On Random Correlation Matrices. 2. The Toeplitz Case
Abstract
The problem of generating random Toeplitz correlation matrices is considered. Several methods are proposed, of which the most promising, as determined by both computational complexity and spectral randomness, seems to be that based on the characteristic function of random discrete probability measure. A number of interesting theoretical issues are recorded for further investigation. This report follows an earlier one devoted to general random correlation matrices. In both cases the intent is to use such matrices to simulate random data for the testing of certain group-theoretic signal processing algorithms. Keywords: Statistical tests, Eigenvalues.
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 23, 1989
- Accession Number
- ADA208229
Entities
People
- Richard B. Holmes
Organizations
- Massachusetts Institute of Technology