期刊名称：Proceedings of the National Academy of Sciences
印刷版ISSN：0027-8424
电子版ISSN：1091-6490
出版年度：1976
卷号：73
期号：2
页码：286-288
DOI：10.1073/pnas.73.2.286
语种：English
出版社：The National Academy of Sciences of the United States of America
摘要：Let X1..., Xn have an arbitrary common marginal distribution function F, and let Mn = max(X1..., Xn). It is shown that EMn [≤] mn, where mn = an + n[unk]an{infty}[1 -F(x)]dx and = F-1(1 - n-1), and that EMn = mn when X1..., Xn are "maximally dependent"; i.e., P(Mn > x) = min{1, n[1 - F(x)]} for all x. Moreover, as n [->] {infty
