On weighted heights of random trees (Q2639445)

To each edge in the family tree of a (typically critical) simple branching process is attached a weight W and these weights are i.i.d. and independent of the tree. The results concern the limiting behaviour of the maximum total weight down any branch of the tree, conditional on the total number n of edges, as n tends to infinity. This depends upon the tail of the distribution of W. If this tail obeys a central-limit type condition, then only the tree structure is relevant and the limit distribution is that of the maximum of a Brownian excursion (for which an interesting straightforward explanation is given in the case of geometric offspring distribution). If the tail of the distribution of W is regularly varying with exponent \(0\leq \alpha <2\), then the tree structure becomes irrelevant and an extreme value limit distribution emerges. There is also a threshold case which is briefly mentioned. It should be noted that the model is identical to the branching random walk, about which there is a body of literature, if we interpret ``weight'' as ``displacement''. This is not mentioned explicitly by the authors.