An asymptotic theory for empirical reliability and concentration processes (Q1083821)

The primary aim of this monograph is to build up a unified asymptotic theory for empirical total time on test, Lorenz, and concentration processes. Rather than using results directly from the theory of general quantile processes, the authors work out special techniques tailored exactly for the present reliability and economic processes. New Lorenz type empirical processes are introduced, for instance, the empirical Shannon and the associated empirical redundancy processes. Similarly, modified total time on test processes are also considered. All these processes fit into the authors' unified theory. The unified theory covers the weak convergence results and provides further insights into the nature of mean residual life processes. The authors' approach is the approximation method which also makes it possible to see clearly what additional assumptions are needed on the underlying distribution in order that the considered empirical processes be similar to corresponding Gaussian processes in a stronger sense than the weak invariance principle obtained in the ''first step'' of this approach. The second ''strong step'' then provides loglog law consistency rates (uniformly) and makes possible the transition of the fluctuational behaviour of the approximating Gaussian processes over to the empirical processes up to the rate of approximation, depending again on the regularity of the underlying distribution. This monograph is therefore methodological, demonstrating the strength of an approach in a non-trivial situation.