The most visited sites of symmetric stable processes (Q1566939): Difference between revisions

Let \(X\) be a symmetric real-valued stable process of index \(\alpha\in (1, 2]\) and \(\{L^x_t; t\geq 0, x\in {\mathbb R}\}\) its jointly continuous local time process. The authors are interested in the so-called most visited sites of \(X\) up to time \(t\), that is the set \(\{x\in{\mathbb R}: L^x_t = \sup_{y\in {\mathbb R}}L^y_t\}\), and more precisely in the maximal most visited site \(V(t)\), defined as the greatest element of the above set (but actually, any point of the set could be chosen for their study). Their main result is the transience of \(V(t)\), and they also give a lower bound for the rate of growth of \(V(t)\), extending a classical result of the first author and \textit{P. S. Griffin} for Brownian motion [Z. Wahrscheinlichkeitstheorie Verw. Geb. 70, 417-436 (1985; Zbl 0554.60076)]. Notice that an upper bound for the rate of growth is also obtained in the above paper, which is not the case in the paper under review. Notice also that the exact rate of growth of \(V(t)\) remains an open problem for Brownian motion. Here, the main tool for the proof is a theorem of Ray-Knight type for symmetric Markov processes which was recently obtained by the second and third authors et al. (Preprint), and which here makes it possible to study the local time process with the help of Gaussian techniques. This paper is very clearly written and remains accessible to the non-specialist.