Some results on the residual life at random time (Q1594868)

Let \(X\) and \(Y\) be nonnegative random variables. The expression \(X_Y=X- Y|X>Y\), called the residual life at random time (RLRT), may be interpreted as the residual life of system \(A\) (of \(X)\) when the system \(B\) (of \(Y)\) fails. Another interpretation of the RLRT \(X_Y\) comes from the queueing theory. Depending on the stochastic orderings defined on random variables (hazard rate ordering, reverse hazard rate ordering, stochastic ordering, increasing convex ordering, Laplace transform ordering), the authors consider several classes of life distributions (denoted as IFR, DFR, DMRL, IMRL, etc.) and investigate their closure properties. Namely, under the assumption that \(Y\) has decreasing reverse hazard rate, the paper shows that when \(X\) is in any of the considered classes IFR, DFR, DMRL, and IMRL of life distributions, then the RLRT \(X_Y\) belongs to the same class as \(X\). Finally, some useful bounds for the distribution function and the moment of \(X_Y\) are derived. Because the idle time in the classical GI/G/1 queueing system can be interpreted as the RLRT \(X_Y\), the obtained results may also be applied to the study of this type of queueing systems.