Title :
Testing proportionality for autoregressive processes
Author_Institution :
CNRS, Univ. de Cergy Pontoise, Neuville Sur Oise, France
fDate :
3/1/2003 12:00:00 AM
Abstract :
We introduce a new hypothesis test to determine wether or not two autoregressive spectral densities are proportional. A test for autoregressive coefficient ity or randomness is deduced. We also derive the exact asymptotic behavior for these tests under parametric alternatives and show that, given a significance level, our tests are the most powerful (MP) tests among all tests.
Keywords :
autoregressive processes; random processes; spectral analysis; AR processes; autoregressive coefficient; autoregressive processes; autoregressive spectral density; exact asymptotic behavior; proportionality testing; randomness; significance level; Autoregressive processes; Biology; Gaussian processes; Helium; Parameter estimation; Reliability engineering; Reliability theory; Robustness; Testing; Time series analysis;
Journal_Title :
Information Theory, IEEE Transactions on
DOI :
10.1109/TIT.2002.808122
