Abstract
We prove that a class of one-dimensional maps with an arbitrary number of non-degenerate critical and singular points admits an induced Markov tower with exponential return time asymptotics. In particular, the map has an absolutely continuous invariant probability measure with exponential decay of correlations for Hölder observations.
Introduction: The Challenge of Mixed Dynamics
In the field of dynamical systems, a central goal is to characterize the long-term statistical behavior of deterministic rules. For one-dimensional interval maps, this typically involves establishing the existence of an absolutely continuous invariant probability measure (acip). When such a measure exists and exhibits rapid "mixing," the system's long-term evolution can be effectively modeled using probability theory, despite its underlying deterministic nature.
Historically, mathematical research has treated two categories of maps separately:
- Maps with critical points: Smooth maps where the derivative vanishes at certain points (e.g., the logistic map). These points cause local contraction.
- Maps with singularities: Maps containing discontinuities or points where the derivative becomes infinite (e.g., Lorenz-like maps). These points cause extreme local expansion.
The primary challenge lies in the simultaneous presence of both types of points. Critical points "contract" space, while singularities "expand" it violently. Both phenomena can disrupt the "bounded distortion" property required for rigorous ergodic proofs. This paper provides a unified mathematical framework to handle an arbitrary number of both critical and singular points, bridging the gap between these seemingly opposite behaviors.
Methodology: Inducing Expansion via Markov Towers
The authors utilize the technique of inducing a Markov tower (specifically a Young tower). Rather than analyzing the map $f$ at every discrete step, they focus on an "induced map" $F = f^T$. This map records the state only when a point returns to a designated reference interval after $T$ iterations.
To control the competing effects of contraction and expansion, the authors implement several key strategies:
- Binding Periods: When an orbit enters a neighborhood of a critical point, it loses expansion. The authors employ a "binding" technique where the orbit shadows the path of the critical point until it recovers its expansive properties.
- Chopping Procedure: To maintain "bounded distortion"—ensuring the map does not stretch parts of an interval infinitely more than others—a systematic algorithm partitions the interval into smaller sub-intervals whenever an orbit approaches a critical or singular point too closely.
- Expansion and Recurrence Assumptions: The framework relies on three core assumptions: expansion outside small neighborhoods of the critical/singular set (H1), a Collet-Eckmann type condition ensuring critical orbits maintain exponential derivative growth (H2), and topological transitivity (H3).
Key Theoretical Results
The "aha moment" of this research is the realization that critical points (where the derivative $\to 0$) and singular points (where the derivative $\to \infty$) can be managed using the same combinatorial machinery. The "chopping" and "binding" logic effectively regulates how the system returns to a state of large-scale expansion.
The paper establishes several landmark results for this class of maps:
- Exponential Return Times: The "return time" $T$ to the reference set has an exponentially decreasing tail ($P(T > n) < Ce^{-\gamma n}$). In ergodic theory, this is the "gold standard" for proving rapid mixing.
- Existence of Chaotic ACIP: The existence of an ergodic acip $\mu$ with a positive Lyapunov exponent is confirmed, proving the system is truly chaotic.
- Exponential Decay of Correlations: For Hölder continuous observations, the correlation between the current state and a future state vanishes exponentially fast. The system "forgets" its initial conditions rapidly.
- Central Limit Theorem: Time averages of functions along the orbits converge to a Gaussian distribution, mirroring the behavior of independent random variables in classical statistics.
Limitations and Outlook
While providing a robust unified framework, the study notes certain boundaries. The results are restricted to non-degenerate points (where the map behaves like $|x-c|^\ell$ near a point $c$) and do not yet cover "flat" critical points. Additionally, the current framework excludes neutral periodic points (where the derivative is exactly 1), which typically result in slower, polynomial decay of correlations. Extending this unified formalism to "intermittent" systems that combine critical, singular, and neutral points remains a promising direction for future research.
Summary updated