Weak convergence of sequences of first passage processes and applications (Q678383): Difference between revisions

The paper investigates the weak convergence of (properly normalized) first passage processes \[ X_n^{-1}(t) = \inf \{x:X_n(x)\geq t\} \] to a limiting process. It is shown how, for a wide range of 'inverse' processes, asymptotic behavior of \(X_n^{-1}\) can be derived from corresponding weak convergence results for \(X_n\). First, the main theorem is proved, in a rather general form. Then a number of special cases is presented and the theorem is applied to them. Thus, the weak convergence results are given for \(U\)-statistic quantile processes, for perturbated quantiles (an inversion of a smooth estimator of the c.d.f), conditional quantile processes, empirical excursion above a given barrier, as well as for stopping rules in the renewal theory and for extreme value processes. The weak convergence results are also adapted to the asymptotic properties of integrated kernel quantiles. Further, in this framework, the strong law of large numbers is proved and a Bahadur-type representation is derived (as well as its consequence, the law of iterated logarithm) for \(U\)-statistic integrated kernel quantiles.