scientific article; zbMATH DE number 1336728
The paper consists of three sections. In the first one the bilinear realizable processes \(y_t\), \(t=0,\pm 1,\pm 2,\dots\) are considered, i.e. the processes which can be represented in the form \( x_t=Ax_{t-1}+Dx_{t-1}v_{t-1}+bv_t-\sigma Db \), \(y_t=c^T x_t\), where \(v_s\) is a Gaussian white noise series, \(Ev_s^2=\sigma^2\), \(A\) and \(D\) are \(m\times m\) matrices, \(c\) and \(b\) are \(m\) dimensional vectors. The spectrum of \(y_t\) is described. In the second section the bispectrum is defined and conditions are deduced in terms of bispectrum of the innovation process under which the best linear and the best quadratic predictors for the third order stationary process are equal. In the third section these results are applied to bilinear realizable processes of the second Hermit degree. For the entire collection see [Zbl 0896.00028].
- On the linear prediction of multivariate \((2,p)\)-bounded processes
- On Wiener–Ito representation and the best linear predictors for bilinear time series
- The exact bispectra for bilinear realizable processes with hermite degree 2
- When the bispectrum is real-valued
- scientific article; zbMATH DE number 957396 (Why is no real title available?)
- On the predictor of non-full-rank bivariate stochastic processes
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 Q4258808)
