Hotelling two sample T2 test for complex normal distribution

A review of literature shows that only one sample test concerning the complex mean of multivariate complex normal (cn) distribution has been considered. however,in real application,many complex variables often appear as data from images or signals that need to be compared,for instance,for object identification. this paper therefore derived a hypothesis testing for two sample means from multivariate cn based on the hotelling t2 test by using the established relationship of multivariate cn to the multivariate real normal distribution. a simulation study was also carried out to investigate the performance of the test,evaluated by the type i error rate against equal means and power of test against non-equal means. results from the empirical study suggest that the test works well in differentiating the means with low type i error rate and good power properties. the interest in the test for complex normal model arose in connection with the study of discriminating the shape of the human maxillary dental arches; an application to such data is also presented. © 2015 pakistan journal of statistics.