Estimating a covariance function having an un unknown scale kakameter
The structure of a family \({\mathcal P}\) of Gaussian measures with zero mean and covariance \(\alpha\) R(s,t) is discussed. For fixed R this family is either homogeneous or composed of singular measures depending on whether the dimension of the reproducing kernel Hilbert space \(H=H(R)\) is finite or not. For the case when dim H\(<\infty\), the MLE \({\hat \alpha}\) is given; when dim H\(=\infty\) an almost sure discriminator is constructed. For the more general setting when R itself depends upon a parameter \(\theta\) the orthogonal decomposition of \({\mathcal P}\) and estimates for both \(\alpha\) and \(\theta\) are obtained.
This page was built for publication: Estimating a covariance function having an un unknown scale kakameter
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3823680)