\bf f}({\bf y}))\rho_n({\bf y})\\
105: &=& \sum_{i} \frac{\rho_n({\bf y_i})}{|D{\bf f}({\bf y}_i)|}\,,
106: \eea
107: where $|D{\bf f}|$ is the Jacobi determinant
108: and the summation extends over all points ${\bf y}_i$
109: that map into ${\bf x}$.
110: Noise can be added as a Gaussian smearing of the propagator,
111: as in a kicked system: the time evolution splits into two parts,
112: the `kick' (\ref{kick}), i.e.~the application of the
113: deterministic map, and a free diffusive spreading. The free
114: diffusion on the $d$-dimensional phase space is described by 
115: a diffusion kernel $K_D$ that solves the appropriate free
116: diffusion equation with $\delta$-function initial conditions.
117: For instance, for free diffusion in Euclidean space the 
118: Fokker-Planck equation for a density $\rho$ is
119: \beq
120: \dot \rho = D \Delta \rho
121: \eeq
122: and the diffusion kernel becomes
123: \beq
124: K_D({\bf y}, {\bf x}, t) =
125: \frac{1}{(2\pi D t)^{d/2}} e^{-({\bf y}-{\bf x})^2/2Dt} \,.
126: \eeq
127: The combined evolution of kick and diffusion for some time 
128: $T$ is then described by 
129: \bea
130: \rho_{n+1}({\bf x}) &=& 
131: \int d{\bf z}\, K_D({\bf x},{\bf z}, T)
132: \int d{\bf y} \, \delta({\bf z} - {\bf f}({\bf y}))\rho_n({\bf y})\\
133: &=&
134: \int d{\bf y}\, K_D({\bf x},{\bf f}({\bf y}),T )\rho_n({\bf y})\\
135: &=& \int d{\bf y}\, K({\bf x}, {\bf y}) \rho_n({\bf y}) \,.
136: \label{evolution}
137: \eea
138: 
139: The evolution kernel $K$ in (\ref{evolution}) can be expanded
140: in terms of left $\langle \lambda|$ and right eigenfunctions 
141: $|\lambda\rangle$ with eigenvalues $\lambda$,
142: \beq
143: K = \sum_\lambda \lambda\,  |\lambda\rangle   \langle\lambda| \,.
144: \eeq
145: The existence of an invariant density on the attractor (the
146: Sinai-Ruelle-Bowen measure) implies the presence of
147: an eigenvalue $\lambda=1$ with right eigenvector $|1\rangle$,
148: the invariant density, and a corresponding left eigenvector
149: $\langle 1|=1$ because of conservation of probability. 
150: Since the kernel $K$ is not self adjoint left- and right-eigenvectors
151: are different. Simple examples show that the 
152: %right I'm sure you mean 'left'!
153: left 
154: eigenvectors
155: develop fractal features \cite{Gaspard}. With noise the 
156: finest scale structures are washed out, but higher levels
157: of the fractal hierarchy survive. 
158: 
159: In order to turn the attractor into a chaotic repellor we
160: punch holes into it. In applications the holes
161: appear through crises and other perturbations and appear on many scales.
162: For the purpose of the present analysis trajectories that enter the holes 
163: may be terminated, since their further evolution does not influence the 
164: escape rate.
165: Let ${\cal O}$ be the domain over which
166: trajectories are taken out and $P$ the elimination projection:
167: \beq
168: P({\bf x}, {\bf z}) = \delta({\bf x}-{\bf z}) \cdot
169: \left\{ \begin{matrix}
170: 1 & {\bf x}\notin {\cal O}\cr
171:                             p & {\bf x}\in {\cal O}
172: \end{matrix}\right. \,.
173: \label{eli_pro}
174: \eeq
175: This projection depends on a parameter $p$ that will be useful in tests 
176: of the perturbation calculations:
177: Trajectories entering ${\cal O}$ continue on with
178: probability $p$ and are taken out with probability $1-p$.
179: For the holes described before we have to take $p=0$. 
180: % taken out with 
181: %probability $0\le 1-p\le1$.
