Predicting Transforms of Stable Noise and other Gaussian Mixtures.
Abstract
Stationary stable processes that are Fourier transforms of symmetric stable independent increments processes are shown to have a.s. finite conditional expectation of X sub t given X sub s and conditional variance of X sub t given X sub(t - delta), X sub(t -2 delta). The associated conditional expectation predictors are nonlinear in (X sub s, s<t) but are mixtures of predictors of the usual type based on the Gaussian model.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jul 01, 1987
- Accession Number
- ADA189280
Entities
People
- Raoul Lepage
Organizations
- University of North Carolina at Chapel Hill