No label defined (Q4036998)

[For the entire collection see Zbl 0755.00019.] Two morphisms \(A,B\) in a category are shift equivalent (SE) if there are morphisms \(R\), \(S\) and a positive integer \(l\) such that \(AR=RB\), \(SA=BS\), \(RS=A^ l\), \(SR=B^ l\). Two morphisms are strong SE if they are related by a string of shift equivalences each of lag 1 [\textit{R. F. Williams}, Ann. of Math., II. Ser. 98, 120-153 (1973); Errata, ibid. 99, 380-381 (1974; Zbl 0282.58008)]. The question is when SE implies strong SE. The positive answer for subshifts of finite type (SFT) (the morphisms are nonnegative integral matrices) is conjectured by Williams [ibid.] (see also \textit{M. Nasu}'s generalization of Williams theory in [Ergodic Theory Dyn. Syst. 6, 265-280 (1986; Zbl 0607.54026)]). The author gives a counter-example to this conjecture in the category of irreducible SFT with Markov measure (i.e. refuting the conjecture of \textit{W. Parry} and \textit{S. Tuncel} [Bull. Lond. Math. Soc. 14, 16-27 (1982; Zbl 0481.28015)]). In the category of reducible SFT the conjecture was refuted by \textit{K. H. Kim} and \textit{F. W. Roush} [J. Am. Math. Soc. 5, No. 1, 213-215 (1992; Zbl 0749.54013)]. (They also proved in [Ergodic Theory Dyn. Syst. 10, No. 2, 381-383 (1990; Zbl 0674.20041)], that for a sofic system (SFT which is a factor of a topological Markov shift) the eventual isomorphism is decidable).