182: The full evolution operator can then be written
183: \bea
184: & & K_P({\bf x}, {\bf y}) = 
185: \int d{\bf z} P({\bf x}, {\bf z}) K({\bf z}, {\bf y})\\
186: &=& K({\bf x}, {\bf y}) -
187: \int_{\cal O} 
188: d{\bf z} \left(
189: \delta({\bf x}, {\bf z}) -P({\bf x}, {\bf z})
190: \right)  K({\bf z}, {\bf y})\\  
191: &=& K({\bf x}, {\bf y}) +\alpha K_1({\bf x}, {\bf y}) \,.
192: \eea
193: The last equation now has the form of an unperturbed propagator $K$ 
194: plus a small perturbation $\alpha K_1$, 
195: where smallness is controlled by the
196: localization in the region ${\cal O}$ and
197: the rate $1-p$ with which points are taken out. $\alpha$ is a formal
198: parameter that helps to organize the familiar perturbation expansion
199: for eigenvalues and eigenvectors. The leading order result contains
200: as usual the diagonal matrix element of the perturbation,
201: \beq
202: \lambda \approx \lambda_0 + \alpha \langle \lambda | K_1 | \lambda \rangle\,.
203: \eeq
204: The deviations of the leading eigenvalue from $\lambda_0=1$ 
205: define a decay rate, $\lambda = \exp(-\gamma)$, where 
206: \beq
207: \gamma \approx -\alpha \langle \lambda | K_1 | \lambda \rangle \,.
208: \label{gamma}
209: \eeq
210: The systems we study here have piecewise smooth invariant densities 
211: and the first order correction to the decay rate has a regular 
212: dependence on the set ${\cal O}$ and the extraction rate $p$.
213: In particular, for the logistic map with holes with $p=0$ 
214: that is studied in 
215: \cite{Paar97} expression (\ref{gamma}) 
216: gives the smooth background. 
217: The simulations by Paar and Pavin~\cite{Paar97} also indicate
218: strong variations in escape rate near short periodic orbits 
219: which are connected to higher orders in perturbation theory
220: and the complicated spatial structures that are characteristic
221: for next to leading eigenvectors~\cite{Gaspard}.
222: 
223: \section{Piecewise linear maps 
224: with non-constant invariant densities\label{sec_plm}}
225: Let us define a family of 1-d maps $x_{n+1} = f_c(x_n)$ with
226: %$f_c: x_n\mapsto x_{n+1}$, $x\in[0,1]$, with real 
227: a parameter $c\in(0,0.5)$,
228: \beq
229: f_c(x) = 
230: \left\{
231: \begin{array}{cl}
232: x/c & 0<x\le c
233: \\
234: 1-2(x-c)c/(1-2c)&  c <x \le 1/2
235: \\
236: c - (2x-1)c/(1-2c)& 1/2 <x \le 1-c  
237: \\
238: 1+(x-1)/c &  1-c <x \le 1  \, .      
239: \end{array}
240: \right. 
241: \label{map}
242: \eeq
243: %
244: The invariant density is a solution to the Frobenius-Perron equation 
245: %\bea
246: %\rho_{n+1}(x)&=&\int_0^1 dy \,\delta[x-f(y)]\rho_{n}(y) \\
247: %&=& \sum_{x=f(x_i)} \frac{\rho_n(x_i)}{|f'(x_i)|},
248: %\label{fpe}
249: %\eea
250: \bea
251: \rho_c(x)&=&\int_0^1 dy \,\delta[x-f_c(y)]\rho_c(y) \\
252: &=& \sum_{x=f_c(x_i)} \frac{\rho_c(x_i)}{|f_c'(x_i)|}\,,
253: \label{fpe}
254: \eea
255: where the sum is taken over all pre-images $x_i$ of $x$.
256: For (\ref{map}) the normalized invariant density is
257: %The normalized solution to Eqn.~\ref{fpe} for map~(\ref{map}) is  
258: \beq
259: \rho_c (x) = 
260: \left\{
261: \begin{array}{cl}
262: \frac{1}{4c(1-c)} & 0 < x < c \; \mbox{and}\; 1-c <x< 1
263: \\
264: \frac{1}{2(1-c)}   &  c <x< 1-c \, .        
265: \end{array}
266: \right.
267: \label{rho_1}
268: \eeq
269: The main results do not depend on the specific value of $c$ so that we can fix $c=0.2$ 
270: and drop the subscript on $f$ and $\rho$. 
