Abstract
The spectral or frequency domain of a multivariate weakly harmonizable process is a vector space of vector functions that are integrable relative to a positive definite matrix bimeasure. It has a norm derived from a Gramian inner product. The completeness property of this space, which is needed for many applications, has been open for some years. An affirmative solution and an application to optimal least-squares estimation of a weakly harmonizable signal are presented here.
Keywords: completeness property, frequency domain, positive definite bimeasure integrals
Full text
PDF

