On the stability, complexity, and distribution of similarity classes of the longest edge bisection process for triangles
Daniel Kalmanovich, Yaar Solomon · 2026
Read On the stability, complexity, and distribution of similarity classes of the longest edge bisection process for triangles by Daniel Kalmanovich, Yaar Solomon (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-2601-13663.txt.
Introduction
Daniel Kalmanovich and Yaar Solomon study what happens when a triangle is cut again and again in 2026. Each cut joins the midpoint of the longest edge to the vertex across from it. Repeating the cut makes an endless family of new triangles. A result by Stynes from 1980 showed a surprise here.
For any start triangle, the endless family falls into just finitely many similarity classes. Shapes in one class are the same form at a new scale. The team studies the stability, complexity, and spread of these classes. Finite classes make the long-run pattern of cuts far more tractable.
Catalog record
| CREATOR | Daniel Kalmanovich, Yaar Solomon |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 48 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2601-13663
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
On the stability, complexity, and distribution of similarity classes of the longest edge bisection process for triangles Daniel Kalmanovich, Yaar Solomon Abstract . The Longest Edge Bisection of a triangle is performed by joining the midpoint of its longest edge to the opposite vertex. Applying this procedure iteratively produces an infinite family of triangles. Surprisingly, a classical result of Stynes (1980) shows that for any initial triangle, this infinite family falls into finitely many similarity classes. While the set of classes is finite, it turns out that a far smaller, stable subset of “fat” triangles effectively dominates the final mesh structure. We prove the following asymptotic area distribution result: for every initial triangle, the portion of area occupied by terminal quadruples tends to one, with the convergence occurring at an exponential rate. In fact, we provide the precise distribution of triangles in every step. We introduce the bisection graph and use spectral methods to establish this result. Given this dominance, we provide a complete characterization of triangles possessing a single terminal quadruple, while conversely exhibiting a sequence of triangles with an unbounded number of terminal quadruples. Furthermore, we reveal several fundamental geometric properties of the points of a terminal quadruple, laying the groundwork for studying the geometric distribution of the entire orbit. Our analysis leverages the hyperbolic geometry framework of Perdomo and Plaza (2014) and refines their techniques. 1. Introduction 1.1. Background The Longest Edge Bisection (LEB) process is a straightforward iterative procedure formally initiated by Rosenberg and Stenger [ 8 ] . In each step, the median to the longest edge is added, splitting the triangle into two daughter triangles. Rosenberg and Stenger established the critical property of non-degeneracy, which guarantees geometric quality. They proved that for an initial triangle \Delta_{0} with smallest angle \alpha_{0} , every angle generated by the infinite process is bounded from below by \frac{\alpha_{0}}{2} . Additionally, a useful triangulation process requires the convergence of the mesh diameter to zero. Kearfott [ 6 ] first proved this for LEB in arbitrary dimensions; later, Stynes [ 9 , 10 ] and Adler [ 1 ] provided sharpened bounds on the convergence rates. A landmark result, known as the Finite Similarity Classes (FSC) Theorem , was established by Stynes [ 10 , Corollary 1] . This theorem asserts that the infinite family of triangles generated by iteratively bisecting a single initial triangle falls into a finite number of similarity classes. This result was later reproved by Adler [ 1 ] using an elegant inductive argument, and subsequently by Perdomo and Plaza [ 7 ] , who recast the analysis within the framework of hyperbolic geometry, a setting that proved to be particularly natural. In their paper, Rosenberg and Stenger identified a specific region where triangles enter a periodic orbit of length 4 under the \mathrm{LEB} process [ 8 , Lemma 4] ; we refer to these orbits as terminal sets. Building on this work and the FSC Theorem, Stynes [ 10 , Corollary 2] demonstrated that the proportion of the area occupied by triangles in these terminal sets converges to 1 . These results pave the way for a