271: %%% The invariant density is
272: %%% \beq
273: %%% \rho = 
274: %%% \left\{
275: %%% \begin{array}{cl}
276: %%% 25/16, & 0 < x < 0.2 \; \mbox{and}\; 0.8 <x< 1
277: %%% \\
278: %%% 5/8,   &  0.2 <x< 0.8  \; .      
279: %%% \end{array}
280: %%% \right.
281: %%% \label{rho_2}
282: %%% \eeq
283: \begin{figure}%[hbt]
284: \begin{center}  
285: %\epsfig{file=/usr/ax1321/faisst/rausch/tent/ernsthaft/data/map_analytic_del_0.2.eps,width=.47\textwidth}%,height=60mm}  
286: %\epsfig{file=map_analytic_del_0.2.eps,width=.47\textwidth}%,height=60mm}  
287: \epsfig{file=fig1.eps,width=.47\textwidth,clip=}%,height=60mm}  
288: %\epsfig{file=/usr/ax1321/faisst/rausch/tent/ernsthaft/data/map_del_0.2.eps,width=.47\textwidth}%,height=60mm}  
289: \end{center}  
290: \caption[]{A graph of the map (\ref{map}) for $c=0.2$ (dashed) and 
291: its invariant density $\rho$ (continuous). 
292: }
293: \label{map_del_0.2}
294: \end{figure}
295: 
296: For the eigenvalue and eigenvector analysis we use a matrix representation of the 
297: density evolution operator $K$. With the help of $N$ characteristic functions $\phi_\nu$, 
298: \beq
299: \phi_\nu(x) = 
300: \left\{
301: \begin{array}{cl}
302: 1, & (\nu-1)/N < x < \nu/N, \, \nu=1,\ldots,N
303: \\
304: 0,   &  \hbox{elsewhere\, ,}        
305: \end{array}
306: \right.
307: \label{phi}
308: \eeq
309: densities can be expanded as $\rho= a_\nu \phi_\nu $ (summation implied) 
310: and the evolution operator $K$ becomes an $N \times N$ matrix. Densities are
311: mapped according to 
312: \beq
313: a_\nu^{(n+1)} = K_{\nu\mu}a_\mu^{(n)}\,,
314: \eeq
315: and the matrix elements $K_{\nu\mu}$ can be calculated from the images of 
316: the step functions: in the support of the characteristic function $\mu$
317: a uniformly distributed ensemble of $5\times 10^5$ points is iterated
318: and the probability to end up in the interval $\nu$ is then the matrix
319: element $K_{\nu\mu}$. The typical size of matrices used in the 
320: calculations is $N=3000$.
321: 
322: The lifetimes for the maps with holes can be obtained from the 
323: eigenvalues of the evolution operator after projection onto 
324: the remaining intervals or directly from integrations of an ensemble 
325: of initial conditions. We followed
326: $10^6$ randomly selected initial points up to a maximal cut-off lifetime of 
327: $10^4$ iterations. The lifetime distributions decayed exponentially and
328: the lifetimes estimated from this decay were within less than $1\%$ of the
329: eigenvalues. 
330: %This provides another test of the reliability of the numerical
331: %representation of the evolution operator.
332: 
333: %--------------------------------------------------
334: \subsection{No holes and no noise}
335: 
336: % shows the right and left eigenfunctions of $K$ corresponding 
337: % to the two eigenvalues of largest absolute value.
338: 
339: \begin{figure}%[hbt]
340: \begin{center}  
341: %\epsfig{file=/usr/ax1321/faisst/rausch/tent/ernsthaft/spektrum_improved_3/data/EV_nonoise_nohole.eps,width=.47\textwidth}%,height=60mm}  
342: %\epsfig{file=EV_nonoise_nohole.eps,width=.47\textwidth}%,height=60mm}  
343: \epsfig{file=fig2.eps,width=.47\textwidth,clip=}%,height=60mm}  
344: \end{center}  
345: \caption[]{Eigenfunctions of map without noise and without holes. 
346: The upper row corresponds to the leading eigenvalue $\lambda_0=1$ and the 
347: lower one to the next to leading eigenvalue $\lambda_1=-0.6$.
348: %Upper (lower) graphs correspond to $\lambda_1=1$ ($\lambda_2=-0.6$).
