Speed of convergence for laws of rare events and escape rates (Q2258838): Difference between revisions

Let \(X_0 , X_1 , \dots\) be a stationary stochastic process on some measure space with measure \(\mathbb P\) and \[ M_n = \max\{X_0 , \dots , X_n \}, n \in \mathbb N. \] The authors obtain the error terms on the rate of convergence to the extreme value laws (EVL), i.e., they obtain an estimate for the difference \[ |\mathbb P(M_n \leq u_n ) - e{}^{-\theta\tau} |, \] where \(\tau > 0\), the sequence \(u_n = u_n (\tau )\) satisfy \[ \lim_{ n \to \infty} n \mathbb P(X_0 > u_n ) = \tau, \] and \[ \theta = \lim_{ n \to \infty} \frac{ \mathbb P(X_0 > u_n , X_1 \leq u_n , \dots , X_n \leq u_n )}{\mathbb P(X_0 > u_n )}, \] if it exists. This estimate is obtained under an assumption on the sequence \(X_n\) weaker than the conditions used in earlier studies. Sharper estimates for EVL and for related hitting time statistics (HTS) also obtained in the context of chaotic dynamical systems such as, for example, Rychlik systems.