Two-sample tests based on empirical Hankel transforms

We study a new Cramér-von Mises type test for the general nonparametric two-sample problem on the nonnegative half-line. The test statistic is based on the empirical Hankel transforms of the sample variables, critical values are obtained by bootstrapping. The test is shown to be consistent against each fixed alternative. A scale invariant version of the test is also considered. A power comparison with the classical Cramér-von Mises test and another new Cramér-von Mises type test is done by simulation.