Excursions of a normal random walk above a boundary (Q1573641): Difference between revisions

A random walk is considered to make an excursion above a boundary \(g(i)\) from steps \(m\) to \(n\) if it is below \(g(m-1)\) after \(m-1\) steps, then above \(g(i)\) for \(i\) between \(m\) and \(n\). Thus the number of excursions is the number of times the walk crosses from below to above the boundary. A normal random walk is a random walk in which the steps are independent, identically distributed normal random variables. This article gives an integral condition on the boundary function for a normal random walk to have finitely many excursions above the boundary, and an asymptotic for the number of excursions in the first \(N\) steps if the expectation is infinite. An important example occurs when the steps are standard normal variables, and \(g(i)=c\sqrt{\log\log x}\). By the law of the iterated logarithm, this walk spends an infinite expected time above this boundary if \(c\leq\sqrt{2}\). However, the expected number of excursions is finite if \(c>1\). Upper and lower bounds are computed for this expected number.