scientific article; zbMATH DE number 3940442

From MaRDI portal
Publication:3711484





We consider the observations \(X_ 1,...,X_ n\) drawn from a stationary Gaussian time series with a spectral density f. Let \(\phi\) be a known functional defined on the set of spectral densities. The problem of estimation of the value \(\phi\) (f) of \(\phi\) at the point f is considered here. In Sect. 1 we indicate some lower bounds for the precision of nonparametric estimators; in Sect. 2 an asymptotically efficient estimator of \(\phi\) (f) is constructed for the case where \(\phi\) is a linear functional; in Sect. 3 the same problem is solved for some nonlinear but sufficiently smooth functionals. It seems that our results are new even in the case of such linear functionals as for example the value of jth covariant: the empirical jth covariation \(R^*_ j=n^{-1}\sum^{n-j}_{k=1}X_ kX_{k+j}\) is the best nonparametric estimator of jth covariation (from an asymptotical point of view).




Cited in
(24)








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 Q3711484)