Table of Contents
Fetching ...

First return systems for some continued fraction maps

Thomas A. Schmidt

TL;DR

The paper proves that for all n≥3 and α∈(0,1), the first expansive power of the interval maps T_{3,n,α} is realized as the first return map of the geodesic flow to a cross-section of the unit tangent bundle on the hyperbolic surface associated with G_{3,n}, by reducing to the m=2 Hecke-group setting and employing Arnoux’s method with planar extensions. It establishes a precise entropy–volume correspondence h(T)μ(Ω) = vol T^1(G\H) for the relevant families, and provides a detailed analysis of the entropy function h(T_{2,n,α}) in the m=2 setting, including continuity, symmetry, maximal values, and asymptotic decay as n→∞. The work also corrects a Mass formula in the Rosen-fraction (m=2) case, develops the machinery of matching intervals and planar extensions, and constructs the necessary bridge from Ω_{2,n,1} to Ω_{3,n,1} to finish the conjecture’s proof. The results deepen the connection between continued-fraction-like dynamics and hyperbolic geometry, with explicit, computable entropy profiles and a robust geometric interpretation via geodesic flows. This yields a comprehensive framework for viewing a broad family of interval maps as cross-section return maps on hyperbolic surfaces, enriching both dynamical-systems and geometric-structures perspectives.

Abstract

We prove a conjecture of Calta, Kraaikamp and the author: For all $n\ge 3$, each member of their one-parameter family of interval maps, denoted $T_{3,n,α}$, has its `first expansive return map' of natural extension given by the first return map under the geodesic flow to a section of the unit tangent bundle of the hyperbolic surface uniformized by the underlying Fuchsian group $G_{3,n}$. To achieve the proof, we first prove the corresponding result for analogous one-parameter families related to the Hecke triangle Fuchsian group $G_{2,n}$. A direct comparison per $n$ of the $α=1$ planar domains allows the Hecke group setting to provide sufficient information to prove the conjecture. We also give details about the entropy functions for the Hecke triangle Fuchsian group maps, $α\mapsto h(T_{2,n,α})$. Each is continuous on $(0,1)$, increasing on $(0,1/2)$, decreasing on $(1/2,1)$, with a central interval of constancy. We give precise formulas for the end points of the central intervals and also give precise formulas for the maximal entropy per family. For fixed $α$, the entropy of $T_{2,n,α}$ goes to zero as $n$ tends to infinity.

First return systems for some continued fraction maps

TL;DR

The paper proves that for all n≥3 and α∈(0,1), the first expansive power of the interval maps T_{3,n,α} is realized as the first return map of the geodesic flow to a cross-section of the unit tangent bundle on the hyperbolic surface associated with G_{3,n}, by reducing to the m=2 Hecke-group setting and employing Arnoux’s method with planar extensions. It establishes a precise entropy–volume correspondence h(T)μ(Ω) = vol T^1(G\H) for the relevant families, and provides a detailed analysis of the entropy function h(T_{2,n,α}) in the m=2 setting, including continuity, symmetry, maximal values, and asymptotic decay as n→∞. The work also corrects a Mass formula in the Rosen-fraction (m=2) case, develops the machinery of matching intervals and planar extensions, and constructs the necessary bridge from Ω_{2,n,1} to Ω_{3,n,1} to finish the conjecture’s proof. The results deepen the connection between continued-fraction-like dynamics and hyperbolic geometry, with explicit, computable entropy profiles and a robust geometric interpretation via geodesic flows. This yields a comprehensive framework for viewing a broad family of interval maps as cross-section return maps on hyperbolic surfaces, enriching both dynamical-systems and geometric-structures perspectives.

Abstract

We prove a conjecture of Calta, Kraaikamp and the author: For all , each member of their one-parameter family of interval maps, denoted , has its `first expansive return map' of natural extension given by the first return map under the geodesic flow to a section of the unit tangent bundle of the hyperbolic surface uniformized by the underlying Fuchsian group . To achieve the proof, we first prove the corresponding result for analogous one-parameter families related to the Hecke triangle Fuchsian group . A direct comparison per of the planar domains allows the Hecke group setting to provide sufficient information to prove the conjecture. We also give details about the entropy functions for the Hecke triangle Fuchsian group maps, . Each is continuous on , increasing on , decreasing on , with a central interval of constancy. We give precise formulas for the end points of the central intervals and also give precise formulas for the maximal entropy per family. For fixed , the entropy of goes to zero as tends to infinity.

Paper Structure

This paper contains 50 sections, 17 theorems, 83 equations, 7 figures.

