二元极值分布的一个性质

A Property for Bivariate Extreme Value Distribution

本文考虑二元极值的相关结构,通过一个变量变换,使变换后的变量基本上是独立的,并给出了它们的随机表示.由此能非常容易地在计算机上产生二元极值分布伪随机向量,以及计算一类常用统计量的数字特征,这是研究某些统计量渐近分布的基础.

The dependence structure of bivariate extreme value distribution is considered. A transform of variables is proposed, such that transformed variates are independent basically. Also we obtain the stochastic representation for bivariate extremes. From these it is easy to generate pseudo random vector of bivariate extreme distributed in computer and to calculate numerical characteristic of a class of usual statistics. These are basics to study asymptotic distribution of some statistics.