Message Board

Dear readers, authors and reviewers,you can add a message on this page. We will reply to you as soon as possible!

2020 Volume 45 Issue 9
Article Contents

Shu-li WU, Zuo-xiang PENG. Tail Behavior and Limiting Distribution of Extremes from α-Skew-Normal Distribution[J]. Journal of Southwest China Normal University(Natural Science Edition), 2020, 45(9): 19-22. doi: 10.13718/j.cnki.xsxb.2020.09.004
Citation: Shu-li WU, Zuo-xiang PENG. Tail Behavior and Limiting Distribution of Extremes from α-Skew-Normal Distribution[J]. Journal of Southwest China Normal University(Natural Science Edition), 2020, 45(9): 19-22. doi: 10.13718/j.cnki.xsxb.2020.09.004

Tail Behavior and Limiting Distribution of Extremes from α-Skew-Normal Distribution

More Information
  • Corresponding author: Zuo-xiang PENG
  • Received Date: 03/12/2019
    Available Online: 20/09/2020
  • MSC: O211

  • In this paper, Mills-type inequalities and Mills-type ratio of α-skew-normal distribution are considered, which are applied to derive the extreme value distribution and its convergence rate with suitable normalizing constants.
  • 加载中
  • [1] ELAL-OLIVERO D. Alpha-Skew-Normal Distribution [J]. Proyecciones (Antofagasta), 2010, 29(3): 224-240.

    Google Scholar

    [2] TARNOPOLSKI M. Analysis of Gamma-Ray Burst Duration Distribution Using Mixtures of Skewed Distributions [J]. Monthly Notices of the Royal Astronomical Society, 2016, 458(2): 2024-2031. doi: 10.1093/mnras/stw429

    CrossRef Google Scholar

    [3] GUI W H. A Generalization of the Slashed Distribution via Alpha Skew Normal Distribution [J]. Statistical Methods & Applications, 2014, 23(4): 547-563.

    Google Scholar

    [4] SHARAFI M, SAJJADNIA Z, BEHBOODIAN J. A New Generalization of Alpha-Skew-Normal Distribution [J]. Communications in Statistics - Theory and Methods, 2017, 46(12): 6098-6111. doi: 10.1080/03610926.2015.1117639

    CrossRef Google Scholar

    [5] 傅华, 彭作祥.偏正态逻辑斯蒂分布的尾部特征[J].西南大学学报(自然科学版), 2014, 36(3): 63-66.

    Google Scholar

    [6] LIAO X, PENG Z X, NADARAJAH S, et al. Rates of Convergence of Extremes from Skew-Normal Samples [J]. Statistics & Probability Letters, 2014, 84: 40-47.

    Google Scholar

    [7] LIAO X, PENG Z X, NADARAJAH S. Tail Properties and Asymptotic Expansions for the Maximum of the Logarithmic Skew-Normal Distribution [J]. Journal of Applied Probability, 2013, 50(3): 900-907. doi: 10.1239/jap/1378401246

    CrossRef Google Scholar

    [8] LEADBETTER M R, LINDGREN G, ROOTZÉN H. Extremes and Related Properties of Random Sequences and Processes[M]. Berlin:Springer Verlag, 1983.

    Google Scholar

  • 加载中
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Article Metrics

Article views(744) PDF downloads(92) Cited by(0)

Access History

Other Articles By Authors

Tail Behavior and Limiting Distribution of Extremes from α-Skew-Normal Distribution

    Corresponding author: Zuo-xiang PENG

Abstract: In this paper, Mills-type inequalities and Mills-type ratio of α-skew-normal distribution are considered, which are applied to derive the extreme value distribution and its convergence rate with suitable normalizing constants.

  • α -偏正态分布(ASN(α))[1]的密度函数fα(x)为

    分布函数记作Fα(x).当α=0时,ASN(0)即为标准正态分布,因此ASN(α)对有偏数据的拟合效果比正态分布更好.文献[2]利用α -偏正态分布与偏正态分布等分布的混合分布对伽马射线爆发持续时间的分布进行拟合.文献[3]将斜切分布中的标准正态分布改为α -偏正态分布从而得到新的斜切分布.文献[4]基于偏正态分布的表达式将α -偏正态分布推广到更一般的形式.