349: }
350: \label{first_two_eigenfunctions}
351: \end{figure}
352: 
353: Without noise and without holes the map (\ref{map}) maps the interval
354: $[0,1]$ into itself. We expect one eigenvalue $1$ with right eigenvector the
355: invariant density (as in Fig.~\ref{map_del_0.2}), and a left eigenvector that
356: is constant because of conservation of probability.
357: The first two pairs of left and right eigenvectors are
358: shown in Fig.~\ref{first_two_eigenfunctions}. The next to leading eigenvalue
359: is $\lambda_1=-0.6$. Its left eigenvector shows the fractal structures
360: one expects for such maps \cite{Gaspard}. 
361: 
362: %The right eigenfunction to eigenvalue $1$ agrees with the invariant density 
363: %from Fig.~\ref{map_del_0.2}.
364: %The left eigenfunction to eigenvalue $1$ is the uniform density as it is 
365: %for any closed system, even in the presence of noise. 
366: 
367: %==================================================
368: \subsection{With holes but without noise}
369: 
370: Pairs of holes of size $\epsilon$ are added symmetrically at the edges of 
371: the invariant density. 
372: `Outer' holes at $(c -\epsilon, c)$ and $(1-c, 1-c+\epsilon)$ lie within the 
373: high density region,  
374: `inner' holes at $(c, c+\epsilon)$ and $(1-c-\epsilon, 1-c)$ in the 
375: low density region.  
376: 
377: The perturbation theory from section~\ref{sec_pt} predicts a linear variation of escape 
378: rate with hole size, so that the ratio of escape rate to hole size should be
379: constant. This is verified in Fig.~\ref{escaperate_nonoise}.
380: \begin{figure}%[hbt]
381: \begin{center}  
382: %\epsfig{file=/usr/ax1321/faisst/rausch/tent/ernsthaft/data/escaperate_nonoise2.eps,width=.47\textwidth}%,height=60mm}  
383: %\epsfig{file=escaperate_nonoise2.eps,width=.47\textwidth}%,height=60mm}  
384: \epsfig{file=fig3.eps,width=.47\textwidth,clip=}%,height=60mm}  
385: %\epsfig{file=spectrum_small5b.eps,width=.47\textwidth}%,height=60mm}  
386: \end{center}  
387: \caption[]{Escape rate $\gamma$ 
388: divided by the hole size $\epsilon$
389: for the map (\ref{map}) with holes but without noise.
390: The dashed lines at $3.125$ and $1.125$ mark the values
391: expected within first order perturbation theory.
392: }
393: \label{escaperate_nonoise}
394: \end{figure}
395: The numerical values for the escape rate $\gamma$ agree well with the expected values,
396: sum of hole sizes times undisturbed density at holes, 
397: $\gamma_{i,o} = 2\epsilon\rho_{i,o}$
398: .   
399: %--------------------------------------------------
400: 
401: With holes we no longer have conservation of probability, and the left eigenvector
402: to the leading eigenvalue will not be constant. It develops a fractal
403: structure, that already for hole size $\epsilon=10^{-3}$ 
404: is difficult to represent numerically. The left eigenstate for 
405: the leading eigenvalue $\lambda_0=0.996855$ 
406: for the map with outer holes 
407: is shown in 
408: Fig.~\ref{first_right_eigenfctn_eps_1.e-3}.
409: The largest eigenvalue corresponds to an escape rate 
410: $\gamma_o = 3.150\times10^{-3}$ in good agreement with the values extracted 
411: from Fig.~\ref{escaperate_nonoise}.
412: 
413: \begin{figure}%[hbt]
414: \begin{center}  
415: %\epsfig{file=/usr/ax1321/faisst/rausch/tent/ernsthaft/spektrum_improved_3/data/first_left_eigenfctn_0.001.eps,width=.47\textwidth}%,height=60mm}  
416: %\epsfig{file=first_left_eigenfctn_0.001.eps,width=.47\textwidth}%,height=60mm}  
417: \epsfig{file=fig4.eps,width=.47\textwidth,clip=}%,height=60mm}  
418: \end{center}  
419: \caption[]{Left eigenfunction of map with outer holes of size 
420: $\epsilon=10^{-3}$ for the leading
421: eigenvalue $\lambda_0$. The hierarchy of peaks follows the pre-images of the holes.
