Construction of a Markov field on an infinite tree (Q919708)

Let T be an infinite tree, let \((X_ t,{\mathcal A}_ t)\) be a measurable space for each \(t\in T\), set \((X,{\mathcal A})=(\prod_{t\in T}X_ t,\otimes_{t\in T}{\mathcal A}_ t),\) and let \(Z_ t\) denote the projection of X onto \(X_ t\), \(t\in T\). By making use of the Ionescu- Tulcea theorem, a probability P on \({\mathcal A}\) is constructed such that \((Z_ t)_{t\in T}\) is a Markov random field with respect to P, i.e. \[ (*)\quad P(Z_ s\in A | \quad (Z_ t)_{t\in T-\{s\}})=P(Z_ s\in A | \quad (Z_ t)_{t\in N_ s})\quad a.s. \] for any \(s\in T\) and \(A\in {\mathcal A}_ s\), where \(N_ s\) stands for the set of all vertices which are neighbours of s. To get (*), several interesting propositions about Markov triples are first proved. The whole construction is finally specialized to introduce the notion of a Galton-Watson random field, a few properties of which are discussed.