Compound Poisson limit theorems for high-level exceedances of some non-stationary processes (Q1433464)

Consider the sequence of random variables \(X_n=\varphi (\xi_n, Y_n)\) where \(\varphi\) is a regular regression function, \(\xi=\{ \xi_n \}\) is a stationary sequence of weakly dependent random variables and \(Y=\{ Y_n\}\) is a non-stationary sequence of random variables satisfying some ergodic conditions. Assume that \(\xi\) and \(Y\) are independent. The following theorem is proved: The point process of high-level exceedances \(N_n(B)=\sum_{i/n\in B} { 1}(X_i>u_n)\) weakly converges to a compound Poisson process as \(u_n\) becomes large.