Invariant measures for typical quadratic maps

Marco Martens, Tomasz Nowicki

This academic paper, published in 1996 by Marco Martens and Tomasz Nowicki, represents a significant milestone in the field of one-dimensional dynamical systems. It provides a rigorous geometric bridge to understand when chaotic systems—specifically quadratic maps—exhibit stable statistical behavior.

TL;DR

The paper establishes a sufficient geometric condition, called "summable scaling factors," for a unimodal map to possess an absolutely continuous invariant probability measure (ACIPM). This result is a cornerstone in proving that "typical" quadratic maps (those we are likely to encounter in nature or simulations) have well-behaved, predictable long-term statistical distributions.

What Problem Does This Paper Address?

In the study of chaos, researchers often ask: "Is the behavior we see 'typical'?" In mathematics, "typical" can mean two different things. It might mean "topologically generic" (true for a dense set of parameters) or "Lebesgue typical" (true for a set of parameters with full measure, like picking a random number and almost never hitting a specific integer).

For the quadratic family of maps—functions like $f(x) = ax(1-x)$ that model everything from population growth to electronic circuits—the goal is to determine if they have an ACIPM. An ACIPM is crucial because it allows us to use the tools of ergodic theory to predict the long-term average behavior of the system. Without it, the system might be "messy" in a way that defies statistical description, such as orbits getting trapped on zero-measure Cantor sets.

The challenge was to find a condition that links the physical geometry of the map to the existence of these "nice" measures.

Methods

The authors employ a purely analytical and geometric approach focused on the critical point (the "peak" of the quadratic curve).

  1. Central Intervals ($U_n$): They define a sequence of nested intervals around the critical point. These are "nice" intervals where the first return map behaves predictably.
  2. Scaling Factors ($\sigma_n$): They introduce scaling factors, defined as the ratio of the lengths of these nested intervals: $\sigma_n = u_{n+1}/u_n$. These factors describe how the geometry of the map "shrinks" as you zoom in on the critical point.
  3. Telemann Decomposition: To bridge geometry and dynamics, they use a technique called Telemann decomposition to track the orbit of the critical point.
  4. Schwarzian Derivative: They utilize the property of a negative Schwarzian derivative (a standard tool in 1D dynamics) to control "distortion"—ensuring that the map doesn't stretch or compress space too wildly in a way that would break the geometric estimates.

Datasets

Not applicable. This is a theoretical mathematics paper; it relies on formal proofs and logical derivation rather than empirical data or computational benchmarks.

Key Insights and Ideas

  • Geometry Governs Expansion: The central "aha moment" is that the rate at which central intervals shrink (the scaling factors) directly controls the expansion of the derivative along the critical orbit. If the intervals shrink fast enough, the system "expands" enough to spread orbits out smoothly, creating an ACIPM.
  • Summability is the Key: The paper proves that if the scaling factors are "summable" (specifically, if the series $\sum \sigma_n^{1/\alpha}$ converges), then the map must have an ACIPM.
  • Finitely Many Central Returns: A major takeaway is that if a map only returns to its "central" zone a finite number of times, it automatically satisfies the condition for having an ACIPM. This covers a vast majority of the parameter space.
  • Closing the Loop on Typicality: By combining their result with work by Mikhail Lyubich, the authors show that for almost every parameter in the quadratic family, the map is either periodic (predictable) or has an ACIPM (statistically predictable).

Key Results

  • Theorem B: For any $S$-unimodal map with a critical exponent $\alpha > 1$, if the scaling factors are summable, an ACIPM exists.
  • Theorem A (The "Typicality" Result): In the quadratic family ($\alpha = 2$), a Lebesgue typical map has a unique Sinai-Bowen-Ruelle (SBR) measure. This means that for almost any starting point, the long-term distribution of the orbit is governed by a single, well-defined measure.
  • Classification: The paper helps classify the quadratic family into three buckets: those with periodic attractors, those that are "infinitely renormalizable" (living on Cantor sets), and those with ACIPMs. They show the third bucket is "typical" for parameters not in the first two.

Limitations and Future Work

  • Specific to Quadratic Maps: While the main theorem (Theorem B) applies to any critical exponent $\alpha > 1$, the proof that this condition is "typical" (Theorem A) was specifically tailored to the quadratic case ($\alpha = 2$).
  • The "Conjecture": The authors include a conjecture that the same "typicality" results should hold for all critical exponents $\alpha > 1$. Proving this for higher-order polynomials or different families of maps remained an open challenge at the time.
  • Edge Cases: The "Johnson-Examples"—maps with infinitely many central returns that don't have ACIPMs—exist as a strange, zero-measure boundary case that continues to interest researchers in singular measures.

Summary updated

Abstract

A sufficient geometrical condition for the existence of absolutely continuous invariant probability measures for S-unimodal maps will be discussed. The Lebesgue typical existence of such measures in the quadratic family will be a consequence.
BibTeX
@inproceedings{martens2018invariant,
  author = {Marco Martens and Tomasz Nowicki},
  title = {{Invariant measures for typical quadratic maps}},
  booktitle = {Astérisque},
  year = {2018},
  doi = {10.24033/ast.475},
}

Loading related content...

Figure