On linear shifts of finite type and their endomorphisms (Q2069809): Difference between revisions

A \textit{linear subshift} on a group \(G\) is a closed \(G\)-invariant subset \(\Sigma=A^G\) for some finite-dimensional vector space \(A\). A \textit{linear cellular automaton} is a \(G\)-equivariant continuous linear map \(\tau\colon B^G\to A^G\), or more generally the restriction of such a map to a subshift \(\Sigma\le B^G\). These notions are the natural linear extension of classical combinatorial subshifts and cellular automata, but they are in fact quite different: the linearity condition is very restrictive (combinatorial maps typically yield \textit{multi}-linear maps and not linear ones). Pursuing the same line of research done in previous papers, the authors explore the restrictions imposed by the linearity condition. A \textit{subshift of finite type} is the kernel of a linear cellular automaton, and a \textit{linear-sofic subshift} is the image of a subshift of finite type by a linear cellular automaton. Subshifts of finite type are characterized (Theorem 1.1) by a ``Noetherianity condition'': they cannot be the intersection of a strictly decreasing sequence of linear subshifts. Thus if \(K[G]\) is one-sided Noetherian (for example, if \(G\) is polycyclic) then all subshifts are of finite type. The authors then study the limit set \(\Omega(\tau)=\bigcap_{n\ge0}\tau^n(\Sigma)\) of a linear cellular automaton \(\tau\colon\Sigma\to\Sigma\). In particular, they show (Theorem 1.8) that if \(\Sigma\) is linear-sofic then \(\tau\) is nilpotent, i.e., \(\tau^n=0\) for some \(n\), if and only if \(\Omega(\tau)=0\). These are linear counterparts of classical results of theory of cellular automata, see for instance Theorem 3.5 in [\textit{K. II. Culik} et al., SIAM J. Comput. 18, No. 4, 831--842 (1989; Zbl 0691.68060)].