A surprising discrepancy in the regularity of conjugacies between generalized interval exchange transformations and their inverses at freezing

This paper demonstrates a surprising asymmetry in the regularity of conjugacies for generalized interval exchange transformations under freezing limits, showing that while the conjugacy can become arbitrarily irregular, its inverse remains uniformly Hölder continuous.

The Gist

Imagine you have a long, colorful ribbon. You cut it into several pieces, shuffle them around in a specific pattern, and tape them back together to form a new ribbon of the same length. This is a basic mathematical game called an Interval Exchange Transformation (IET). It's like a perfect, mechanical dance where every piece moves exactly the same distance.

Now, imagine a slightly more chaotic version of this game. Instead of just shuffling the pieces, you also stretch some of them and shrink others as you move them. This is called a Generalized Interval Exchange Transformation (GIET), or more specifically, an Affine one (AIET). It's the same dance, but the dancers are stretching their arms and legs as they move.

The Big Question: How Smooth is the Connection?

Mathematicians have long known that if you have this chaotic, stretching dance (the AIET), you can usually find a "translator" to explain how it relates back to the perfect, non-stretching dance (the IET). This translator is a map called a conjugacy (let's call it $h$).

Think of $h$ as a rubber sheet that you stretch over the chaotic dance to make it look like the perfect dance.

• If you look at the rubber sheet from the chaotic side to the perfect side, how "rough" or "smooth" is it?
• If you look at it from the perfect side back to the chaotic side (the inverse, $h^{-1}$), how rough is it?

Usually, mathematicians expected that if the rubber sheet is very rough in one direction, it would be equally rough in the other. They thought the "smoothness" (mathematically called Hölder regularity) would be a two-way street.

The Surprise: A One-Way Street of Roughness

This paper, by Krzysztof Frączek and Łukasz Kotlewski, discovers a shocking exception to that rule. They found a specific family of these stretching dances where the "roughness" behaves completely differently depending on which way you look.

Here is the analogy:
Imagine a fractal coastline.

• If you try to walk along the coastline in one direction (the conjugacy $h$), the path becomes so jagged and broken that you can barely take a step without tripping. As the "stretching" parameter in their experiment gets larger (approaching what they call a "freezing" or zero-temperature limit), this path becomes infinitely jagged. The smoothness drops to zero.
• However, if you turn around and walk back along that same coastline in the opposite direction (the inverse $h^{-1}$), the path remains surprisingly smooth and walkable. It never gets too jagged; it stays within a safe, predictable level of roughness.

The Main Discovery:
The authors proved that for certain self-similar, hyperbolic dances, you can make the connection to the perfect dance arbitrarily terrible (infinitely rough) in one direction, while the connection in the opposite direction remains perfectly decent (uniformly smooth).

How They Did It: The "Freezing" Experiment

To find this, the authors used a concept from physics called thermodynamic formalism.

• Imagine the stretching of the ribbon is controlled by a "temperature" dial.
• They turned this dial up to "infinity" (a "zero-temperature" or "freezing" limit).
• As the system "froze," the chaotic stretching became extreme.
• Using complex math involving "Gibbs measures" (which are like probability maps of where the dancers are most likely to be), they calculated exactly how the smoothness changed.

They found that as the "temperature" dropped:

1. The smoothness of the map $h$ (chaotic $\to$ perfect) vanished, dropping to zero.
2. The smoothness of the map $h^{-1}$ (perfect $\to$ chaotic) stayed high, bounded by a specific positive number.

The "Why" and the "How Much"

The paper doesn't just say "it happens"; it gives a precise recipe for how much it happens.

• They calculated the exact rate at which the roughness increases in the bad direction.
• They calculated the exact "safety limit" of smoothness in the good direction.
• They even built a concrete example using a 5-piece ribbon shuffle (a 5-IET) and used a computer to prove that the "safety limit" is about 0.64. This means the inverse map is definitely smooth enough to be useful, while the forward map becomes a mess.

Summary in Plain English

Think of a funhouse mirror.

• Usually, if a mirror distorts your reflection badly in one direction, it distorts it just as badly if you look at it from the other side.
• This paper found a magical, mathematical funhouse mirror where, if you look at it from the "stretching" side, your reflection is a terrifying, jagged monster.
• But if you look at it from the "perfect" side, your reflection is still a recognizable, smooth human face.

The authors showed that this extreme asymmetry isn't just a fluke; it's a fundamental property of these specific mathematical systems, and they provided the exact formulas to predict exactly how distorted the reflection gets as you crank up the "stretching" knob.

Technical Summary

Technical Summary: A Surprising Discrepancy in the Regularity of Conjugacies Between Generalized Interval Exchange Transformations and Their Inverses at Freezing

Problem Statement
The paper addresses a fundamental problem in one-dimensional dynamics concerning the regularity of conjugacies between Generalized Interval Exchange Transformations (GIETs) and Interval Exchange Transformations (IETs). While GIETs are semi-conjugate to IETs under specific combinatorial conditions (8-complete Rauzy–Veech combinatorics), the nature of this conjugacy $h$ and its inverse $h^{-1}$ remains a central question in Herman's rigidity program.

