On palindromic factorization of words

Given a finite word u, we define its palindromic length |u|pal to be the least number n such that u=v1v2Â?vn with each vi a palindrome. We address the following open question: let P be a positive integer and w an infinite word such that |u|pal≤P for every factor u of w. Must w be ultimately periodic? We give a partial answer to this question by proving that for each positive integer k, the word w must contain a k-power, i.e., a factor of the form uk. In particular, w cannot be a fixed point of a primitive morphism. We also prove more: for each pair of positive integers k and l, the word w must contain a position covered by at least l distinct k-powers. In particular, w cannot be a Sierpinski-like word.

[1]  Jean Savinien,et al.  A characterization of subshifts with bounded powers , 2011, Discret. Math..

[2]  M. Lothaire Algebraic Combinatorics on Words , 2002 .

[3]  Giuseppe Pirillo,et al.  Episturmian words and some constructions of de Luca and Rauzy , 2001, Theor. Comput. Sci..

[4]  Lucian Ilie,et al.  The "runs" conjecture , 2011, Theor. Comput. Sci..

[5]  Antonio Restivo,et al.  Periodicity and the Golden Ratio , 1998, Theor. Comput. Sci..

[6]  Olexandr Ravsky On the Palindromic Decomposition of Binary Words , 2003, J. Autom. Lang. Comb..

[7]  Wojciech Rytter,et al.  On the maximal sum of exponents of runs in a string , 2012, J. Discrete Algorithms.

[8]  Luca Q. Zamboni,et al.  Palindromic richness , 2008, Eur. J. Comb..

[9]  Jarkko Peltomäki,et al.  Introducing privileged words: Privileged complexity of Sturmian words , 2013, Theor. Comput. Sci..

[10]  Jean-Paul Allouche,et al.  Palindrome complexity , 2003, Theor. Comput. Sci..