Biased random walks on Galton-Watson trees with leaves (Q662434)

Let \(Z\) be a random variable having the offspring distribution in a Galton-Watson process with a single ancestor, and put \(f(s):=\operatorname{E}s^Z\). It is assumed that \(\operatorname{E}Z\in (1,\infty)\) and that \(\operatorname{P}\{Z=0\}>0\). Denote by \(q\) the extinction probability of the process, and let \(T\) denote the corresponding Galton-Watson tree conditioned on the event of nonextinction. The present paper investigates the \(\beta\)-biased \((\beta>1)\) random walk \((X_n)_{n\in N_0}\) on \(T\) with \(X_0=0\). Assuming that \(\operatorname{E} Z^2<\infty\) and that \(\beta>1/f'(q)\) (the sub-ballistic regime), the authors prove that the laws of \(\big(n^{-1/\gamma}\Delta_n\big)\) and \(\big(n^{-\gamma}|X_n|\big)\) are tight, where \(\gamma:=-\log f'(q)/\log \beta\) and \(\Delta_n:=\inf\{i\in\mathbb N_0: |X_i|=n\}\). It is further shown that, for \(\beta\) large enough, the sequence \(n^{-1/\gamma}\Delta_n\) only converges in distribution along the explicitly given subsequence, the limit being an infinitely divisible law \(\mu\). Several important properties of \(\mu\) are pointed out. As the authors state in the abstract, ``key tools for the proof are the classical Harris decomposition for Galton-Watson trees, a new variant of regeneration times and the careful analysis of triangular arrays of i.i.d. heavy-tailed random variables''.