On the boundary at infinity for branching random walk (Q6177617)

Given a transient, irreducible Markov chain \(M= ((X_n)_{n\ge 0},\mathbf{P}_X)\) with a countable state space \(S\) and a supercritical Bienaymé-Galton-Watson process satisfying the \(X\log X\) condition, i.e., an offspring distribution \(\mu\) with \(\overline{\mu}:=\sum_n n\,\mu(n)>1\) and \(\sum_n(n\log n)\mu(n)<\infty\), consider the branching random walk \(((B_n)_{n\ge 0}\mathbb{P}_X)\) constructed according to the following rules: All particles alive have positions in \(S\) . When a particle branches, it is replaced at its position according to \(\mu\) by a set of particles which as part of the same time step independently of each other take a step in \(S\) according to \(M\). Start the process with one particle at \(o\in S\) and let \(X\) be the limit point of the trajectory \((X_n)_n\) on the Martin boundary. The authors show that \(B_n/\overline{\mu}^n\) converges almost surely weakly to a random measure \(\mathbf{W}\) on the Martin compactification \(\mathcal{M}\) of \(M\) that is supported on the active part of the Martin boundary of \(M\) and satisfies \(\mathbb{E}_o\mathbf{W}(A)=\mathbf{P}_o (X\in A)\) for every Borel set \(A\subseteq\mathcal{M}\). In addition, several open problems regarding \(\mathbf{W}\) are discussed. Parallel to this investigation similar results allowing the consideration of spatially dependent branching and of compactifications other than the Martin one were obtained by \textit{V. A. Kaimanovich} and \textit{W. Woess} [Ann. Inst. Henri Poincaré, Probab. Stat. 59, No. 4, 1951--1983 (2023; Zbl 1530.60083)].