Standard intuition suggests that the Hölder regularity of a homeomorphism and its inverse should degenerate simultaneously. However, this paper investigates whether this symmetry holds in low-regularity regimes. Specifically, the authors seek to determine if there exist natural families of Affine Interval Exchange Transformations (AIETs) where the supremal Hölder exponent of the conjugacy $H(h)$ and that of its inverse $H(h^{-1})$ exhibit a significant, asymmetric disparity. Previous results (e.g., Cobo) suggest that for typical IETs, if the log-slope vector is not stable, neither the conjugacy nor its inverse is Hölder, while stable vectors yield $C^{1+\delta}$ regularity. The authors hypothesize that such a disparity can be observed by restricting attention to specific, non-typical settings: self-similar IETs of hyperbolic periodic type.

Methodology
The authors employ a synthesis of dynamical systems theory and thermodynamic formalism to analyze the asymptotic behavior of these conjugacies as a parameter $t$ tends to infinity (the "freezing" or zero-temperature limit).

1. Setting: The study focuses on AIETs $f_{t\omega}$ belonging to a one-parameter family $Aff(T, t\omega)$, where $T$ is a self-similar IET of hyperbolic periodic type and $\omega$ is a central log-slope vector (invariant under the self-similarity matrix $M$).
2. Thermodynamic Formalism: The problem is recast using the thermodynamic formalism for shifts of finite type (SFT). The Rauzy–Veech renormalization scheme is modeled by a shift space $X$ and a locally constant potential $\Phi_\omega$ derived from the log-slope vector.
3. Zero-Temperature Limits: The authors utilize results from the theory of zero-temperature limits (freezing) for Gibbs measures. As the inverse temperature parameter $t \to +\infty$, the behavior of the system is governed by maximizing and minimizing subshifts ($X(\Phi_\omega)$ and $X(-\Phi_\omega)$) and the associated ergodic averages of the potential.
4. Explicit Formulas: Building on recent explicit formulas for Hausdorff dimensions of invariant and conformal measures (from [1]), the authors relate these dimensions to the supremal Hölder exponents of the conjugacy and its inverse. The conjugacy $h$ is identified as the distribution function of the invariant measure $\mu$, while $h^{-1}$ corresponds to the conformal measure $\nu$.

Key Contributions and Results
The paper establishes a strong asymmetry in the regularity of conjugacies and their inverses in the freezing limit. The main results are summarized in Theorem 1.1 and Theorems 5.3–5.4:

• Asymptotic Disparity: For the family $f_{t\omega}$ as $t \to +\infty$:

  • The supremal Hölder exponent of the inverse conjugacy, $H(h^{-1}_{t\omega})$, converges to a strictly positive constant:
    $$\lim_{t \to +\infty} H(h^{-1}_{t\omega}) = \frac{h_{top}(X(\Phi_\omega))}{h_{top}(X)}$$
    where $h_{top}$ denotes topological entropy.
  • In contrast, the supremal Hölder exponent of the conjugacy, $H(h_{t\omega})$, decays to zero. Specifically, the product $t \cdot H(h_{t\omega})$ converges to a finite, non-zero constant:
    $$\lim_{t \to +\infty} t \cdot H(h_{t\omega}) = \frac{h_{top}(X)}{\Phi_\omega - \Phi_\omega}$$
    (where $\Phi_\omega$ and $\Phi_\omega$ are the maximal and minimal ergodic averages of the potential).

• Hausdorff Dimensions: The paper provides sharp asymptotics for the Hausdorff dimensions of the invariant measure $\mu_{t\omega}$ and the conformal measure $\nu_{t\omega}$, showing that $\dim_H(\nu_{t\omega})$ converges to the same limit as $H(h^{-1}_{t\omega})$, while $\dim_H(\mu_{t\omega})$ decays at a rate of $1/t$.

• Explicit Example: In Section 6, the authors construct a concrete example using a 5-IET of hyperbolic periodic type. Through computational analysis of the associated graph and subshifts, they demonstrate a case where the limiting constant for the inverse conjugacy is approximately $0.64$ (specifically $\log 3 / \log \theta_0$), while the conjugacy's regularity vanishes.

Significance
The paper claims to exhibit a "strongly asymmetric behavior" contrary to the usual expectation that Hölder regularities degenerate simultaneously. By constructing a specific one-parameter family of AIETs, the authors demonstrate that it is possible for the inverse conjugacy to remain uniformly Hölder (with a positive exponent) while the conjugacy itself becomes arbitrarily irregular (with an exponent tending to zero).

This result challenges the assumption of symmetry in the regularity of dynamical conjugacies in low-regularity regimes. The authors note that this phenomenon is specific to the self-similar setting with hyperbolic periodic types and central log-slope vectors, distinguishing it from typical scenarios where such disparities are unlikely to occur. The work provides precise quantitative descriptions of this disparity using thermodynamic formalism, linking the decay rates of regularity to the topological entropies of maximizing subshifts.