scientific article; zbMATH DE number 18823
[For the entire collection see Zbl 0732.00014.] Let \(X({\mathbf t}),\;{\mathbf t}\in [0,1]^ n\), be a sample continuous centered Gaussian field. For some positive \(\alpha_ 1,\dots,\alpha_ k, u\) denote \(g_ u({\mathbf t}) = (u^{-2/\alpha_ 1}{\mathbf t}^ 1,\dots,u^{-2/\alpha_ k}{\mathbf t}^ k)\) where \({\mathbf t} =({\mathbf t^ 1},\dots,{\mathbf t}^ k)\), \({\mathbf t}^ 1 = (t_ 1,\dots,t_{l_ 1})\), \({\mathbf t}^ 2 = (t_{l_ 1+1},\dots,t_{l_ 1+l_ 2}),\dots\) and \(\sum_{i=1}^ k l_ i = n\). For the probability measures \[ P_{u,T}(\cdot)= P\bigl( Y_{u,T} \in \cdot \mid \max_{{\mathbf h} \in [0,1]^ n} X({\mathbf h})>u\bigr) \] defined on the Borel \(\sigma\)- field of \(C([0,1]^ n)\) the limit behaviour is studied as \(u\to \infty\) and after that \(T \to \infty\). Here \(Y_{u,T}({\mathbf t})=u(X(g_ u({\mathbf t} +T\tau_ u))-u)\), \(T>0\), and the random vector \(\tau_ u\) is the first in the lexicographic sense of \(\mathbf k\) such that \(\max_{{\mathbf h}\in \Delta_{u,T}({\mathbf k})} X{\mathbf h} >u\) (one uses the collection of \(n\)-dimensional parallelepipeds of the form \(\Delta_{u,T}({\mathbf k})=g_ k(T_{\mathbf k} +[0,1]^ n)\), \({\mathbf k}=(k_ 1,\dots,k_ n) \in \mathbb{N}^ n\)). Certain condition involving \(\alpha_ 1,\dots,\alpha_ k\) is imposed on the covariance function \(\gamma({\mathbf t})\) near \({\mathbf t}=0\).
- scientific article; zbMATH DE number 3956125
- Probabilities of high excursions of Gaussian fields
- Limit distributions of characteristics of exceeding a level by a Gaussian field
- On probability of high extremes of Gaussian fields with a smooth random trend
- The limit theorems on extremes for Gaussian random fields
This page was built for publication:
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3976237)
