A Recursive Block–Pillar Structure in the Kolakoski Sequence K ⁡ ( 1 , 3 ) K(1,3)

William Cook · 2025

Read A Recursive Block–Pillar Structure in the Kolakoski Sequence K ⁡ ( 1 , 3 ) K(1,3) by William Cook (2025) 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-2504-13433.txt.

Introduction

William Cook reveals in 2025 a nested block structure inside a classic number sequence built from ones and threes. That sequence, named for Kolakoski, is known to be ordered, unlike its two symbol cousin. Its symbol share links to a special cubic number.

He defines block and pillar sequences by a mutual recursion. Blocks are shown to be prefixes of the full sequence that grow toward it. Length and symbol count rules follow the same growth number in the limit. If block and pillar densities settle, they must match the known symbol share. The frame gives a fresh, constructive view of the sequence's order.

Catalog record

CREATORWilliam Cook
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE35 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2504-13433

Browse the full library · About the Tower of Babel

Full text

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

A Recursive Block–Pillar Structure in the Kolakoski Sequence K ⁡ ( 1 , 3 ) K(1,3) William Cook Abstract The Kolakoski sequence K ⁡ ( a , b ) K(a,b) over { a , b } \{a,b\} is the unique sequence starting with a a that equals its own run-length encoding. While the classical case K ⁡ ( 1 , 2 ) K(1,2) ( A000002 [ 5 ] ) remains deeply enigmatic [ 2 ] , generalisations K ⁡ ( a , b ) K(a,b) exhibit markedly different behaviours depending on the parity of a a and b b . The sequence K ⁡ ( 1 , 3 ) K(1,3) ( A064353 [ 6 ] ), a ’same-parity’ case using the alphabet { 1 , 3 } \{1,3\} , is known to possess significant structure; notably, a related sequence is morphic [ 1 ] , leading to a calculable symbol density distinct from 1 / 2 1/2 . The frequency of ’1’ in K ⁡ ( 1 , 3 ) K(1,3) is known to be d ≈ 0.397215 d\approx 0.397215 [ 1 , p. 171, Eq. (6)] . This paper reveals a complementary structural property: K ⁡ ( 1 , 3 ) K(1,3) admits an explicit nested block–pillar recursion. We introduce block sequences B n B_{n} and pillar sequences P n P_{n} satisfying B n + 1 = B n ⋅ P n ⋅ B n , P n + 1 = 𝒢 ⁡ ( R ⁡ ( P n ) , 3 ) , B_{n+1}=B_{n}\!\cdot\!P_{n}\!\cdot\!B_{n},\quad P_{n+1}=\mathcal{G}(R(P_{n}),3), where 𝒢 \mathcal{G} is the run-length generation operator and R ⁡ ( P n ) R(P_{n}) denotes the sequence P n P_{n} interpreted as a run-length vector. We prove that every B n B_{n} is a prefix of K ⁡ ( 1 , 3 ) K(1,3) and that B n + 1 = 𝒢 ⁡ ( R ⁡ ( B n ) , 1 ) B_{n+1}=\mathcal{G}(R(B_{n}),1) , demonstrating how the recursion reflects the Kolakoski property. This provides a direct, constructive recursive definition of K ⁡ ( 1 , 3 ) K(1,3) . Furthermore, we derive exact recurrences for prefix lengths and symbol counts, prove exponential growth governed by the Pisot number α \alpha (the real root of x 3 − 2 ​ x 2 − 1 = 0 x^{3}-2x^{2}-1=0 [ 1 , p. 170] ), and establish a recurrence for the symbol density. We show that if the block and pillar densities converge, they must converge to the same value d d . The quantitative analysis aligns perfectly with the known properties of K ⁡ ( 1 , 3 ) K(1,3) derived from substitution dynamics [ 1 ] . This highlights the structural regularity expected of same-parity Kolakoski sequences and offers an alternative, constructive perspective to its known morphic generation. 1 Introduction For distinct symbols a,b\in\mathbb{N} , the Kolakoski sequence K(a,b) is the unique infinite sequence over the alphabet \{a,b\} beginning with a such that the sequence equals its own run-length encoding [ 4 , 7 ] . That is, if R(S) denotes the sequence of run lengths of a sequence S , then K(a,b) equals the sequence generated by applying the run lengths R(S) while alternating between symbols a and b , starting with a . The archetypal case K(1,2) ( A000002 [ 5 ] ), K(1,2)=1\,2\,2\,1\,1\,2\,1\,2\,2\,1\,2\,2\,\dots, has resisted decades of analysis despite its simple definition [ 2 ] . Fundamental questions, such as whether the asymptotic frequency of 1 s exists and equals 1/2 [ 3 ] , remain