arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1904.08730math.STstat.OTstat.TH
收藏

服从指数化Gumbel-II分布分量的最高与最低次序统计量的若干序性质

Some ordering properties of highest and lowest order statistics with exponentiated Gumble type-II distributed components

IIT Kharagpur(印度理工学院哈格普尔分校)

Surojit Biswas

AI总结:

本文针对服从指数化Gumbel-II分布的独立随机变量,利用向量优化与多元链优化理论,建立了高低次序统计量的多种随机序关系,为联合参数异质性下的可靠性比较提供了通用方法。

摘要:

本文研究独立指数化Gumbel-II分布随机变量产生的最高与最低次序统计量的随机比较。首先,利用向量优化理论,针对最低和最高次序统计量关于外形状参数与内形状参数建立了全面的随机序关系。特别地,对于最低次序统计量,推导了通常随机序与似然比序的结果,这进一步蕴含了失效率序、平均剩余寿命序及增凸序。其次,证明最低次序统计量满足比例失效率结构,此外还建立了星序与分散序。对于最高次序统计量,得到了逆失效率序的结果,并证明一般情况下不满足比例失效率结构;而且,与若干经典指数化模型不同,在内形状参数异质性下,最高次序统计量在通常随机序下未必可比较。最后,利用多元链优化理论扩展分析,研究尺度参数与外形状参数同时变化的影响,在此框架下,针对最低和最高次序统计量建立了通常随机序结果,为联合参数异质性下提供了更通用的可靠性比较。

英文摘要:

This paper studies stochastic comparisons of the highest and lowest order statistics arising from independent Exponentiated Gumbel Type-II distributed random variables. First, using vector majorization, we establish a comprehensive set of stochastic orderings for the lowest and highest order statistics with respect to the outer and inner shape parameters. In particular, for the lowest order statistic, we derive results on the usual stochastic order and likelihood ratio order, which further imply failure rate, mean residual life, and increasing convex orderings. Second, we show that the lowest order statistics satisfy a proportional failure rate structure. In addition, we also establish star and dispersive orderings. For the highest order statistic, we obtain results in the reversed failure rate order, and we show that, in general, a proportional failure rate structure does not hold. Moreover, unlike several classical exponentiated models, the highest order statistics are not necessarily comparable in the usual stochastic order under heterogeneity of the inner shape parameters. Finally, we extend the analysis using multivariate chain majorization to study the effect of simultaneous variation in scale and outer shape parameters. Under this framework, we establish usual stochastic ordering results for both the lowest and highest order statistics. It provides more general reliability comparisons under joint parameter heterogeneity.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