陈敏琼. 关于能量距离的两点注记[J]. 应用概率统计, 2018, 34(5): 463-474. DOI: 10.3969/j.issn.1001-4268.2018.05.003
引用本文: 陈敏琼. 关于能量距离的两点注记[J]. 应用概率统计, 2018, 34(5): 463-474. DOI: 10.3969/j.issn.1001-4268.2018.05.003
CHEN Minqiong. Two Notes on Energy Distance[J]. Chinese Journal of Applied Probability and Statistics, 2018, 34(5): 463-474. DOI: 10.3969/j.issn.1001-4268.2018.05.003
Citation: CHEN Minqiong. Two Notes on Energy Distance[J]. Chinese Journal of Applied Probability and Statistics, 2018, 34(5): 463-474. DOI: 10.3969/j.issn.1001-4268.2018.05.003

关于能量距离的两点注记

Two Notes on Energy Distance

  • 摘要: 本文讨论了能量距离的两个问题.类似Brownian协方差的讨论提出了Brownian距离的定义,并证明了Brownian距离与能量距离的一致性. 给出了配对变量的能量距离的表示,并探讨了将能量距离用于配对样本同分布的检验问题时原假设下的渐近分布理论.最后通过一个简单的数值模拟说明基于能量距离的配对样本的分布差异的检验方法比传统的t检验及Wilcoxon符号秩检验更有效.

     

    Abstract: The definition of Brownian distance is presented and it's proved that Brownian distance coincides with the energy distance with respect to Brownian motion. Energy distance for dependent random vectors is also given and the asymptotic distribution is derived under null hypothesis. A simple numerical simulation result shows that the method for paired-sample test based on energy distance can distinguish the distributions of the paired variables more effectively than the classical t-test and Wilcoxon signed rank test.

     

/

返回文章
返回