Extremal clustering in non-stationary random sequences (Q825998)

The paper contributes to a better understanding of the phenomenon of \textit{clustering of extremes} in sequences of dependent random variables. The authors analyze the case of non-stationary and identically distributed sequences of random variables which satisfy \textit{the asymptotic independence of maxima} condition. The main result of the paper regards the asymptotic distribution of the sample maxima. For the proof, the standard technique of block-clipping is used. Under appropriate restrictions, the parameter that measures the average extremal clustering of the sequence is then estimated, with a focus on periodic sequences. The limiting distribution of the normalized interexceedance times is also highlighted. This refined study is complemented by the presentation of two examples of non-stationary Markov sequences with a periodic dependence structure and common marginal distributions.