deeper, quantitative analysis of the \mathrm{LEB} process as a dynamical system. While it is known that the variety of shapes is finite and that terminal sets asymptotically exhaust the area of the initial triangle, this does not describe the rate of convergence or the precise distribution of triangle shapes. Several fundamental questions naturally arise: At what rate does the total area of the mesh become dominated by these stable shapes? What are the specific geometric properties of these terminal sets? Furthermore, how does the number of distinct terminal sets depend on the geometry of the initial triangle? Answering these questions is essential for a comprehensive understanding of the stability and complexity of the \mathrm{LEB} process. We provide a precise formulation of these problems in § 1.2 and address them as the main contribution of this paper. Partial results in these directions were previously noted in [ 4 ] . 1.2. Preliminaries and main results The 2014 paper [ 7 ] by Perdomo and Plaza provided a new proof of the FSC theorem, employing the machinery of hyperbolic dynamics in a moduli space of triangle shapes, modeled in the hyperbolic upper half-plane \mathbb{H} , to analyze the bisection process as a dynamical system. Given a triangle \Delta_{0} , by (1) scaling the edge lengths so that the longest edge has length 1 , (2) mapping the longest edge to the interval [0,1] on the real axis, and (3) placing the third vertex z opposite to the longest edge, on the upper half plane, and so that the shortest edge is on the left, we bijectively identify every similarity class of triangles with a unique point z in the domain D\subseteq\mathbb{H} defined by (see Figure 1 ). D=\left\{z\in\mathbb{C}\ \middle|\ \mathrm{Re}(z)\leq\frac{1}{2},\>\mathrm{Im}(z)>0,\>\left|{z-1}\right|\leq 1\right\}. Relying on this identification, we refer to a point z\in D as a triangle. For such z\in D , the bisection by the median to the longest edge defines two function L,R:D\to D , where L(z) is the left triangle and R(z) is the right triangle . As explained in [ 7 ] , the region D is subdivided into 6 subregions by the geodesics \mathrm{Re}(z)=\frac{1}{4} , \left|{z}\right|=\frac{1}{2} and \left|{z-\frac{1}{2}}\right|=\frac{1}{2} (see Figure 1 ), and the functions L and R are defined in a piecewise manner, by a Möbius or anti-Möbius transformation on each one of these regions. The formulas and the geometric interpretation of these transformations are presented in Table 1 . An illustrative application presenting the geometry of the various transformations is available at https://danielkalmanovich.wixsite.com/website/leb . The reader is referred to [ 2 , §3] for more details on the action by Möbius transformations on the upper half-plane. \mathrm{Re} \mathrm{Im} 0.5 1 0.5i 1i 0 (a) The normalized region D I II III IV V VI \mathrm{Re}(z)=\frac{1}{4} \left|{z}\right|=\frac{1}{2} \left|{z-\frac{1}{2}}\right|=\frac{1}{2} (b) The subdivision of D into 6 subregions Figure 1 . The normalized region D and its subdivision. Region Function Geometric interpretation in D I L(z)=\frac{1}{2\overline{z}} An inversion with respect to the circle |z|=\frac{\sqrt{2}}{2} . II L(z)=\frac{-1}{2z-1} An elliptic Möbius map with a fixed point at \frac{1}{4}+\frac{\sqrt{(}7)}{4}i\in D . Points rotate counterclockwise by an angle of \arccos\left(-\frac{3}{4}\right) about the fixed point. III L(z)=\frac{2\bar{z}}{2\bar{z}-1} A hyperbolic anti-Möbius map. Points spiral from the repeller at z=0 to the attractor at z=\frac{3}{2} . IV L(z)=\frac{2z-1}{2z} An elliptic Möbius map with a fixed point at \frac{1}{2}+\frac{1}{2}i\in D . Points rotate clockwise by an angle of \frac{\pi}{2} about the fixed point. V L(z)=2z A hyperbolic Möbius map that dilates by 2 . VI L(z)=1-2\bar{z} A hyperbolic anti-Möbius map. A reflection across \mathrm{Re}(z)=\frac{1}{3} combined with a dilation by a factor of 2 . \text{I}\cup\text{II}\cup\text{III} R(z)=\frac{-1}{2z-2} An elliptic Möbius map with a fixed point at \frac{1}{2}+\frac{1}{2}i\in D . Points rotate counterclockwise by an angle of \frac{\pi}{2} about the fixed point. \text{IV}\cup\text{V}\cup\text{VI} R(z)=\frac{2\bar{z}-1}{2\bar{z}-2} An inversion with respect to the circle |z-1|=\frac{\sqrt{2}}{2} . Table 1. The piecewise functions L(z) and R(z) . As we will show, an exceptional property of the dynamics of the \mathrm{LEB} process is that, regardless of the initial triangle, the process eventually produces predominantly triangles from a certain subregion of \text{I}\cup\text{IV} , with