422: }
423: \label{first_right_eigenfctn_eps_1.e-3}
424: \end{figure}
425:  
426: %The largest eigenvalue $0.996845$ corresponds to an escape rate of 
427: %$3.155\cdot10^{-3}$ and has to be compared with 
428: %$3.125\cdot10^{-3}$ from Fig.~\ref{escaperate_nonoise}. (1\% error, seems ok)
429: 
430: 
431: %==================================================
432: \subsection{Noisy map without holes}
433: For the map with noise we add Gaussian distributed random numbers at
434: each time step,
435: \beq
436: x_{n+1} = f(x_n) + \xi_n \, ,      
437: \eeq
438: where the $\xi_n$ are independent and identically distributed according to 
439: \beq
440: p(\xi) = \frac{1}{\sqrt{\pi \sigma^2}} e^{-\xi^2/\sigma^2} \, ,      
441: \eeq
442: with $\sigma=\sqrt{2Dt}$ the amplitude of the noise.
443: Since it then is possible to leave the interval $[0,1]$ 
444: we close the system periodically by mapping points outside the interval
445: back in using the modulo operation.
446: %For the case of a the noisy map a Gaussian random
447: %displacement of appropriate variance is added to the final point.
448: The case with noise is the more regular one, and if the width of the
449: characteristic function for the hole region is a fraction of the noise 
450: level the vectors converge rather reliably, as a comparison
451: between the results for matrix sizes $N=3000$ and $N=4000$ for noise 
452: amplitude $10^{-3}$ showed.
453: 
454: %The operator K is constructed by first doing the deterministic step and then adding noise. 
455: The eigenfunctions and eigenvalues in the presence of noise of amplitude
456: $\sigma=10^{-3}$ are shown in 
457: Fig.~\ref{first_two_eigenfcts_noise_1.e-3}.
458: The first eigenvalue remains at $1$, the first left eigenvector is uniform, but the 
459: step in the first right eigenvector is smoothed out. With increasing noise
460: amplitude this transition region becomes wider, as evidenced by the magnifications in Fig.~\ref{smooth2}.
461: \begin{figure}%[hbt]
462: \begin{center}  
463: %\epsfig{file=/usr/ax1321/faisst/rausch/tent/ernsthaft/spektrum_improved_3/data/EV_noise0.001_nohole.eps,width=.47\textwidth}%,height=60mm}  
464: %\epsfig{file=EV_noise0.001_nohole.eps,width=.47\textwidth}%,height=60mm}  
465: \epsfig{file=fig5.eps,width=.47\textwidth,clip=}%,height=60mm}  
466: \end{center}  
467: \caption[]{Eigenfunctions for the noisy map without holes. The noise
468: amplitude is $\sigma=10^{-3}$. The upper row
469: corresponds to the leading eigenvalue $\lambda_0=1$, the lower to the 
470: next to leading eigenvalue $\lambda_1=-0.6073$.
471: }
472: \label{first_two_eigenfcts_noise_1.e-3}
473: \end{figure}
474: 
475: \begin{figure}%[htb]
476: \begin{center}  
477: %\epsfig{file=/usr/ax1321/faisst/rausch/tent/ernsthaft/spektrum_improved_2/data/smooth2.eps,width=.47\textwidth}%,height=60mm}  
478: %\epsfig{file=smooth2.eps,width=.47\textwidth}%,height=60mm}  
479: \epsfig{file=fig6.eps,width=.47\textwidth,clip=}%,height=60mm}  
480: \end{center}  
481: \caption[]{Magnification of the invariant density for the noisy map without holes. 
482: Noise amplitudes are $\sigma=10^{-5}$ (sharpest transition), $10^{-3}$, $10^{-2}$, and $10^{-1}$ (widest transition).
483: }
484: \label{smooth2}
485: \end{figure}
486: %==================================================
487: \subsection{Noisy map with holes}
488: We finally come to our model for a noisy repellor, the noisy map with holes.
489: Fig.~\ref{escaperate_eps_1.e-3} shows for holes of size 
490: % inconsistent: $0.001$ 
491: $\epsilon = 10^{-3}$
492: the change 
493: in the escape rates as a function of noise amplitude. 
494: For outer holes the escape rate decreases, for inner holes it increases, until
495: for a noise amplitude of about $10^{-2}$ they almost coincide.
