Farey graphs and geodesic expansions of complex continued fractions

Hitoshi Nakada, Rie Natsui, and Jörg Thuswaldner · 2026

Read Farey graphs and geodesic expansions of complex continued fractions by Hitoshi Nakada, Rie Natsui, and Jörg Thuswaldner (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-28468.txt.

Introduction

Nakada, Natsui, and Thuswaldner build Farey graphs for complex continued fractions over five number fields built from square roots of negative integers. They work with Schmidt's Farey polygons lifted into three-dimensional hyperbolic space. Those shapes recover known tilings of the plane by Hecke groups. They also give polyhedra that tile the same space under Bianchi groups. The graphs then support geodesic continued fraction expansions, through a method broader than past work.

Catalog record

CREATORHitoshi Nakada, Rie Natsui, and Jörg Thuswaldner
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE87989
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2603.28468

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Farey graphs and geodesic expansions of complex continued fractions Hitoshi Nakada, Rie Natsui, and Jörg Thuswaldner Abstract We discuss complex Farey graphs for the Euclidean imaginary quadratic number fields ℚ ⁡ ( − d ) \mathbb{Q}(\sqrt{-d}) , d ∈ { 1 , 2 , 3 , 7 , 11 } d\in\{1,2,3,7,11\} . We study hyperbolic versions of A. Schmidt’s Farey polygons living in 3 3 -dimensional hyperbolic space ℍ 3 \mathbb{H}^{3} . Using these Farey polygons we recover tessellations of the hyperbolic plane ℍ 2 \mathbb{H}^{2} that are defined by the action of the Hecke groups H 4 H_{4} and H 6 H_{6} and have been studied earlier by I. Short and M. Walker. Moreover, hyperbolic Farey polygons allow us to define polyhedra that induce Farey tessellations of ℍ 3 \mathbb{H}^{3} by the action of certain Bianchi groups. Using complex Farey graphs we consider geodesic complex continued fraction expansions. Our method provides a different and more general approach as the one from the discussion by M. Hockman. Dedicated to Vitaly Bergelson on the occasion of his 75th birthday Introduction The main purpose of this paper is to give a brief explanation of some relations between complex versions of the Farey graph, the Farey tessellation, and complex continued fractions for Euclidean imaginary quadratic number fields. In the case of the Gaussian field, Hockman H-1 ; H-2 already discussed this topic. We consider all Euclidean imaginary quadratic number fields by a different approach. The notion of Farey sequence, Farey graph, and Farey tessellation, as well as their connection with continued fraction expansions is well-studied in the field of the rational numbers \mathbb{Q} , see e.g. , B-H-S ; J-S-W ; M . We give a brief sketch of this classical situation. Usually, one starts with 0=\tfrac{0}{1} and 1=\tfrac{1}{1} , which constitute the Farey sequence of level 1 . Then \tfrac{0}{1},\,\,\tfrac{1}{2}=\frac{0+1}{1+1},\,\,\tfrac{1}{1} is of level 2 , and \tfrac{0}{1},\,\,\tfrac{1}{3}=\tfrac{0+1}{1+2},\,\,\tfrac{1}{2},\,\,\tfrac{2}{3}=\tfrac{1+1}{2+1},\,\,\tfrac{1}{1} is of level 3 . We define the Farey sequence of level k recursively from that of level k-1 by inserting the mediant fraction \tfrac{a+c}{b+d} between consecutive rational numbers \tfrac{a}{b} and \tfrac{c}{d} in the sequence of level k-1 . Every