On Prediction of Harmonizable Stable Processes.
Abstract
Spectral and time domain criteria for a harmonizable stable process to be regular are given, which provide an orthogonal moving average representation. Also criteria for such processes to have linear predictor filters are obtained; these include the positivity of the distance and of the angle between past and future. In the process, the notion of angle between isotropic complex stable random variables is introduced and studied. Keywords: Harmonizable stable processes; Moving average representation; Linear predictors; Angle; Isotropic complex random variables.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jul 01, 1985
- Accession Number
- ADA161412
Entities
People
- A. G. Miamee
- S. Cambanis
Organizations
- University of North Carolina at Chapel Hill