Asymptotically most powerful rank tests for multivariate randomness against serial dependence (Q1262052)

The authors introduced a class of linear serial multirank statistics for the problem of testing the null hypothesis that a multivariate series of observations is a white noise (with unspecified density function) against alternatives of ARMA dependence. They study the asymptotic behavior of the proposed test statistics, both under the null and local alternatives hypotheses. These statistics lead to some asymptotically optimal tests against specified local alternatives of ARMA dependence. For the normally distributed errors it is shown that a van der Waerden type test is asymptotically equivalent to the classical test based classical autocovariances.