State Space Models for Gaussian Stochastic Processes.
Abstract
A comprehensive theory of stochastic realization for multivariate stationary Gaussian processes is presented. It is coordinate-free in nature, starting out with an abstract state space theory in Hilbert space, based on the concept of splitting subspace. These results are then carried over to the spectral domain and described in terms of Hardy functons. Each state space is uniquely characterized by its structural function, an inner function which contains all the systems theoretical characteristics of the corresponding realizations. Finally coordinates are introduced and concrete differential-equation-type representations are obtained. This paper is an abridged version of a forthcoming paper, which in turn summarizes and considerably extends results which have previously been presented in a series of preliminary conference papers. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Jul 05, 1980
- Accession Number
- ADA092388
Entities
People
- Anders Lindquist
- Giorgio Picci
Organizations
- University of Kentucky