节点文献
高斯向量序列的几乎处处中心极限定理
Almost Surely Central Limit Theorem on the Maximum of Gaussian Vector Sequence
【作者】 陈志成;
【导师】 彭作祥;
【作者基本信息】 西南大学 , 概率论与数理统计, 2007, 硕士
【摘要】 本文主要分为三部分,第一部分给出了多维随机变量最大值的几乎处处中心极限定理.主要结论如下:定理A设{Xi}i=1∞是标准化的d维非平稳高斯随机变量序列,若满足(2.1)和(2.2),对实数向量序列uni={uni(p),p=1,…,d},i=1,2,…,n当n→∞时,有(?)且λn(p)≥c(log n)1/2,c>0,则其中λn=(λn(1),…,λn(d)),λn(p)=(?)uni(p),p=1,…,d.定理B设X1,X2,…是标准化的d维非平稳高斯随机变量序列,若满足(2.1),(2.2),(2.3)和n(1-Φ(λn(p)))有界,则定理C设X1,X2,…是标准化的d维非平稳高斯随机变量序列,若满足(2.1),(2.2)且当n→∞时,n(1-Φ(λn(p)))→Υp,p=1,…,d,则定理D设X1,X2,…是标准化的d维非平稳高斯随机变量序列,若满足(?);存在(?),当n→∞时,(2.5),(2.9)成立,且当0≤Υp<∞,n(1-Φ(un(p)))→Υp,p=1,…,d.则定理E设{Yn}n=1∞为d维高斯随机变量序列,其中Y=Xn+(?)n,{Xn}n=1∞满足定理2.1.1,(?)n,mn*分别满足(2.6)和(2.7),若存在c>0,有λn(p)≥c(log n)1/2,则其中Mn*=(Mn*(1),…,Mn*(d)),Mn*(p)=(?) Yi(p),p=1,…,d.bn*=bnId,Id=(?),x=(x1,…,xd).定理F设{Yn}n=1∞为d维随机变量序列,其中Yn=Xn+(?)n,{Xn}n=1∞满足定理2.1.2,mn和mn*分别满足(2.6)和(2.7)且存在D>0,有(2.8)成立,则文章第二部分讨论了高斯随机变量最大与最小值的几乎处处收敛性,结论如下:定理G设X1,X2,…是标准化的d维非平稳高斯随机变量序列,若满足;存在γ>2(1+(?))/1-(?),当n→∞时,(2.5),(2.9)成立.当0≤Υp,ηp<∞,n(1-Φ(υn(p)))→Υp,nΦ(υn(p))→ηp,p=1,…,d,则文章第三部分主要给出了独立同分布随机变量最大值的几乎处处局部中心极限定理,结论如下:定理H设{Xn}n=1∞是独立同分布随机变量序列,EX1=0,实数列{un},{vn}满足vn<bn<un,n(1-F(vn))有界且F(un)-F(vn)>C/nlogεn,则其中Pk=P(υk≤(?)k<uk),(?)k=(?)Xi.
【Abstract】 This thesis is composed of almost surely central limit theorem on the maxima of weak dependent nonstationary Gaussian vector sequence under some conditions, and the almost surely local central limit theorems of the maximum of independent and identically distributed random variables. The main results are:Theorem A Suppsc X1,X2,…be standardized nonstationary Gaussian d dimensional random vectors satisfying (2.1) and (2.2). Let be constants such that, and let Uni(P),P = 1,…,d such thatλn(p)≥c(logn)1/2 for some c>0.ThenTheorem B Suppse X1, X2,…be standardized nonstationary Gaussian d dimensional random vectors satisfying (2.1), (2.2) and (2.3). Letλn(p)= Uni(p) be constants such that n(1 -Φ(λn(p))) is bounded. ThenTheorem C Suppse X1, X2,…be standardized nonstationary Gaussian d dimensional random vectors satisfying (2.1) and (2.2). Letλn(p) be constants such that n(1 -Φ(λn(p)))→τp as n→∞for someτ-≥O. ThenThorem D Suppse X1,X2,…be standardized nonstationary Gaussian d dimensionalrandomveetorswithδ= ma 1, and (2.5),(2.9) hold for some n→∞for someτp≥0, p = 1,…, d, thenTheorem E Suppse X1, X2,.. be d dimensional Gaussian random vectors with Yn = Xn+mn where satisfy Theorem 2.1.1 and mn, satisfy (2.6) and (2.7) respectively. Ifλn(p)≥c(logn)1/2 for some c>0, then whereTheorem F Suppse X1, X2,…be d dimensions random variables with Yn = Xn + mn where satisfy Theorem 2. 1. 2 and mn和satisfy (2. 6) and (2. 7) respectively. If (2. 8) holds for some D>0, then Theorem G Suppse X1, X2,…be standardized d dimensional nonstationary Gaussian random vectors withδ=1, and (2.5),(2.9) hold for some as for some O≤τp,ηp<∞, thenTheorem H Let X1, X2,…be independent identically distributed random variables with EXi = 0, i = 1,2,….{un}, {un}. If F(un) - F(un)>andn(1 - F(un)) is bounded as Un<bn<un, thcnwhere
【Key words】 Gaussian vector sequence; extreme distribution; almost surely central limit theorem; local central limit theorem;
- 【网络出版投稿人】 西南大学 【网络出版年期】2007年 06期
- 【分类号】O211.4
- 【下载频次】76