Marginally closed processes with local interaction (Q1201758)

The authors continue the study of marginally closed locally interacting processes introduced by \textit{I. A. Ignatyuk}, \textit{V. A. Malyshev} and \textit{S. A. Molchanov} [Sel. Math. Sov. 8, No. 4, 351-384 (1989; Zbl 0701.60100)]. These are discrete time Markov processes of configurations on \(\mathbb{Z}^ \nu\) with values in a denumerable set \(S\) with the property that the finite-dimensional distributions at time \(t\) of order \(m\) can be expressed as linear combinations of the finite-dimensional distributions at time \(t-1\) of order \(\leq m\). The main results are derived for conditionally independent marginally closed processes. The authors study their ergodic behaviour and derive their invariant measures for translational invariant initial distributions, which for dimensions \(\nu\geq 3\) have to satisfy a decay condition of the correlation functions. The proofs are based on a path expansion. Also several examples of non-conditionally independent marginally closed processes are discussed.