Papers
Topics
Authors
Recent
Search
2000 character limit reached

OCF-Based Alternating Algorithm in Octagonal Dynamics

Updated 8 July 2026
  • The paper shows that the octagon additive continued fraction algorithm accelerates diagonal changes by grouping multiple elementary geometric moves into a single macro-step.
  • It unifies symbolic dynamics from the octagon Farey map with explicit geometric updates on quadrangulations, linking continued fractions to translation surface renormalization.
  • It highlights an alternating structure that interleaves staircase moves, symmetry operations, and relabeling to efficiently update directions on the octagon.

An OCF-based alternating algorithm is the renormalization procedure on the regular octagon in which the octagonal continued fraction of a direction determines a sequence of accelerated geometric updates on quadrangulations of the associated translation surface. In the formulation analyzed in "Octagonal continued fraction and diagonal changes" (Artigiani, 2020), the relevant continued-fraction mechanism is the octagon Farey map of Smillie–Ulcigrai, and the underlying fine-scale geometric dynamics are the diagonal changes introduced by Delecroix and Ulcigrai. The central result is that the octagon additive continued fraction algorithm is not a distinct mechanism from diagonal changes: it is their acceleration, so that one continued-fraction step groups finitely many elementary geometric moves into a single macro-step (Artigiani, 2020).

1. Geometric and dynamical setting

The ambient object is the translation surface XOX_O obtained by gluing opposite sides of a regular octagon OO. Its Veech group contains the dihedral symmetry group D8D_8 and the involution

γ=(12(1+2) 01),γ2=id.\gamma=\begin{pmatrix} -1 & 2(1+\sqrt2)\ 0 & 1 \end{pmatrix}, \qquad \gamma^2=\mathrm{id}.

The space of directions is Σ+\Sigma_+, the upper half of the circle, parametrized either by angle θ[0,π]\theta\in[0,\pi] or by inverse slope

u=cot(θ).u=\cot(\theta).

The upper half-circle is partitioned into eight sectors

Σj={θS1:jπ8θ(j+1)π8},j=0,,7.\overline{\Sigma}_j=\left\{\theta\in S^1:\frac{j\pi}{8}\le \theta\le \frac{(j+1)\pi}{8}\right\},\qquad j=0,\dots,7.

Sector Σ0\overline{\Sigma}_0 is a fundamental domain for the action of D8D_8 on OO0. For each sector OO1, a symmetry OO2 maps OO3 linearly onto OO4, giving the folding map

OO5

This setting matters because the algorithm acts simultaneously on two levels. On the symbolic side, it advances a continued-fraction coding of directions. On the geometric side, it updates quadrangulations and saddle connections on the octagon surface. A plausible implication is that the algorithm should be understood less as a purely number-theoretic expansion than as a renormalization rule for directions on a translation surface (Artigiani, 2020).

2. Octagonal continued fractions and the Farey map

The octagon Farey map is defined by

OO6

Equivalently, on each sector OO7, it is the branch

OO8

In inverse slope coordinate OO9, each branch is a Möbius transformation

D8D_80

where

D8D_81

The map is continuous and piecewise projective or linear in angle coordinates, and it is expanding except at sector endpoints (Artigiani, 2020).

The itinerary of a direction under D8D_82 is its octagon continued-fraction expansion. If

D8D_83

then

D8D_84

The map shifts this expansion: D8D_85

A direction can have at most two expansions, with the usual endpoint ambiguity. The explicit identifications are:

  • if D8D_86 is even, then

D8D_87

  • if D8D_88 is odd, then

D8D_89

Also,

γ=(12(1+2) 01),γ2=id.\gamma=\begin{pmatrix} -1 & 2(1+\sqrt2)\ 0 & 1 \end{pmatrix}, \qquad \gamma^2=\mathrm{id}.0

In the OCF-based alternating algorithm, the symbols γ=(12(1+2) 01),γ2=id.\gamma=\begin{pmatrix} -1 & 2(1+\sqrt2)\ 0 & 1 \end{pmatrix}, \qquad \gamma^2=\mathrm{id}.1 are the control data. Each digit selects a branch γ=(12(1+2) 01),γ2=id.\gamma=\begin{pmatrix} -1 & 2(1+\sqrt2)\ 0 & 1 \end{pmatrix}, \qquad \gamma^2=\mathrm{id}.2, and therefore selects a prescribed finite block of geometric operations. This is the precise sense in which the procedure is OCF-based (Artigiani, 2020).

