随机变量的几种收敛及其相互关系1论文随机变量的几种收敛及其相互关系1摘要概率是对大量随机现象的考察中显现出来的,而对于大量的随机现象的描述就要采用极限的方法。概率统计中的极限定理研究的是随机变量序列的某种收敛性,对随机变量收敛性不同定义将导致不同的极限定理,而随机变量的收敛性的确可以有各种不同的定义。主要讨论了依概率收敛与依分布收敛,r阶收敛与几乎处处收敛,几乎处处收敛与依概率收敛之间的关系。给出了由依概率收敛推出几乎处处收敛的条件和由依概率收敛推出r阶收敛的条件,从而比较完全地说明了随机变量序列的各种收敛性之间的关系。本论文将对随机变量的几种收敛作出较为简单扼要的介绍和讨论.论文结构如下:一、随机变量的几种收敛的概念理论;二、随机变量的几种收敛之间的关系;从以上几个方面对随机变量的几种收敛理论简明扼要地分析,说明随机变量序列收敛理论在实际问题中的应用范围之广,在实际生活中的重要性。关键词:r阶收敛;几乎处处收敛;依概率收敛;依分布收敛。随机变量的几种收敛及其相互关系2AbstractTheProbabilityisthestudyofalargenumberofrandomphenomenaemerge,butforalargenumberofrandomphenomenashoulduseextrememethodsdescribed.Probabilityandstatisticsinthelimittheoremisasequenceofrandomvariablesconvergence,convergenceofrandomvariableswithdifferentdefinitionsleadtodifferentlimittheorem,andindeedtheconvergenceofrandomvariablescanhavedifferentdefinitions.Mainlydiscussedconvergenceinprobabilityandconvergenceindistribution,convergenceinorderrandalmosteverywhereconvergence,almostsureconvergenceandconvergenceinprobabilityrelationship.Convergenceinprobabilityisgivenbythelaunchofalmosteverywhereconvergenceofconditionsandtheconvergenceinprobabilitybytheintroductionofr-orderconvergenceconditions,whichmorecompletelydescribesthevariousrandomvariablesconvergencerelationship.Thispaperwillmaketheconvergenceofseveralrandomvariablesismorebriefpresentationsanddiscussions.Paperisstructuredasfollows:1.Convergenceofrandomvariablestheconceptoftheory;2.theconvergenceofseveralrandomvariablesbetween;Fromtheaboveaspectsofthetheoryofrandomvariablesofseveralbriefanalysisofconvergenceshowsthattheconvergencetheoryofrandomvariablesintheactualproblemsinthewiderangeofapplications,inreallifeimportance.Keywords:convergenceinorderr;almosteverywhereoralmostsurely;convergenceinprobability;convergenceindistribution.随机变量的几种收敛及其相互关系3目录引言:····························································································41几种收敛性定义··············································································42依概率收敛与依分布收敛的关系························································53r阶收敛与几乎处处收敛的关系························································114依概率收敛与r阶收敛的关系··························································135几乎处处收敛与依概率收敛和依分布收敛的关系··································17总结·······························································································19四种收敛性······