Some comparisons for Gaussian processes

From MaRDI portal



Abstract: Extensions and variants are given for the well-known comparison principle for Gaussian processes based on ordering by pairwise distance.


The author establishes extensions of the known comparison principles for Gaussian processes. Two of them are: Let \(\{X_i, i\in I\}\) and \(\{Y_i, i\in I\}\) be two mean-zero Gaussian processes indexed by a denumerable set \(I\), and \(\{m_i, i\in I\}\) be arbitrary constants. If \(E(X_i- X_j)^2\leq E(Y_i- Y_j)^2\) for all \(i,j\in I\), then \[ E\sup_i \{X_i+ m_i\}\leq E\sup_i \{Y_i+ m_i\}. \] If additionally \(EX^2_i\leq EY^2_i\) for all \(i\in I\), then \[ \int^\infty_t P\Biggl(\sup_i X_i> s\Biggr) ds\leq \int^\infty_t P\Biggl(\sup_i Y_i> s\Biggr) ds \] for all \(t> 0\).




Cited in
(22)








This page was built for publication: Some comparisons for Gaussian processes

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4501096)