Asymptotic distributions of the correlation integral based statistics
Author:
Dejian Lai a
| Affiliation: | a School of Public Health, University of Texas Houston Health Science Center, Houston, TX |
DOI:
10.1080/10485259908832757
Publication Frequency:
8 issues per year
Subjects:
Mathematical Economics;
Mathematical Finance;
Medical Statistics;
Statistical Theory & Methods;
Statistics;
Statistics for the Biological Sciences;
Stochastic Models & Processes;
Formats available:
PDF
(English)
View Article:
View Article (PDF)
Abstract
Data generated from some deterministic dynamical systems may appear random. This phenomenon is called chaos. Many classical test statistics cannot distinguish this type of data from IID random observations. Recently, some new test statistics based on correlation integrals were proposed in the literature. However, the statistical properties of these new test statistics have not been studied. This paper studies the asymptotic properties of these test statistics under the IID null hypothesis. Establishing the asymptotic distributions of the test statistics facilitates statistical hypothesis testing. The asymptotic null distributions derived are supported by Monte Carlo simulations.
|
| Keywords: Asymptotic normality; BDS statistic; Chaos; nonparametric tests; time series |
| view references (22) |

Download Citation


CiteULike
Del.icio.us
BibSonomy
Connotea