················································································19四种收敛蕴涵关系·············································································19致谢·····························································································21参考文献·························································································22随机变量的几种收敛及其相互关系4引言:概率论最早产生于17世纪,本来是保险事业的发展而产生的,但是来自于赌博者的请求,却是数学家们思考概率论中问题的源泉。然而其公理体系只在20世纪的20至30年代才建立起来并得到迅速发展,在过去的半个世纪里概率论在越来越多的新兴领域显示了它的应用性和实用性。概率论是根据大量同类随机现象的统计规律,对随机现象出现某一结果的可能性作出一种客观的科学判断,对这种出现的可能性大小做出数量上的描述;比较这些可能性的大小、研究它们之间的联系,从而形成一整套数学理论和方法。特别值得一提的是,概率论是今天数理统计的基础,其结果被用做问卷调查的分析资料或者对经济前景进行预测。概率论中的重要概念——概率的收敛性,寻找概率收敛中的随机变量序列收敛性的相互性质以及收敛性之间的相互关系,弄清楚它们之间的关系在理论和应用上都是很有意义的。1几种收敛性定义定义1.1(r阶收敛)设对随机变量nX,及X有||,||rrnEXEX,其中0r为常数,如果lim0rnnEXX则称{nX}r阶收敛于X,并记为rnXX.当2p是,2lim0nnEXX,称{,1}nXn均方收敛到X。记为..msnXX.例1.1设{,1kXkn}相互独立,且满足1(1)nPXn,1(0)(1)nnPXnn,()0X。则21(0)0nEXn,故2lim00nnEX,即..0msnX.定义1.2(几乎处处收敛)如果(lim)1nnPXX则称{nX}以概率1收敛于X,又称{nX}必乎处处收敛于X,并记为..asnXX.随机变量的几种收敛及其相互关系5例1.2设{nX,1n},,XY是定义在[0,1]上博雷尔概率空间(,,)FP=([0,1],[0,1],)FP上的随机变量,满足:[0,1],()1Y。而()1X,若B={[0,1]上理点};()0X,若B={[0,1]上有理点全体}。而()1X,若1,12;()0nX,若10,2n。则易知(:()())()0PXYPB。(:lim()())nnXY;(:lim()())nnXXB,但1B,故..asnXX。定义1.3(依分布收敛)设随机变量nX,X的分布函数分别为()nFx及()Fx。若对()nFx的每个连续点x有lim()(),nnFxFx则称{nX}依分布函数收敛于X(()nFx弱收敛到()Fx)。记为LnXX,或者()()WnFxFx。例1.3nZ,nY的记号同林德伯格-莱维(Lindeberg-Levy)定理,令Z~2(0,1)N,则LnZZ,即xR,有lim()()nnPZxx。定义1.4(依概率收敛)如果对于任意ε0,lim(||)0nnPXX则称{Xn}依概率收敛于X,并记为PnXX或limnnpXX.例1.4设{,1kYkn}独立同分布,且1Y~[0,1]U,令1/nnkkXYn,则由大数定律可知1()2PnXn.2依概率收敛与依分布收敛的关系随机变量序列依概率收敛和依分布收敛是概率论中两种较重要的收敛形式,弄清楚它们之间的关系是本节要讨论的.本节约定所涉及定义1.3,定义1.4。定理2.1若随机变量序列{}nX依概率收敛于某随机变量X,则{}nX依分布收敛于X.但定理2.1的逆不成立。证明设xx,则随机变量的几种收敛及其相互关系6{nXx}={nXx,Xx}{nXx,Xx}{}{,}nnnXxXxXx从而()()(,)nnnFxFxPXxXx设PnXX,则(,)(||)0nnnPXxXxPXXxx因而有()lim()nnnFxFx同理可证,对xx,有lim()()nnnFxFx所以对xxx,有()lim()lim()()nnnnnnFxFxFxFx如果x是()Fx的连续点,则令x,x趋于x,得()lim()nnFxFx即LnXX.反之不然,例如,若样本空间12{,},12()()1/2PP,定义随机变量()X如下:12()1,()1XX,则()X的分布律为(())1/2PXk,1k,1,如果对一切n,令()()nXX,则显然()()LnXX。但是对于任意的02,(|()()|)()1nPXXP所以{()nX}不依概率收敛于()X。但是在特殊场合有下面结果:对于常数C,则PnXC与LnXC等价。事实上,若LnXC,则0,(||)()()nnnPXCPXCPXC1(0)()nnFCFC随机变量的几种收敛及其相互关系71100从而PnXC。反之,若PnXC,则由定理2.1得LnXC。例2.1设12,,,XXX为独立同分布的随机变量,公共的分布列为(0)(1)1/2.PXPX显然:(n=1,2,)nX与X的分布函数相同,故{nX}依分布收敛X.但对于任意0E1和0R12,对一切n,有()(1,0)(0,1)nnnPXXEPXXPXX(1)(0)(0,1)