二阶矩有限时弱鞅和相依序列的不等式及强大数律-概率论与数理统计专业论文.docxVIP

摘要 概率极限理论的中心研究课题是随机变量序列的收敛性及随机变量和的强 大数定律，要想得到更好的性质通常的方法是运用概率不等式进行证明. 为了研究随机变量收敛性的问题，许多学者给出了经典的不等式，例如 Hájek 和 Rényi (1955) 给出一个重要的不等式，后人把此不等式一步步推广，归纳出一 个 Hájek- Rényi 型不等式，得到广泛的研究和应用.近期，Prakasa Rao (2000)用 Hájek–Rényi 不等式证明了相依序列的收敛性问题，Sung (2008)把 Hájek–Rényi 不等式进一步推广，并给出更加广泛的应用，其结果优于 Prakasa Rao (2000) 结 果. 但是从证明中我们发现当中出现了系数上的错误. Chow (1960) 证明了下鞅的一个极大值不等式，它包含了 Hájek–Rényi 不等 式. Christofides (2000) 证明了下鞅的极大值不等式可以推广到弱鞅的情形， Wang (2004)又证明了 PA 序列和更广的弱鞅序列的 Doob 不等式. 近期，Hu 等 (2009) 用另外的方法给出了一个 Hájek–Rényi 型不等式对于 PA 序列，此结果包 含了 Sung (2008) 的结论，并且给出收敛速度和上确界积分有限.另外指出 Sung (2008) 在证明中出现的一些问题. 通常情况下,大数定律通常是在 r 阶矩 (r ??1) 时给出的，而本文首先仅在 一阶矩有限 0 ??r ??1 时给出关于弱鞅的不等式同时还给出弱鞅和 PA 序列的强 大数律和强收敛速度.再从概率不等式出发,在二阶矩限制条件下,得到一个随机 变量序列的 Hájek–Rényi 型不等式,并应用此不等式证明随机变量序列部分和的 几乎处处收敛性,同时给出随机变量序列部分和的推广性质和收敛速度,可以证明 本文的结论优于 Sung (2008) 的主要结论. 应用到随机变量序列收敛性的证明,从 而推广了随机变量序列的一些收敛性质. 关键词：弱鞅; 相依序列; Hájek–Rényi 型不等式; 强大数律 Abstract The limit theorem is the central subject of the probability limit theory. We use probability inequality in order to obtain a lot of perfect results. In order to study random variable convergence inequality, a lot of inequalities has already been given by many scholars. Hájek and Rényi (1955) gave a important inequality named Hájek-Rényi inequality. Many scholars have improved them and obtained a class of Hájek-Rényi-type inequality and their application. Recently, in 2000, Prakasa Rao (2000) proved the associated random variables convergence by using the Hájek-Rényi-type inequality. Sung (2008) further generalized Hájek-Rényi-type inequality, gave a more extensive application which was better than the results of Prakasa Rao (2000). But there were several mistakes in the proof of main theorems. Chow (1960) have proved the maximal inequality for submartingale which contains Hájek-Rényi inequality. Christofides (2000) have given that maximal inequality for demimartingale and demisubmartingale, Wang (2004) have proved the Doob inequality for PA sequence and demimartingale.