Ordering properties of the smallest order statistics from generalized Birnbaum-Saunders models with associated random shocks (Q683864)

This paper deals with lifetimes of systems consisting of \(n\) components, either in series or in parallel. Each component can be impacted by a random shock. When the impact is not effective the lifetime of the component follows a non-negative generalized Birnbaum-Saunders distribution. When the impact is effective the component has a zero lifetime. A Bernoulli random variable is used to play the role of the impact. All the involved random variables are assumed to be independent. Usual stochastic order results are established for the smallest and for the largest order statistics when the matrix of the parameters of both the above distributions is conveniently changed. The proofs of the results follow from a clever use of two lemmas in [\textit{N. Balakrishnan} et al., ``Stochastic comparisons of series and parallel systems with generalized exponential components '', IEEE Trans. Reliab. 64, No. 1, 333--348 (2015; \url{doi:10.1109/TR.2014.2354192})] referred to in the paper. We have here a nice contribution in reliability/survival analysis. One point is that the statement ``since \(h\) is strictly increasing and concave, \(h^{-1}\) is strictly increasing and concave'' in the proof of Theorem 3.1 is not correct, but hopefully, it will be conveniently fixed soon.