496: This is exactly what one would expect from the changes in invariant density
497: shown in Fig.~\ref{smooth2}: the density on the upper level decreases
498: and the one on the lower increases with the corresponding changes in 
499: lifetime. 
500: If we had not closed the interval to a circle the loss of trajectories at the 
501: end of the intervals would have swamped this effect and the 
502: escape rate would have increased monotonically with noise level.
503: 
504: \begin{figure}%[hbt]
505: \begin{center}  
506: %\epsfig{file=/usr/ax1321/faisst/rausch/tent/ernsthaft/data/escaperate_eps_1.e-3.eps,width=.47\textwidth}%,height=60mm}  
507: %\epsfig{file=escaperate_eps_1.e-3.eps,width=.47\textwidth}%,height=60mm}  
508: \epsfig{file=fig7.eps,width=.47\textwidth,clip=}%,height=60mm}  
509: \end{center}  
510: \caption[]{Escape rate $\gamma$ 
511: divided by the hole size $\epsilon = 10^{-3}$
512: for the noisy map 
513: with holes. When the holes
514: are in the high density region (outer holes) the escape rate decreases, and
515: when they are in the low density region (inner holes) it increases. When noise
516: level $\sigma$ and hole size are comparable there is no difference between inner and 
517: outer hole placement anymore.
518: }
519: \label{escaperate_eps_1.e-3}
520: \end{figure}
521: 
522: The associated eigenfunctions are shown in Fig.~\ref{first_two_eigenfcts_eps_1.e-3_noise_1.e-3} for 
523: outer holes of size $10^{-3}$ together with a noise amplitude 
524: $10^{-3}$. Compared to the noise free case
525: structures are smoothed out: for instance, the amplitudes of left eigenfunctions decrease considerably.
526: 
527: \begin{figure}%[hbt]
528: \begin{center}  
529: %\epsfig{file=/usr/ax1321/faisst/rausch/tent/ernsthaft/spektrum_improved_3/data/EV_noise0.001_eps0.001.eps,width=.47\textwidth}%,height=60mm}  
530: %\epsfig{file=EV_noise0.001_eps0.001.eps,width=.47\textwidth}%,height=60mm}  
531: \epsfig{file=fig8.eps,width=.47\textwidth,clip=}%,height=60mm}  
532: \end{center}  
533: \caption[]{Eigenfunctions of the map with noise amplitude $\sigma=10^{-3}$ and 
534: with outer holes of size $\epsilon = 10^{-3}$. The eigenvalues are
535: $\lambda_1=0.99745$ for the upper graph and $\lambda_2=-0.61138$ for the lower one.
536: }
537: \label{first_two_eigenfcts_eps_1.e-3_noise_1.e-3}
538: \end{figure}
539: 
540: \subsection{Failure of perturbation theory}
541: 
542: First order perturbation theory does not always work that well.
543: Consider another map, $x_{n+1}=g(x_n)$ on the interval $[0,1]$ with
544: \beq
545: g(x) = 
546: \left\{
547: \begin{array}{cl}
548: 1/3+2x & 0<x\le 1/3
549: \\
550: 3(1-x)/2 &  1/3<x \le 1
551: \end{array}
552: \right. \, .      
553: \label{g}
554: \eeq
555: Its invariant density is (Fig.~\ref{asymm}): 
556: \beq
557: \rho(x) = 
558: \left\{
559: \begin{array}{cl}
560: 3/4, & 0 < x \le 1/3
561: \\
562: 9/8,   &  1/3 <x \le 1        
563: \end{array}
564: \right. \, .      
565: \label{rho_3}
566: \eeq
567: 
568: \begin{figure}[b]
569: \begin{center}  
570: %%%\epsfig{file=/usr/ax1321/faisst/rausch/tent/data/asymm.eps,width=.47\textwidth}
571: %\epsfig{file=/usr/ax1321/faisst/rausch/tent/data/asymm2.eps,width=.47\textwidth}
572: %\epsfig{file=asymm2.eps,width=.47\textwidth}
573: \epsfig{file=fig9.eps,width=.47\textwidth,clip=}
574: \end{center}  
575: \caption[]{The map (\ref{g}) (dashed) and its invariant density (continuous).
