In this paper we study the problem of measuring and testing joint independence for a collection of multivariate random variables. Using the emerging theory of optimal transport (OT) based multivariate ranks, we propose a distribution-free test for multivariate joint independence. Towards this we introduce the notion of rank joint distance covariance (RJdCov), the higher-order rank analogue of the celebrated distance covariance measure, that captures the dependencies among all the subsets of the variables. The RJdCov can be easily estimated from the data without any moment assumptions and the associated test for joint independence is universally consistent. We can calibrate the test without any knowledge of the (unknown) marginal distributions (due to the distribution-free property), both asymptotically and in finite samples. In addition to being distribution-free and universally consistent, the proposed test is also statistically efficient, that is, it has non-trivial asymptotic (Pitman) efficiency. We demonstrate this by computing the limiting local power of the test for both mixture alternatives and joint Konijn alternatives. We also use the RJdCov measure to develop a method for independent component analysis (ICA) that is easy to implement and robust to outliers and contamination. Extensive simulations are performed to illustrate the efficacy of the proposed test in comparison to other existing methods. Finally, we apply the proposed test to learn the higher-order dependence structure among different US industries based on stock prices.
翻译:在本文中,我们研究了为收集多种变式随机变量而衡量和测试联合独立的问题。使用基于最佳运输(OT)的多元变数排名的新理论,我们建议对多种变式联合独立进行无分配的测试。为此,我们引入了等级联合距离共差概念(RJdCov),即所庆祝的远距离共变差措施的高等级类比,以捕捉所有变量组群之间的依赖性。RJdCov可以在不作任何时刻假设的情况下从数据中轻易地估算出数据,而相关的联合独立独立测试则是普遍一致的。我们可以在对(由于无分配特性)的边际分布没有任何了解的情况下校准测试。我们还可以使用不易知知的(由于无分配特性)边际分配和有限样本中的边际分布测试。除了无分配和普遍一致的排序概念(RJCov),拟议的测试在统计上也非常高效,也就是说,它具有非边际的随机性(Pitman)效率。我们通过计算混合替代品和共同 Konijn替代品的当地测试力量的有限能力来证明这一点。我们也可以校准测试测试测试测试测试测试测试测试测试。我们还使用较易的RJC的系统,在进行这种结构中采用较强的测试方法。我们为最终的测试方法。我们采用的测试方法来进行测试方法,以便进行更易的测试方法来进行试验。我们的测试。