Abstract
The paper presents a new nonparametric test for independence of two vectors. The idea is based on zonotope approach by G. Koshevoy, H. Oja and others, see [4, 5]. Under the independence hypothesis the test statistic converges in distribution to the supremum of a certain Gaussian field, and its asymptotic distribution is found using the theory of extrema of random Gaussian fields developed by V. Piterbarg and Yu. Tyurin, see [6, 8]. In contrast to traditional correlation coefficients the formula is not symmetric.
Similar content being viewed by others
References
A. V. Bulinsky and A. N. Shiryaev, Theory of Random Processes (Fizmatlit, Moscow, 2003) [in Russian].
M. Ghosh, “Rank Statistics and Limit Theorems”, in Nonparametric Methods. Handbook of Statistics, Ed. by K. Krishnaiah and P. K. Sen (North-Holland, Amsterdam, 1984), Vol. 4, p. 145–171.
O. Kallenberg, Foundations of Modern Probability (Springer, New York, 1997).
G. Koshevoy, J. Möttönen, and H. Oja, “On the Geometry of Multivariate L1 Objective Functions”, Allgemeines Statist. Archiv. 88, 137–154 (2004).
J. Möttönen, G. Koshevoy, H. Oja, and Y. Tyurin, Multivariate Tests for Independence Based on Zonotopes, Manuscript (2005).
V. I. Piterbarg, Asymptotic Methods in the Theory of Gaussian Processes and Fields (Amer. Math. Soc., Providence, RI, 1996).
V. Piterbarg and Yu. Tyurin, “Testing for Homogeneity of Two Multivariate Samples: a Gaussian Field on a Sphere”, Math. Methods Statist. 2, 147–164 (1993).
V. I. Piterbarg and Yu. N. Tyurin, “Multivariate Rank Correlations: A Gaussian Field on a Direct Product of Spheres”, Theory Probab. Appl. 45, 246–257 (2000).
T. Romanova, “Numerical Analysis of Piterbarg-Tyurin Procedure for Testing Homogeneity and Independence”, Theory Probab. Appl. 45, 681–687 (2000).
J. Serfling, Approximation Theorems of Mathematical Statistics (Wiley, New York, 1980).
Author information
Authors and Affiliations
Corresponding author
About this article
Cite this article
Sukhanova, E.M. A test for independence of two multivariate samples. Math. Meth. Stat. 17, 74–86 (2008). https://doi.org/10.3103/S1066530708010067
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.3103/S1066530708010067