    在极值理论中,给定分布的序列的极值分布与分布尾的特征紧密相关.本文首先研究ASN(α)在α≠0时的尾部特征,包括Mills型不等式与Mills型比率.在此基础上,得到α -偏正态分布极值的极限分布,并建立相应的极值分布的收敛速度.相关研究参见文献[5-7].

1.   ASN(α)的Mills型不等式及比率
  • 本节给出ASN(α)的Mills型不等式、1-Fα的渐近展开、ASN(α)极值的极限分布与规范化常数,最后给出了极值分布的收敛速度.

    定理1   当α≠0且$ x>\sqrt{1+\sqrt{2}}$时,有以下不等式成立

      当x>0时,有$x\left(1-F_{a}(x)\right) <\int_{x}^{\infty} t f_{a}(t) \mathrm{d} t $.令函数

    可得g(t)最大值为$ {1 + \sqrt 2 }$,由分部积分得

    $ x>\sqrt{1+\sqrt{2}}$时,由式(3)得式(2)右端不等式.

    另一方面,因为$x^{-2}\left(1-F_{a}(x)\right)>\int_{x}^{\infty} t^{-2} f_{a}(t) \mathrm{d} t $,那么由分部积分可得

    g(t)最小值为$ 1 - \sqrt 2 $可得

    那么

    得式(2)左端不等式成立.定理证毕.

    对标准正态分布,当x充分大时,有$1-\mathit{\Phi}(x)=\frac{\varphi(x)}{x}\left(1+O\left(x^{-2}\right)\right) $,其中Φ(x),φ(x)分别为标准正态分布的分布函数与密度函数.对于ASN(α)有类似的结果.

    定理2  当x充分大时,有

      由分部积分得

    使用洛必达法则知,

    故对充分大的x,式(3)成立.定理证毕.

    定理1与定理2均可得下面的Mills型比率.

    推论1   对充分大的x,有

2.   ASN(α)极值的极限分布与收敛速度
  • 本节将给出ASN(α)极值的极限分布以及在相应规范化常数下极值分布的收敛速度.

    定理3   取规范化常数anbn

    $ F_{a}^{n}\left(a_{n} x+b_{n}\right) \rightarrow \mathit{\Lambda}(x)=\exp (-\exp (-x))$,并且有以下收敛速度

      令$u_{n}=\alpha_{n} x+\beta_{n} $,由文献[8]中定理1.5.1,可取$1-F_{a}\left(u_{n}\right)=\frac{1}{n} \exp (-x) $,则当$n \to \infty $${u_n} \to \infty $.由式(4)可得当n充分大时,有$ \frac{n}{u_{n}} f_{a}\left(u_{n}\right)\left(1+O\left(\left(u_{n}\right)^{-2}\right)\right)=\exp (-x)$,那么

    两边同时除以un2得到

    可得$u_{n}^{2} \sim 2 \log n $,进一步得$\frac{2 \log n}{u_{n}^{2}}=1+O\left(\frac{\log \log n}{\log n}\right) $.因此,

    将式(9),(10)代入式(7)中,计算可得

    那么

    由文献[8]定理1.5.1有$F_{a}^{n}\left(\alpha_{n} x+\beta_{n}\right) \rightarrow \mathit{\Lambda}(x) $,又因为$a_{n}^{-1} \alpha_{n}=1, a_{n}^{-1}\left(\beta_{n}-b_{n}\right) \rightarrow 0 $,根据文献[8]定理1.2.3可得

    下证式(6).令$u_{n}^{\prime}=a_{n} x+b_{n} $,那么由式(5)可得

    由式(4)可得,当n充分大时

    带入式(11),可得

    则对于$ \tau(x)=\exp (-x)$,有

    最后,由文献[8]定理2.4.2可得

    定理证毕.

Reference (8)

Catalog

    /

    DownLoad:  Full-Size Img  PowerPoint
    Return
    Return