Key Result

Theorem 1

For all $n\ge 3$ and for all $\alpha \in (0,1)$, the product $h(T_{3,n, \alpha})\, \mu(\Omega_{3, n, \alpha})$ equals the volume of the unit tangent bundle of the hyperbolic orbifold uniformized by the triangle group $G_{3,n}$. Furthermore, the first pointwise expansive power of $T_{3, n, \alpha}$ h

Figures (7)

  • Figure 1: Birkhoff sums approximation to entropy functions $\alpha \mapsto h(T_{2,n,\alpha})$, for $n\in \{3, 5,8,50\}$ and $0.1< \alpha \le 1/2$. Horizontal dashed lines at maximum value as per Theorem \ref{['t:mIsTwoSimpleEntropyBehavior']}; vertical dashed lines at $1- R_n/t_{2,n}$ per $n$.
  • Figure 2: The $\mu$-mass of the Rosen planar extension $\Omega_{2s+3}$ of BKS (in red) is comparable to that of $\Omega_{2s+2}$ (in blue), with both going to infinity as $s\to \infty$. See § \ref{['ss:Correction']}.
  • Figure 3: On the left: The planar domain $\Omega_{2,3,1/5}$, and its partitioning 'blocks'; each $\mathcal{B}_i$ fibers over the cylinder where $T_{2,3,1/5}$ is given by $x \mapsto A^iC\cdot x$. The $T_{2,3,1/5}$-orbit of $r_0(1/5) = 1/5$ is labeled by $r_0, r_1$; the $T_{2,3,1/5}$-orbit of $\ell_0(1/5) = -4/5$ is labeled by $\ell_0,\ell_1, \cdots$. One has $r_2(\alpha) = \ell_5(\alpha)$ for other $\alpha$ in the matching interval $J_{k,v} = J_{5,1}$, but here the special equality arises: $r_1 = \ell_4 = 0$. The $y$-values of the horizontal boundaries are derived from appropriate $T_{2,3,\eta_{5,1}}$- and $T_{2,3,\eta_{5,1}}$-orbits, see the discussion of § \ref{['sss:OneFifth']}. The blue rectangle indicates the region $\Omega^{+}_{\alpha, d}$ of Lemma \ref{['l:firstReturnTo2ndQuadrant']}. On the right: The image of $\Omega_{2,3,1/5}$ under $\mathcal{T}_{2,3,1/5}$ is $\Omega_{2,3,1/5}$ itself.
  • Figure 4: Decreasing mass along $J_{1,5}$. With $m=2, n=3$, both $\alpha = 0.18$ and $\alpha' = 0.2$ lie in the matching interval $J_{1, 5}$. Along any matching interval, the coordinates of all tops and bottoms of the rectangles which comprise the various $\Omega_{2,n,\alpha}$ are constant (here $y_1, \dots y_5$ and $y_{-2}, y_{-1}$). The function $\mathcal{T}_A$ maps the rectangle $\mathcal{D} = [\ell_0, \ell'_0)\times [r_{-\overline{S}-1}, y_1]$ to $\mathcal{A} = [r_0, r'_0)\times [r_{-1}, y_{\overline{S}+1}]$, where here we use simplified notation of $\ell_0$ for $\ell_0(\alpha)$ and $\ell'_0$ for $\ell'_0(\alpha)$, and similarly for $r_0, r'_)$. To pass from $\Omega_{2,n,\alpha}$ to $\Omega_{2,n,\alpha'}$, we delete the initial $\mathcal{T}_{2,n,\alpha}$-orbit of $\mathcal{D}$ and add $\mathcal{A}$ and its initial $\mathcal{T}_{2,n,\alpha}$-orbit. Since each step preserves $\mu$-measure, the sign of the change in mass is that of $(\overline{S}(k) - \underline{S}(k,v))$. See Lemma \ref{['c:massIncreasesWhere']}.
  • Figure 5: The domain $\Omega_{2,5,0}$ is the union of the gray and the hatched regions. These latter are excised to leave the planar domain of the two-dimensional system defined by the interval map given by by accelerating $T_{2,5,0}$ by applying appropriate powers of $W$ for $x$ in $(\ell_0, \epsilon_0) = (-t_{2,5}, \epsilon_0)$, where $\epsilon_0 = W^{-1}\cdot 0$. The negative sloped hatched rectangles form $\mathcal{D}_1$ of \ref{['e:deletionRects']}, the positive sloped hatched rectangles form $\mathcal{D}_2$.
  • ...and 2 more figures

Theorems & Definitions (35)

  • Theorem 1
  • Theorem 2
  • Definition 2.1
  • Theorem 2.2
  • Theorem 2.3: Nakada 2010
  • Definition 2.4
  • Definition 2.5
  • Definition 2.6
  • Definition 2.7
  • Theorem 3
  • ...and 25 more