Bilinear stochastic models and related problems of nonlinear time series analysis. A frequency domain approach (Q1297718): Difference between revisions

The present work is intended to be a systematic statistical analysis of bilinear processes in the frequency domain. The first two chapters are devoted to foundations: basic theory of nonlinear functions of stationary Gaussian processes, Hermite polynomials, cumulants, higher order spectra, multiple Itô integrals and finally chaotic Wiener-Itô spectral representation of subordinal processes. Chapter 3 contains the results concerning bilinear processes. To offer an easier understanding of the technique of chaotic representation, three levels of bilinear processes are considered: the simple bilinear model, the general bilinear model with scalar value, and the multiple bilinear model. In each case explicit assumptions of second order stationarity and expressions for the second order spectrum are given. The generalized autoregressive conditionally heteroscedastic model is investigated by the same method as the bilinear one, and its basic properties are shown. There are two chapters for general nonlinear time series problems. Chapter 4 covers non-Gaussian estimation. Here the bispectrum is dealt with. Explicit expressions are given for the asymptotic variance and the asymptotic normality and consistency is shown. The following chapter considers a bispectrum based test for checking linearity. Some applications are given in the last chapter.