rational number in [0,1] appears in the Farey sequence of level k for some k\geq 1 , and each appearing fraction \tfrac{a}{b} satisfies \gcd(a,b)=1 . It is easy to extend Farey sequences to all rational numbers by taking \mathbb{Z} as the Farey sequence of level 1 and iterating in the same way. One can check inductively that for fractions \tfrac{a}{b} and \tfrac{c}{d} that are “Farey neighbors” in some level k we always have |bc-ad|=1 . We now discuss the idea of Farey neighbors from a different point of view. Setting i=\sqrt{-1} let \mathbb{H}^{2}=\{z=x+yi:x,y\in\mathbb{R}\,\,\mbox{and}\,\,y>0\} be the hyperbolic plane with the Poincaré metric ds^{2}=\tfrac{dx^{2}+dy^{2}}{y} . The boundary of \mathbb{H}^{2} , denoted by \partial\mathbb{H}^{2} , is \mathbb{R}\cup\{\infty\} . A geodesic curve S is either a half-circle or a straight half-line, perpendicular to the \partial\mathbb{H}^{2} , i.e. , \begin{split}S&=\big\{x+yi\in\mathbb{C}\,:\,y>0\text{ and }|x+yi-a|^{2}=r^{2}\big\}\;\text{for some $a\in\mathbb{R}$ and $r>0$, or}\\ S&=\{x+yi\in\mathbb{C}\,:\,y>0\}\;\text{for some $x\in\mathbb{R}$}.\end{split} We denote S by (a-r,a+r) in the first case and (\infty,x) in the second case. We interpret {\rm PSL}(2,\mathbb{Z})=\left\{\begin{pmatrix}a&b\\ c&d\end{pmatrix}:a,b,c,d\in\mathbb{Z},ad-bc=1\right\}/\left\{\begin{pmatrix}\pm 1&0\\ 0&\pm 1\end{pmatrix}\right\} as a set of Möbius transformations defined by A(z)=\tfrac{az+b}{cz+d} for A=\begin{pmatrix}a&b\\ c&d\end{pmatrix} (note that A and -A define the same Möbius transformation and and we may choose either of these representatives to define the fractional transformation). The Farey graph is the graph with vertices in \mathbb{Q}^{*}=\mathbb{Q}\cup\{\infty\} and set of edges \mathcal{E}=\{A(\infty)\to A(0):A\in{\rm PSL}(2,\mathbb{Z})\} . Because \det A=1 one can show that \frac{a}{b},\frac{c}{d}\in\mathbb{Q} are connected by an edge if and only if |bc-ad|=1 . Thus \frac{a}{b} and \frac{c}{d} are connected by an edge in \mathcal{E} if and only if \frac{a}{b} and \frac{c}{d} are Farey neighbors in a Farey sequence of some level k . In this paper, we regard the Farey graph just as the set of edges \mathcal{E} and often (if the direction of the edge plays no role) interpret \alpha\to\beta\in\mathcal{E} as the geodesic (\alpha,\beta) in \mathbb{H}^{2} . Since the edges \infty\to 0 and 0\to\infty correspond to the identity matrix and to \begin{pmatrix}0&-1\\ 1&0\end{pmatrix} , respectively, z_{1}\to z_{2}\in\mathcal{E} is equivalent to z_{2}\to z_{1}\in\mathcal{E} . 1 1 1 Although this means that \mathcal{E} is an undiected graph it is convenient for us to use directed edges. Consider the hyperbolic triangle \mathcal{C}=\left\{z=x+yi:0<x<1,\,\,y>0,\,\,(x-\tfrac{1}{2})^{2}+y^{2}>\tfrac{1}{2}\right\} . It is well known that \bigcup_{A\in{\rm PSL}(2,\mathbb{Z})}A\cdot{\mathcal{C}}^{\mathrm{cl}}=\bigcup_{A\in{\rm PSL}(2,\mathbb{Z})}\{A(z):z\in{\mathcal{C}}^{\mathrm{cl}}\}=\mathbb{H}^{2}, where {B}^{\mathrm{cl}} is the closure of a given set B . Also we see that \mathcal{C}\cap A\mathcal{C} is either equal to \mathcal{C} or to the empty set. In other words, \mathcal{C} tessellates \mathbb{H}^{2} by the action of the group {\rm PSL}(2,\mathbb{Z}) . Because the edges of \mathcal{E} are equal to the edges of the hyperbolic triangles A\cdot C , A\in{\rm PSL}(2,\mathbb{Z}) , this