3. Diagonal changes and quadrangulations

The fine-scale geometric mechanism is the diagonal changes algorithm on translation surfaces, specialized here to the regular octagon. Its elementary objects are wedges, admissible quadrilaterals, staircases, and staircase moves.

A wedge on a translation surface γ=(12(1+2) 01),γ2=id.\gamma=\begin{pmatrix} -1 & 2(1+\sqrt2)\ 0 & 1 \end{pmatrix}, \qquad \gamma^2=\mathrm{id}.3 is a pair of saddle connections

γ=(12(1+2) 01),γ2=id.\gamma=\begin{pmatrix} -1 & 2(1+\sqrt2)\ 0 & 1 \end{pmatrix}, \qquad \gamma^2=\mathrm{id}.4

such that γ=(12(1+2) 01),γ2=id.\gamma=\begin{pmatrix} -1 & 2(1+\sqrt2)\ 0 & 1 \end{pmatrix}, \qquad \gamma^2=\mathrm{id}.5 and γ=(12(1+2) 01),γ2=id.\gamma=\begin{pmatrix} -1 & 2(1+\sqrt2)\ 0 & 1 \end{pmatrix}, \qquad \gamma^2=\mathrm{id}.6 start at the same singularity, γ=(12(1+2) 01),γ2=id.\gamma=\begin{pmatrix} -1 & 2(1+\sqrt2)\ 0 & 1 \end{pmatrix}, \qquad \gamma^2=\mathrm{id}.7 is left-slanted, γ=(12(1+2) 01),γ2=id.\gamma=\begin{pmatrix} -1 & 2(1+\sqrt2)\ 0 & 1 \end{pmatrix}, \qquad \gamma^2=\mathrm{id}.8, γ=(12(1+2) 01),γ2=id.\gamma=\begin{pmatrix} -1 & 2(1+\sqrt2)\ 0 & 1 \end{pmatrix}, \qquad \gamma^2=\mathrm{id}.9 is right-slanted, Σ+\Sigma_+0, and the two edges are sides of an embedded triangle in Σ+\Sigma_+1 (Artigiani, 2020).

An admissible quadrilateral is one whose left- and right-slanted edges alternate cyclically around it. A quadrangulation Σ+\Sigma_+2 is a decomposition of Σ+\Sigma_+3 into admissible quadrilaterals. Each quadrilateral has bottom sides forming a wedge Σ+\Sigma_+4, top sides, and a diagonal Σ+\Sigma_+5. It is called left-slanted if its diagonal is left-slanted and right-slanted if its diagonal is right-slanted.

The elementary geometric update is the diagonal change. If a quadrilateral Σ+\Sigma_+6 is left-slanted, the base wedge is replaced by

Σ+\Sigma_+7

If Σ+\Sigma_+8 is right-slanted, it is replaced by

Σ+\Sigma_+9

A left staircase is a cyclic chain of quadrilaterals glued so that top-left sides identify with next bottom-right sides; a right staircase is defined symmetrically by top-right with next bottom-left. If all quadrilaterals in the staircase are left-slanted or right-slanted, the staircase is well slanted. A staircase move is simultaneous diagonal change in all quadrilaterals of a well-slanted staircase (Artigiani, 2020).

The quadrangulation also has a combinatorial encoding. For a labeled quadrangulation θ[0,π]\theta\in[0,\pi]0 with θ[0,π]\theta\in[0,\pi]1 quadrilaterals, the datum is a pair of permutations

θ[0,π]\theta\in[0,\pi]2

where the top-left side of θ[0,π]\theta\in[0,\pi]3 glues to the bottom-right side of θ[0,π]\theta\in[0,\pi]4, and the top-right side of θ[0,π]\theta\in[0,\pi]5 glues to the bottom-left side of θ[0,π]\theta\in[0,\pi]6. The train-track relations are

θ[0,π]\theta\in[0,\pi]7

For a cycle θ[0,π]\theta\in[0,\pi]8 of θ[0,π]\theta\in[0,\pi]9 or u=cot(θ).u=\cot(\theta).0, a staircase move changes the combinatorics by explicit permutation updates and transforms the length data linearly via

u=cot(θ).u=\cot(\theta).1

These definitions are the non-accelerated substrate of the algorithm. The OCF-based alternating algorithm packages such staircase and diagonal changes into sector-dependent blocks rather than executing them one by one (Artigiani, 2020).

4. Acceleration and the alternating structure

The main theorem states:

The octagon additive continued fraction algorithm defined in Smillie–Ulcigrai is an acceleration of the diagonal changes algorithm for the octagon (Artigiani, 2020).