open. However, the broader family of K(a,b) sequences exhibits a crucial dichotomy based on the parity of a and b . Sequences where a and b have different parity (like K(1,2) ) tend to be complex and poorly understood. In contrast, sequences where a and b share the same parity (both odd or both even) are often more structured and analytically tractable [ 1 ] . This article focuses on K(1,3) ( A064353 [ 6 ] ), a key example of the ’same-parity’ (odd-odd) case: K(1,3)=1\,3\,3\,3\,1\,1\,1\,3\,3\,3\,1\,3\,\dots. Reflecting the expected regularity of same-parity sequences, K(1,3) is known to possess significant structure. It is closely related to the sequence \operatorname{Kol}(3,1) (its mirror image starting with 3), which is known to be generatable by a primitive Pisot substitution rule and classifiable as a (deformed) model set with pure point diffraction spectrum [ 1 ] . These properties imply unique symbol frequencies exist. The frequency of ’1’ in K(1,3) is known to be d=(3-\alpha)/2\approx 0.397215 , where \alpha\approx 2.20557 is the real root of the polynomial x^{3}-2x^{2}-1=0 [ 1 , p. 171, Eq. (6)] . This non-trivial frequency confirms K(1,3) is structurally different from a random sequence. (Note that sequences like K(2,4) are trivial rescalings of K(1,2) and thus inherit its complexity, e.g., [ 2 , p. 1] ). In this paper, we reveal a different, yet comparably structured, property of K(1,3) . We demonstrate that K(1,3) admits an explicit block–pillar recursion, allowing arbitrarily long prefixes to be generated via a direct, nested construction. We define sequences (blocks B_{n} and pillars P_{n} ) and show they satisfy a simple mutual recursion that is consistent with the Kolakoski property itself, applied alternately to the blocks and pillars using a generation operator \mathcal{G} . This recursive structure not only provides a generative mechanism but also allows for a detailed quantitative analysis, including deriving relations for the asymptotic symbol frequency and growth rate. This provides a constructive, recursive description of K(1,3) itself, complementing the known morphic generation and further highlighting the profound structural differences between K(1,3) and the classical K(1,2) . The structure of the paper is as follows: Section 2 introduces the necessary notation and the recursive definitions. Section 3 establishes basic properties of the blocks and pillars. Section 4 presents the main theorem proving the connection to K(1,3) and the Kolakoski property. Section 5 performs a detailed quantitative analysis of lengths, counts, density, and growth rate. Section 6 discusses the implications of these findings, connects them to known results, and suggests future directions. 2 Definitions Throughout, concatenation of finite sequences is written \!\cdot\! and indexing starts at 1 . The alphabet considered for K(1,3) is \Sigma=\{1,3\} . We denote the number of occurrences of symbol x in a sequence W by N_{x}(W) . We use angle brackets \langle a_{1},\dots,a_{n}\rangle for sequences viewed as vectors, particularly run-length vectors, and standard sequence notation S[1..n] for prefixes. We use R(W) to denote the sequence W interpreted as a run-length vector. Definition 2.1 (Generation operator) . For a finite run‑length vector R=\langle r_{1},\dots,r_{m}\rangle of positive integers and a starting symbol s\in\{1,3\} , define the generation operator \mathcal{G} as: \mathcal{G}(R,s)=s^{r_{1}}\!\cdot\!(4{-}s)^{r_{2}}\!\cdot\!s^{r_{3}}\!\cdot\!\cdots, where x^{k} denotes k consecutive copies of the symbol x , and (4-s) provides the alternate symbol within the alphabet \{1,3\} . The parity of the run index determines the symbol used, starting with s for the first run. The length of the generated sequence is \lvert\mathcal{G}(R,s)\rvert=\sum_{i=1}^{m}r_{i} . Definition 2.2 (Blocks and pillars) . Let R_{0}=R(K(1,3)[1..5])=\langle 1,3,3,3,1\rangle be the run-length vector corresponding to the first 5 terms of K(1,3) ( A064353 [ 6 ] ). Define the initial block sequence B_{1}=\mathcal{G}(R_{0},1) and the initial pillar sequence P_{1}=\langle 3\rangle : \displaystyle B_{1} \displaystyle=\mathcal{G}\!\bigl(\langle

Read the complete text