Statistics of the occupation time of renewal processes (Q5949324): Difference between revisions

The authors systematically study renewal theory for broad distributions. By this we mean the following. Let \((\tau_i)_{i\in\mathbb N}\) be an i.i.d.~sequence of positive random variables whose upper tails behave as \(P(\tau_1>t)\sim {\text{ const}}\times t^{-\theta}\) as \(t\to\infty\), with a parameter \(\theta\in(0,2)\). Consider the random walk \((t_n) _{n\in\mathbb N_0}\) given by \(t_n=\tau_1+\dots+\tau_n\) and define \(N_t\) as the largest \(n\) with \(t_n\leq t\). The times \(t_1,t_2,\dots\) are interpreted as the successive times at which a given one-dimensional stochastic process crosses the origin, and \(t_{N_t}\) is the number of crossings before time \(t\). The case \(\theta=1/2\) corresponds to simple random walk respectively Brownian motion and is well studied and classical. The authors motivate the consideration of other values of \(\theta\) with certain physical models related to phase persistence. The paper under review provides a host of formulas for distributions (and Laplace transforms, fluctuations, scaling limit distributions, etc.) of variables related to the renewal sequence \((t_n)_n\), the most important of which are \(t_{N_t}\), \(t-t_{N_t}\), \(t_{N_t+1}-t\), and the vector \((t_{N_t},\dots,t _{N_{t+t'}})\). All the methods of the proofs are well-known or straightforward extensions of known proofs, and the authors only indicate them.