Equivalently, the octagon Farey map u=cot(θ).u=\cot(\theta).2 is obtained by composing a finite number of elementary diagonal changes until the next continued-fraction step is reached. Thus diagonal changes are the fine-scale procedure, while the octagon Farey map records their net effect after a complete block.

This is the precise content of acceleration in the paper. The diagonal changes algorithm produces a detailed path in the move graph; the octagon Farey map gives the induced transformation after traversing one whole block of that path. The relation is explicitly compared with the classical fact that the Gauss map is an acceleration of the one-dimensional Farey map (Artigiani, 2020).

The procedure is also described as alternating. The relevant alternation is not between two optimization blocks or two players, but among different kinds of geometric operations:

  • direct staircase changes,
  • symmetry operations,
  • relabelings,
  • and changes in the type of staircase, left or right.

This is especially visible in even sectors, where orientation-reversing symmetries must be inserted to match the branch structure of the octagon Farey map. The paper therefore interprets the accelerated procedure as an alternating algorithm because one effective step is a composition of different operation types rather than a repetition of a single uniform move (Artigiani, 2020).

A plausible implication is that “alternating” here names a structural property of the renormalization block: symbolic normalization, geometric diagonal change, and combinatorial relabeling are interleaved within each continued-fraction step.

5. Sector-wise decomposition into finite move words

The regular octagon has a special quadrangulation u=cot(θ).u=\cot(\theta).3. If a direction u=cot(θ).u=\cot(\theta).4 lies in sector u=cot(θ).u=\cot(\theta).5, the initial quadrangulation is

u=cot(θ).u=\cot(\theta).6

The branch u=cot(θ).u=\cot(\theta).7 opens the sector and the paper works with

u=cot(θ).u=\cot(\theta).8

This opened quadrangulation has combinatorial datum

u=cot(θ).u=\cot(\theta).9

Because even sectors involve orientation-reversing symmetries, two auxiliary operations are introduced:

  1. Symmetry, which interchanges left and right vectors in every quadrilateral.
  2. Relabeling, because after a cycle of moves the combinatorics may return only up to a permutation of labels.

The paper gives a reduced graph of allowed moves in Σj={θS1:jπ8θ(j+1)π8},j=0,,7.\overline{\Sigma}_j=\left\{\theta\in S^1:\frac{j\pi}{8}\le \theta\le \frac{(j+1)\pi}{8}\right\},\qquad j=0,\dots,7.0 and identifies basic move words such as

  • Σj={θS1:jπ8θ(j+1)π8},j=0,,7.\overline{\Sigma}_j=\left\{\theta\in S^1:\frac{j\pi}{8}\le \theta\le \frac{(j+1)\pi}{8}\right\},\qquad j=0,\dots,7.1,
  • Σj={θS1:jπ8θ(j+1)π8},j=0,,7.\overline{\Sigma}_j=\left\{\theta\in S^1:\frac{j\pi}{8}\le \theta\le \frac{(j+1)\pi}{8}\right\},\qquad j=0,\dots,7.2,
  • Σj={θS1:jπ8θ(j+1)π8},j=0,,7.\overline{\Sigma}_j=\left\{\theta\in S^1:\frac{j\pi}{8}\le \theta\le \frac{(j+1)\pi}{8}\right\},\qquad j=0,\dots,7.3,
  • symmetry.

For each sector Σj={θS1:jπ8θ(j+1)π8},j=0,,7.\overline{\Sigma}_j=\left\{\theta\in S^1:\frac{j\pi}{8}\le \theta\le \frac{(j+1)\pi}{8}\right\},\qquad j=0,\dots,7.4, a finite word in these elementary moves has total action equal to the branch Σj={θS1:jπ8θ(j+1)π8},j=0,,7.\overline{\Sigma}_j=\left\{\theta\in S^1:\frac{j\pi}{8}\le \theta\le \frac{(j+1)\pi}{8}\right\},\qquad j=0,\dots,7.5 of the octagon Farey map. The paper gives explicit examples (Artigiani, 2020):

