Multivariate maxima of moving multivariate maxima

From MaRDI portal



Abstract: We define a class of multivariate maxima of moving multivariate maxima, generalising the M4 processes. For these stationary multivariate time series we characterise the joint distribution of extremes and compute the multivariate extremal index. We derive the bivariate upper tail dependence coefficients and the extremal coefficient of the new limiting multivariate extreme value distributions.


Let \(\mathbf{Z}_{m,n}=(Z_{m,n,1},\dots,Z_{m,n,d})\), \(m\in\mathbb N\), \(n\in\mathbb Z\), be an array of independent copies of a random vector \(\mathbf Z\) with standard Fréchet margins and copula \(C_{\mathbf Z}\). A (stationary) multivariate maxima of moving multivariate maxima (M5) process is defined by NEWLINE\[NEWLINEY_{n,j}:=\max_{m\in\mathbb N}\max_{k\in\mathbb Z} \alpha_{m,k,j} Z_{m,n-k,j},\;j=1,\dots,d,\;n\in\mathbb N,NEWLINE\]NEWLINE where the nonnegative constants \(\alpha_{m,k,j}\) satisfy NEWLINE\[NEWLINE\sum_{m\in\mathbb N}\sum_{k\in\mathbb Z} \alpha_{m,k,j}=1\;\text{for}\;j=1,\dots,d.NEWLINE\]NEWLINE Put \(\mathbf{Y_n}=(Y_{n,1},\dots,Y_{n,d})\). Under the condition that NEWLINE\[NEWLINEC_{\mathbf{Z}}(u_1,\dots,u_d) ge\prod_{i=1}^du_i,\;(u_1,\dots,u_d)\in[0,1]^d,NEWLINE\]NEWLINE the following results are derived:NEWLINENEWLINE the limiting joint distribution of the componentwise maxima of \(\mathbf{Y}_1,\dots,\mathbf{Y}_n\), the multivariate extremal index, the bivariate upper tail dependence coefficient and the extremal coefficient of the limiting multivariate extreme value distribution.











This page was built for publication: Multivariate maxima of moving multivariate maxima

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q449003)