576: %done {\bf Diese Abb sollte an 1 angepasst werden: also invariante Dichte = Linie,
577: %Abbildung=gestrichelt}
578: }
579: \label{asymm}
580: \end{figure}
581: 
582: \begin{figure}[htb]
583: \begin{center}  
584: %\epsfig{file=/usr/ax1321/faisst/rausch/tent/ernsthaft/spektrum_improved_3/data/EV_asymmetricmap_nonoisenohole.eps,width=.47\textwidth}%,height=60mm}  
585: %\epsfig{file=EV_asymmetricmap_nonoisenohole.eps,width=.47\textwidth}%,height=60mm}  
586: \epsfig{file=fig10.eps,width=.47\textwidth,clip=}%,height=60mm}  
587: \end{center}  
588: \caption[]{Eigenfunctions for the leading eigenvalue $\lambda_0=1$ and the next to leading eigenvalue $\lambda_1=-2/3$
589: for map (\ref{g}). }
590: \label{eigenfctns_asymm_map_noholes}
591: \end{figure}
592: 
593: Eigenfunctions for the two leading eigenvalues are shown in 
594: Fig.~\ref{eigenfctns_asymm_map_noholes}.
595: Again we take a `left' hole at $[1/3-\epsilon,1/3]$ in the low density region
596: and a `right' hole at $[1/3,1/3+\epsilon]$ in the high density region.
597: First order perturbation theory predicts the decay rates $\gamma$ 
598: as function of hole size $\epsilon$ to be
599: \bea 
600: \gamma_r &=& \epsilon \rho_r = 9/8 \epsilon \\
601: \gamma_l &=& \epsilon \rho_l = 3/4 \epsilon \, . 
602: \eea 
603: The numerical values extracted from Fig.~\ref{escaperates_vs_eps_nonoise} 
604: are 
605: \bea 
606: \gamma_r &=& %\epsilon \rho_r = 
607: 0.87 \epsilon \\
608: \gamma_l &=& %\epsilon \rho_l = 
609: 0.75 \epsilon \, .  
610: \eea 
611: In agreement with perturbation theory the escape rate is proportional
612: to the hole size. 
613: The pre-factors from the two calculations agree for the left hole in the low, but
614: disagree for the right one in the high density region. 
615: The deviation becomes smaller when the holes are
616: partially closed, i.e., the parameter $p$ in (\ref{eli_pro}) is set to $0.99$.
617: Then the perturbative and numerical escape rates are in better 
618: agreement. This failure of the perturbative estimate seems to be
619: closely connected with the presence of a periodic orbit on the 
620: border of the interval. If the opening is shifted to slightly
621: larger values then the slope agrees again with the perturbative results.
622: \begin{figure}%[hbt]
623: \begin{center}  
624: %\epsfig{file=/usr/ax1321/faisst/rausch/tent/ernsthaft/escape_rate/data/escaperate_vs_eps_nonoise.eps,width=.47\textwidth}
625: %\epsfig{file=escaperate_vs_eps_nonoise.eps,width=.47\textwidth}
626: \epsfig{file=fig11.eps,width=.47\textwidth,clip=}
627: \end{center}  
628: \caption[]{Escape rates $\gamma$ 
629: divided by hole size $\epsilon$ and by $(1-p)$
630: for the map (\ref{g}) vs.~hole size for two placements
631: of the hole and two extraction rates $p$. 
632: The dashed horizontal lines mark the theoretical values $9/8$ and $3/4$. For the left hole the escape rate is in
633: agreement with perturbation theory (circles, for p=0). For the opening to the
634: right at $p=0$ (squares) the pre-factor does not agree with perturbation theory.
635: For $p=0.99$ the values move up to the first order perturbation theory
636: result (triangles). If the position of the opening is moved away from the
637: critical point $x=1/3$, e.g.~to the interval $[0.35,0.35+\epsilon]$, then
638: the perturbative results is also obtained for $p=0$ (diamonds).
639: }
640: \label{escaperates_vs_eps_nonoise}
641: \end{figure}
642: 
643: \section{Final remarks\label{sec_fr}}
644: Within the simple models studied here the hypothesis that the variation of 
645: lifetimes can be related to inhomogeneities in the invariant density of the 
646: unperturbed attractor could be confirmed. Such inhomogeneities are most 
647: pronounced near bifurcations and crises \cite{Grebogi83}, 
648: as in the work of Franaszek \cite{Franaszek91}.