Sector Word of moves Matrix information
Σj={θS1:jπ8θ(j+1)π8},j=0,,7.\overline{\Sigma}_j=\left\{\theta\in S^1:\frac{j\pi}{8}\le \theta\le \frac{(j+1)\pi}{8}\right\},\qquad j=0,\dots,7.6 Σj={θS1:jπ8θ(j+1)π8},j=0,,7.\overline{\Sigma}_j=\left\{\theta\in S^1:\frac{j\pi}{8}\le \theta\le \frac{(j+1)\pi}{8}\right\},\qquad j=0,\dots,7.7 matrix Σj={θS1:jπ8θ(j+1)π8},j=0,,7.\overline{\Sigma}_j=\left\{\theta\in S^1:\frac{j\pi}{8}\le \theta\le \frac{(j+1)\pi}{8}\right\},\qquad j=0,\dots,7.8 given explicitly
Σj={θS1:jπ8θ(j+1)π8},j=0,,7.\overline{\Sigma}_j=\left\{\theta\in S^1:\frac{j\pi}{8}\le \theta\le \frac{(j+1)\pi}{8}\right\},\qquad j=0,\dots,7.9 Σ0\overline{\Sigma}_00 matrix Σ0\overline{\Sigma}_01 explicitly given
Third sector Σ0\overline{\Sigma}_02 matrix Σ0\overline{\Sigma}_03

For the first sector, the matrix is

Σ0\overline{\Sigma}_04

The fourth, fifth, sixth, and seventh sectors are likewise described by explicit words and matrices Σ0\overline{\Sigma}_05, and the appendix contains drawings of the actual geometric staircase moves (Artigiani, 2020).

These sector words are the operational core of the algorithm. One continued-fraction digit chooses one sector, and the sector chooses one finite word of elementary moves. That is the exact mechanism by which OCF digits control alternating geometric renormalization.

6. Coding, execution, and interpretation

The key dynamical relation

Σ0\overline{\Sigma}_06

means that each digit Σ0\overline{\Sigma}_07 specifies the next normalization sector. Geometrically, the paper describes the process as follows (Artigiani, 2020):

  1. Σ0\overline{\Sigma}_08 determines the initial placement of the quadrangulation,

Σ0\overline{\Sigma}_09

  1. Applying D8D_80 opens the relevant sector and puts the surface into the standard position

D8D_81

  1. The next digit D8D_82 determines which sequence of diagonal changes to perform on D8D_83.
  2. Repetition produces a sequence of quadrangulations and saddle connections approximating the direction D8D_84.

The paper also gives a concrete implementation viewpoint. For an irrational direction D8D_85, one determines the sector D8D_86, initializes D8D_87, applies the corresponding sector word, updates quadrangulation and length data by the associated matrix product, computes the new direction D8D_88, and repeats (Artigiani, 2020).

This suggests that the OCF-based alternating algorithm is simultaneously:

  • a symbolic algorithm, because it shifts the expansion D8D_89;
  • a geometric algorithm, because it acts on wedges, quadrangulations, and saddle connections;
  • and a linear-algebraic algorithm, because the length data evolve by explicit matrices attached to move words.

The paper emphasizes that intermediate vectors, analogous to intermediate convergents in the classical Farey–Gauss setting, are naturally represented by the finer diagonal changes. A plausible implication is that the accelerated continued-fraction description suppresses this intermediate structure, while the non-accelerated diagonal-change model retains it (Artigiani, 2020).

7. Significance and relation to adjacent notions

The principal significance of the OCF-based alternating algorithm is conceptual: it identifies the octagonal continued fraction of Smillie–Ulcigrai with an accelerated geometric renormalization procedure rather than an isolated symbolic construction. In this perspective, octagonal continued-fraction digits are not merely labels of sectors; they are instructions for finite blocks of diagonal changes, possibly augmented by symmetry and relabeling (Artigiani, 2020).

This addresses a natural misconception. The octagon Farey map might appear to define a self-contained continued-fraction algorithm unrelated to the local geometry of quadrangulations. The paper shows the opposite: each branch OO00 is realized by a finite word in the diagonal-change calculus, so the continued fraction is a symbolic code for an underlying geometric process (Artigiani, 2020).

The term “OCF-based alternating algorithm” therefore refers most precisely to a procedure in which:

  • the OCF component is the octagonal continued fraction

OO01

  • the based component is the selection of sector branches OO02,
  • and the alternating algorithm component is the composition of diagonal changes, symmetry, relabeling, and left/right staircase-type transitions into accelerated blocks.

Within the scope of the paper, the bottom-line characterization is concise: the octagon Farey map

OO03

is an acceleration of the diagonal changes algorithm on the regular octagon, and the octagonal continued-fraction digits provide the symbolic coding of that geometric alternating renormalization process (Artigiani, 2020).

Definition Search Book Streamline Icon: https://streamlinehq.com
References (1)

Topic to Video (Beta)

No one has generated a video about this topic yet.

Whiteboard

No one has generated a whiteboard explanation for this topic yet.

Follow Topic

Get notified by email when new papers are published related to OCF-based Alternating Algorithm.