Elsevier

Theoretical Computer Science

Volume 936, 10 November 2022, Pages 172-183
Theoretical Computer Science

On the 2-abelian complexity of generalized Cantor sequences

https://doi.org/10.1016/j.tcs.2022.09.025Get rights and content

Highlights

  • We prove that abelian complexity of 2-block coding of generalized Cantor sequences satisfying certain conditions is regular.

  • We build a bridge between 2-abelian complexity of generalized Cantor sequences and abelian complexity of its 2-block coding.

  • We prove that 2-abelian complexity of generalized Cantor sequences satisfying certain conditions is regular.

Abstract

In this paper, we study the generalized Cantor sequence c, which is an -automatic sequence. We prove that the abelian complexity of the 2-block sequence of c is -regular if the factor set of the sequence c is mirror invariant. As a consequence, we show that the 2-abelian complexity of a generalized Cantor sequence satisfying certain conditions is -regular.

Introduction

In recent years, abelian complexity of infinite sequences has been a widely studied topic. Early in the seventies of the last century, Coven and Hedlund [5] showed that the periodic and Sturmian words both can be characterized by using Parikh vectors. In 2011, Richomme, Saari and Zamboni [13] formally introduced abelian complexity through Parikh vectors. In the same paper, they characterized Thue-Morse minimal subshift by its factor complexity and abelian complexity simultaneously. Thereafter, the abelian complexity of some notable words, such as the regular paperfolding word and the Rudin-Shapiro word, were studied in [10] and [9] respectively. Madill and Rampersad [10] raised a question that whether the abelian complexity of any b-automatic sequence is always b-regular. In 2016, Blanchet-Sadri, Seita and Wise [3] proved that the abelian complexity of binary word w is -regular if w is a fixed point of arbitrary -uniform morphism φ satisfying that ||φ(0)|0|φ(1)|0|1.

The k-abelian complexity, as a generalization of abelian complexity, was firstly introduced in 1981 by Karhumäki [7]. The formal definition of the k-abelian complexity was given in 2013 by Karhumäki, Saarela and Zamboni [8]. Meanwhile, they computed the k-abelian complexity of all Sturmian words for every k1 and characterized ultimately periodic words by the k-abelian complexity. In 2014, Vandomme, Parreau and Rigo [12] conjectured that the 2-abelian complexity of the Thue-Morse word is 2-regular. An affirmative answer has been given independently by Greinecker [6] and by Parreau, Rigo, Rowland and Vandomme [11]. It is notable that in [11] they also proved that the 2-abelian complexity of the period-doubling word is 2-regular and raised the following general conjecture.

Conjecture 1

The 2-abelian complexity of any b-automatic sequence is b-regular.

For the general automatic sequence, its 2-abelian complexity is hard to investigate. In 2018, Chen, Lü and Wu [4] studied the k-abelian complexity of the Cantor sequence and proved that the k-abelian complexity of the Cantor sequence is always 3-regular for every k1. Here we will consider a class of automatic sequences which are generalizations of the Cantor sequence. First we give the definition. Let k2 be an integer and {ni}1ik1 be (k1) positive integers. Set =k+i=1k1ni. The generalized Cantor sequence c is defined as the fixed point of the morphism σ:110n110n2110nk11,00 beginning with 1, i.e.,c:=σ(1)=c(0)c(1)=10n110n2110nk110. Note that c is the Cantor sequence if k=2 and n1=1.

For any infinite sequence w, we say the factor set of w is mirror-invariant if the reversal of u is also a factor of w whenever a finite word u is a factor of w. It is not hard to obtain that the factor set of the generalized Cantor sequence c is mirror-invariant if and only if σ(1) is a palindrome, i.e., σ(1)=σ(1)R where uR denotes the reversal of u. By the definition of the morphism σ, |σ(0)|0|σ(1)|0=k2. Following from [3] or Lemma 4 in Section 3, the abelian complexity of any generalized Cantor sequence c is -regular. In this paper, we focus on the 2-abelian complexity of c whose factor set is mirror-invariant. The first theorem is about the abelian complexity of the corresponding 2-block coding word x:=block(c,2) (see Definition 4).

Theorem 1

If the factor set of the generalized Cantor sequence c is mirror-invariant, then the abelian complexity of the 2-block coding sequence x of c is ℓ-regular.

Before stating the second theorem, we need some notations. LetA:={i|0i1,c(i)=1}={0}{m+1+j=1mnj|1mk1}. Let L1:={+ji+1|i,jAandi>j} and L2:={+ji+1|i,jAandij}. For any two sets A,BZ and any two numbers x,yZ, setxA+yB:={xa+yb|aA,bB}, and xAyB:=xA+(y)B. For convenience, write xAy:=xAy{1}. Now we have the following result about the 2-abelian complexity of c.

Theorem 2

If the factor set of the generalized Cantor sequence c is mirror invariant and L1(L2)=, then the 2-abelian complexity of c is ℓ-regular.

This paper is organized as follows. In Section 2, we will recall some basic definitions and notations. In Section 3, Theorem 1 will be proved. In the last section, we will prove Theorem 2.

Section snippets

Preliminary

An alphabet Σ is a finite and non-empty set (of symbols) whose elements are called letters. A finite word over the alphabet Σ is a concatenation of letters in Σ. Let Σ denote the set of all finite words over Σ including the empty word ε. When these finite words are of the same length , we write Σ instead of Σ. The concatenation of two finite words u=u(0)u(1)u(m)Σm+1 and v=v(0)v(1)v(n)Σn+1 is defined byuv=u(0)u(1)u(m)v(0)v(1)v(n). For any ωΣ, let |ω| be the length of ω. We adopt the

The proof of Theorem 1

For the generalized Cantor sequence c, let x=x(0)x(1) denote the 2-block coding word block(c,2), in other words, for every n0,x(n)=2c(n)+c(n+1). Note that ni1 for every 1ik1, and thus Fc(2)={00,01,10}. Hence x(n){0,1,2} for every n0. First we give some properties of factors of the word x.

Lemma 1

If u is any factor of length n2 of c, let u=block(u,2). Then|u|1=|u|1pref1(u)and|u|2=|u|1suff1(u).

Proof

It follows from block(01,2)=1 and block(10,2)=2 that |u|1=|u|01 and |u|2=|u|10. Moreover, since

The proof of Theorem 2

The following lemma gives an equivalent definition of k-abelian equivalence.

Lemma 7

[8], Lemma 2.3

Let u,v be two words of length at least k1 over the alphabet Σ. Then, ukv if and only if prefk1(u)=prefk1(v) (resp. suffk1(u)=suffk1(v)) and |u|w=|v|w for every wΣk.

First we investigate the relation between the 2-abelian equivalence of factors of c and the abelian equivalence of corresponding coding factors of x.

Proposition 3

For all u,vFc(n) with n2, let u=block(u,2) and v=block(v,2). Then u2v if and only if u1v

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgement

We warmly thank the referees for their careful reading of the manuscript and their comments.

References (13)

There are more references available in the full text version of this article.

Cited by (0)

This work was supported by NSFC (Nos. 11701202, 11801203, 11871295).

View full text