shows that for any A_{1},\,A_{2}\in{\rm PSL}(2,\mathbb{Z}) the intersection (A_{1}(\infty),A_{1}(0))\cap(A_{2}(\infty),A_{2}(0)) is either equal to (A_{1}(\infty),A_{1}(0)) or empty, 2 2 2 Note that \partial\mathbb{H}^{2} is not part of the geodesic curves (\alpha,\beta) forming the edges \alpha\to\beta of \mathcal{E} . , thus the Farey graph is a planar graph. An illustration of a part of the Farey graph, which also shows Figure 1: The classical Farey tessellation of \mathbb{H}^{2} . the Farey tessellation \mathcal{T}=\{A\cdot C\,:\,A\in{\rm PSL}(2,\mathbb{Z})\} of \mathbb{H}^{2} into hyperbolic triangles, is provided in Figure 1 . It is well known that this tessellation is strongly related to the tessellaton of \mathbb{H}^{2} by the orbit of a fundamental domain of the modular group (see e.g. Series Series ). For a rational number r=\tfrac{p}{q}\in\mathbb{Q} with \gcd(p,q)=1 , the Ford circle \mathcal{F}(r) associated with r is the Euclidean circle defined by \mathcal{F}(r)=\Bigg\{x+yi\in\mathbb{C}:\,\,y\geq 0,\,\,\bigg|x+yi-\bigg(r+\frac{i}{2q^{2}}\bigg)\bigg|=\frac{1}{2q^{2}}\Bigg\}, Figure 2: Ford circles (black) and their interplay with the Farey graph (shaded). and \mathcal{F}(\infty)=\{z=x+i:x\in\mathbb{R}\}\cup\{\infty\} . One can show that the set of Ford circles is equal to \{A\cdot\mathcal{F}(\infty):A\in{\rm PSL}(2,\mathbb{Z})\} . The (degenerate) Ford circle \mathcal{F}(\infty)=\mathcal{F}\big(\frac{1}{0}\big) intersects \mathcal{F}\big(\frac{0}{1}\big) in i , which is contained in the geodesic \big(\frac{1}{0},\frac{0}{1}\big) . Because for each edge \tfrac{a}{b}\to\tfrac{c}{d}\in\mathcal{E} there is A=\begin{pmatrix}a&c\\ b&d\end{pmatrix}\in{\rm PSL}(2,\mathbb{Z}) such that \tfrac{a}{b}=A(\infty) and \tfrac{c}{d}=A(0) , we see that \tfrac{a}{b}\to\tfrac{c}{d}\in\mathcal{E} implies that A\cdot\mathcal{F}\big(\frac{1}{0}\big)=\mathcal{F}\big(\tfrac{a}{b}\big) and A\cdot\mathcal{F}\big(\frac{0}{1}\big)=\mathcal{F}\big(\tfrac{c}{d}\big) are tangent and intersect in a point contained in the geodesic \big(\tfrac{a}{b},\tfrac{c}{d}\big) . Moreover, if \tfrac{a}{b}\to\tfrac{c}{d}\not\in\mathcal{E} , then one can check by simple calculation that \mathcal{F}\big(\tfrac{a}{b}\big) and \mathcal{F}\big(\tfrac{c}{d}\big) are disjoint. See Figure 2 for an illustration of this interplay between Ford circles and the Farey graph. It is well known that the Farey graph, the Farey tessellation, and Ford circles have tight relations to continued fraction algorithms. In particular, one can prove that a continued fraction expansion [a_{0};a_{1},\ldots,a_{n}]=\displaystyle a_{0}+\frac{1}{a_{1}+\displaystyle\frac{1}{\ddots+\displaystyle\frac{1}{a_{n}}}} (1) with a_{0},a_{1},\ldots,a_{n}\in\mathbb{Z} corresponds to a walk 3 3 3 Recall that a walk in a graph is a consecutive sequence of edges. A path is a walk in which no vertex occurs more than one time. The shortest walk from a vertex a to a vertex b (which has to be a path) is called a geodesic path . Keep in mind the ambigous use of the word geodesic throughout the paper. of length n+1 in the Farey graph leading from \infty to x=[a_{0};a_{1},\ldots,a_{n}] . This is explored in Beardon et al. B-H-S . It is natural to ask which continued fraction expansion is the shortest in the sense that it corresponds to the shortest walk from \infty to x in the Faray

Read the complete text