649: The placement of holes from the outside is less artifical than it may seem. In 
650: the case of riddling bifurcations \cite{riddles} line attractors are broken up by
651: holes that appear near periodic points that are no longer transversally stable, 
652: and both position and widths can
653: be controlled externally. Thus the observations discussed here have
654: some bearing on the effects of noise on the lifetimes in riddled attractors.
655: 
656: 
657: Support from the Deutsche Forschungsgemeinschaft is greatfully acknowledged.
658: 
659: 
660: 
661: 
662: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%  
663: % References
664: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
665: 
666: \begin{thebibliography}{99}  
667: 
668: \bibitem{Haenggi90}
669: P.~H\"anggi{,} P.~Talkner{,} M.~Borkovec.
670: % \newblock Reaction-rate theory: fifty years after kramers.
671: \newblock { Rev. Mod. Phys.}, {\bf 62}, 251, (1990). 
672: 
673: \bibitem{Martin}
674: B. Shraiman, C.E. Wayne and P.C. Martin,
675: Phys. Rev. Lett. {\bf 46}, 935 (1981).
676: % Scaling theory for noisy period-doubling transitions to chaos
677: 
678: \bibitem{Haenggi98} 
679: L.~Gammaitoni{,} P.~H\"anggi{,} P.~Jung{,} F.~Marchesoni.
680: \newblock {Rev. Mod. Phys.}, {\bf 70}, 223--287, (1998).
681: %p. H\"anggi,
682: %his Rev. Mod. Phys. on stochastic resonance
683: 
684: \bibitem{arnold83}
685: L.~Arnold{,} H.~Crauel{,} V.~Wihstutz.
686: \newblock {SIAM J. Control Optim.}, {\bf 21}, 451, (1983)  
687: 
688: \bibitem{Franaszek91}
689: M.~Franaszek.
690: \newblock {Phys. Rev. A}, {\bf 44}, 4065, (1991).   
691: 
692: M.~Franaszek{,}~L. Fronzoni.
693: \newblock {Phys. Rev. E}, {\bf 49}, 3888, (1994).
694: 
695: \bibitem{Tel}
696: T. T\'el, Directions in Chaos, edited by Hao Bai-Lin
697: (World Scientific, Singapore, 1990), Vol 3, pg. 149
698: % a general review of repellors
699: 
700: \bibitem{Gaspard}
701: P. Gaspard, I. Claus, T. Gilbert and J.R. Dorfman,
702: Phys. Rev. Lett. {\bf 86}, 1506 (2001).
703: 
704: P. Gaspard, Chaos, scattering and statistical mechanics,
705: (Cambridge University Press, Cambridge, 1998).
706: 
707: S. Tasaki and P. Gaspard, J. Stat. Phys. {\bf 81}, 935 (1995)
708: 
709: P. Gaspard, G. Nicolis, A. Provata and S. Tasaki, 
710: Phys. Rev. E {\bf 51}, 74 (1995).
711: 
712: \bibitem{Paar97}
713: V.~Paar{,}~N. Pavin.
714: \newblock {Phys. Rev. E}, {\bf 55}, 4112, (1997).
715: 
716: \bibitem{Grebogi83}
717: C.~Grebogi{,} E.~Ott{,} J.A.~Yorke,
718: \newblock {Physica D}, {\bf 7}, 181 (1983)
719: % their crisis paper
720: 
721: \bibitem{riddles}
722: A.S. Pikovsky and P. Grassberger, J. Phys. A {\bf 24}, 4587 (1991)
723: 
724: E. Ott, J.C. Alexander, I. Kan, J.C. Sommerer and J.A. Yorke,
725: Physica D {\bf 76} 384 (1994)
726: 
727: \bibitem{Arnold91}
728: L.~Arnold{,} P.~Boxler.
729: %\newblock Lyapunov exponents.
730: \newblock In {\em Lect. Notes Math.}, {\bf 1486}, 159, Springer, (1991). 
731: 
732: \bibitem{Crauel00}
733: H.~Crauel.
734: \newblock {Arch. Math.}, {\bf 75}, 472 (2000).
735: 
736: 
737: \end{thebibliography}  
738: 
739: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%  
740: % Figures  
741: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
742: 
743: \end{document}
744: 
745: