Capacity of the range of branching random walks in low dimensions (Q2135117)

Consider a critical Galton-Watson process with branching law \(\mu\). Due to criticality each associated tree is finite. Let \((\mathcal{T}_n)_{n\ge 0}\) be a sequence of independent such trees attached to points of a fixed infinite ray (spine) in \(\mathbb{Z}^d\), resulting in a forest \(\mathcal{T}\), which itself is a tree rooted at the root of the spine. Let \(\mathbf{P}_\mu\) be its law. The branching random walk is now constructed as tree-indexed random walk \((V_u)_{u\in\mathcal{T}}\) in \(\mathbb{Z}^d\) as follows: Let \(\theta\) be a probability distribution on \(\mathbb{Z}^d\) and attach to all edges of \(\mathcal{T}\) i.i.d. random variables distributed according to \(\theta\) and independent of \(\mathcal{T}\), one to each edge. Define \(V_\varnothing:=0\) and for each \(u\in\mathcal{T}\setminus\{\varnothing\}\) let \(V_u\) be the sum of those random variables which are attached to the edges in the (unique) simple path relating \(u\) to the root \(\varnothing\). The range studied under the law \(\mathbf{P}_{\mu,\theta}\) of \((V_u)_{u\in\mathcal{T}}\) is \(R_n:= \{V_u:u\in\bigcup^{n-1}_{j=0}\mathcal{T}_j\}\). Admitting dimensions \(d=3,4\), and \(5\), assuming that \(\mu\) not only has mean \(1\), \(\mu=\delta_1\), but also finite variance, assuming further that \(\theta\) is symmetric, irreducible, with some finite exponential moments, and letting \(\eta\) be an aperiodic, irreducible probability measure on \(\mathbb{Z}^d\) with mean \(0\) and finite \((d+1)\)th moment, it is shown that a.s. under \(\mathbf{P}_{\mu,\theta}\) the capacity of \(R_n\) with respect to \(\eta\), as \(n\to\infty\), equals \(n^{(d-1)/2+o(1)}\), where \(o(1)\to 0\) a.s., as \(n\to\infty\). For a related paper by \textit{T. Bai} and \textit{Y. Hu} on the capacity of the range of tree-indexed random walks see [Ann. Appl. Probab. 32, No. 3, 1557--1589 (2022)].