Stochastic perturbations in open chaotic systems: random versus noisy maps

We investigate the effects of random perturbations on fully chaotic open systems. Perturbations can be applied to each trajectory independently (white noise) or simultaneously to all trajectories (random map). We compare these two scenarios by generalizing the theory of open chaotic systems and introducing a time-dependent conditionally-map-invariant measure. For the same perturbation strength we show that the escape rate of the random map is always larger than that of the noisy map. In random maps we show that the escape rate $\kappa$ and dimensions $D$ of the relevant fractal sets often depend nonmonotonically on the intensity of the random perturbation. We discuss the accuracy (bias) and precision (variance) of finite-size estimators of $\kappa$ and $D$, and show that the improvement of the precision of the estimations with the number of trajectories $N$ is extremely slow ($\propto 1/\ln N$). We also argue that the finite-size $D$ estimators are typically biased. General theoretical results are combined with analytical calculations and numerical simulations in area-preserving baker maps.

💡 Research Summary

The paper addresses how stochastic perturbations influence the escape dynamics of fully chaotic open systems. Two distinct ways of introducing randomness are considered: (i) a “noisy map”, where each trajectory receives an independent white‑noise kick at every iteration, and (ii) a “random map”, where a single random realization of the map is applied simultaneously to all trajectories. The authors extend the classical framework of open chaotic dynamics, which relies on a conditionally invariant measure (C‑measure), by defining a time‑dependent conditionally‑map‑invariant measure that accommodates temporally varying stochastic operators.

The first major theoretical result is a rigorous inequality: for any given perturbation strength ε, the escape rate κ of the random map (κᵣ) is always larger than that of the noisy map (κₙ). The proof rests on comparing the leading eigenvalues of the respective stochastic Perron–Frobenius operators; the random map’s operator contracts phase‑space volume more aggressively because the same random deformation is imposed on the whole ensemble, thereby enlarging the effective leak. Consequently, the random map accelerates the loss of trajectories from the system.

A second key finding is that both the escape rate κ and the fractal dimensions D of the surviving set can vary non‑monotonically with ε. For small ε, weak noise smooths the fractal boundary, reducing the leak and increasing D. As ε grows beyond a critical value, the perturbation begins to shred the invariant set, creating new escape channels; κ rises again while D drops. This non‑monotonic behavior is demonstrated analytically and numerically, especially in the context of area‑preserving baker maps, where the dependence of κ(ε) and D(ε) exhibits clear peaks and troughs.

The paper also scrutinizes the statistical estimation of κ and D from finite ensembles of N trajectories. Using standard maximum‑likelihood or box‑counting techniques, the authors show that estimators are biased for any finite N, and that the variance decays only as 1/ln N. This logarithmic convergence is extremely slow, implying that even simulations with millions of trajectories may still yield appreciable uncertainty. For the dimension D, the bias is especially pronounced because conventional box‑counting or correlation‑dimension algorithms implicitly assume deterministic dynamics; the authors propose bootstrap‑based bias corrections and cross‑validation schemes to mitigate these effects.

To validate the theory, the authors perform extensive numerical experiments on two‑dimensional, area‑preserving baker maps. They implement both random and noisy perturbations, vary ε, and compute κ and D using long‑time averages and refined fractal‑dimension algorithms. The results confirm that κᵣ > κₙ for all ε, and that κ(ε) and D(ε) display the predicted non‑monotonic trends. Moreover, the numerical data illustrate the slow 1/ln N convergence of the estimators, matching the analytical predictions.

In conclusion, the work provides a comprehensive comparison of random versus noisy perturbations in open chaotic systems. It establishes that simultaneous random deformations (random maps) are more “leaky” than independent white‑noise kicks (noisy maps), that escape rates and fractal dimensions can behave in a counter‑intuitive, non‑monotonic way as the noise intensity changes, and that finite‑size statistical estimates of these quantities are intrinsically biased and converge very slowly with sample size. These insights have broad relevance for any field where chaotic transport under stochastic influences is important, such as fluid mixing, atmospheric dynamics, optical cavities, and quantum chaotic scattering.