張 玉,沈愛婷
(安徽大學 數(shù)學科學學院,安徽 合肥 230601)
?
NSD隨機變量加權和的強收斂性質(zhì)
張 玉,沈愛婷
(安徽大學 數(shù)學科學學院,安徽 合肥 230601)
負超可加相依(negatively superadditive dependent,NSD)隨機變量是一類包含獨立隨機變量和負相協(xié)(negatively associated,NA)隨機變量在內(nèi)的非常廣泛的相依變量。文章利用NSD隨機變量的三級數(shù)定理和隨機變量的截尾技術,在較弱的條件下建立了NSD隨機變量加權和的若干強收斂性質(zhì)。所得結果推廣了獨立隨機變量和NA隨機變量的相應結果。
加權和;NSD隨機變量;收斂性質(zhì);NSD三級數(shù)定理;隨機變量截尾
定義1[1]稱隨機變量{Xi,1≤i≤n}為負相協(xié)(NA)的,若對任意2個非空不交子集A,B?{1,2,…,n}均有Cov (f1(Xi,i∈A) ,f2(Xj,j∈B))≤0,其中fi(i=1,2)是使此式有意義且對各變元非降(或同時對各變元非升)的函數(shù)。稱隨機變量序列{Xn,n≥1}是NA的,如果其每個有限子集是NA的。
文獻[2-7]系統(tǒng)研究了NA隨機變量的概率極限性質(zhì)及其應用,文獻[8]提出了一類比NA隨機變量范圍更為廣泛的相依隨機變量——負超可加相依(negatively superadditive dependent,NSD)隨機變量,此概念是建立在超可加函數(shù)的基礎上。
定義2[2]稱函數(shù)φ:Rn→R是超可加的,如果x,y∈Rn滿足:
其中,x∨y表示對各分量取大者;x∧y表示對各分量取小者。
定義3[8]稱隨機向量X=(X1,X2,…,Xn)為NSD,如果滿足:
文獻[8]舉例說明了NSD不能推出NA,文獻[9]指出NA是NSD的。文獻[10-11] 研究了NSD隨機變量的概率極限性質(zhì)以及統(tǒng)計大樣本性質(zhì)。本文在已有結果的基礎上,利用NSD隨機變量的三級數(shù)定理,進一步研究NSD隨機變量的若干強收斂性。
引理2[8](NSD隨機變量的基本性質(zhì)) 如果隨機變量(X1,X2,…,Xn)是NSD的,且g1,g2,…,gn都是非降函數(shù),則(g1(X1),g2(X2),…,gn(Xn))也是NSD的。
(1)
由于{gn(t),n≥1}為非負函數(shù)列,且對每個n≥1,gn(t)單調(diào)不減,結合(1)式有:
(2)
(3)
(4)
由Cr不等式、gn(t)單調(diào)不減、(4)式以及(1)式得:
(5)
因此由(5)式得:
(6)
定理2 設{Xn,n≥1}是NSD隨機變量序列,{an,n≥1}是正常數(shù)序列,{gn(t),n≥1}是非負函數(shù)列,且對每個n≥1,gn(t)單調(diào)不減。假定存在δ>0, 使得gn(t)≥δt,0 (7) (8) 再由gn(t)≥δt,0 (9) 由于gn(t)≥δt,0 (10) 由Cr不等式和(10)式得: (11) 由(11)式和(7)式得: (12) 推論2 設{Xn,n≥1}是NSD隨機變量序列,{an,n≥1}是正常數(shù)序列。若存在常數(shù)0<β≤1,使得: 定理3 設{Xn,n≥1}是NSD隨機變量序列,{an,n≥1}是正常數(shù)序列,{gn(t),n≥1}是非負函數(shù)列,且存在常數(shù)β≥1 使得gn(t)≥δtβ,t>0。如果 (13) 證明 由(13)式和引理4可得: (14) 由于gn(t)≥δtβ,故由(14)式得: (15) 根據(jù)(13)式和(15)式,以及gn(t)≥δtβ得: (16) 再由Cr不等式及gn(t)≥δtβ得: 由此及(13)式、(14)式得: (17) [1] KHURSHEED A, LAI SAXENA K M.Positive dependence in multivariate distributions[J].Communications in Statistics:Theory and Methods,1981,10(12):1183-1196. [2] 凌能祥,杜雪樵.NA樣本下單邊截斷型分布族位置參數(shù)的經(jīng)驗Bayes估計[J].合肥工業(yè)大學學報(自然科學版),2002,25(5):743-747. [3] 王小明.NA序列部分和的完全收斂性[J].應用數(shù)學學報,1999,22(3):407-412. [4] SHAO Q M.A comparison theorem on moment inequalities between negatively associated and independent random variables[J].Journal of Theoretical Probability,2000,13(2):343-356. [5] 劉立新,吳榮.NA隨機變量序列的強大數(shù)律和完全收斂[J].應用概率統(tǒng)計,2001,17(3):315-320. [6] 吳永鋒.關于NA序列的強大數(shù)定律[J].合肥工業(yè)大學學報(自然科學版),2008,31(5):825-827. [7] 張世兵,杜雪樵.負相協(xié)重尾隨機變量加權和的尾概率等價關系[J].合肥工業(yè)大學學報(自然科學版),2009,32(9):1454-1456. [8] HU T Z.Negatively superadditive dependence of random variables with applications[J].Chinese Journal of Applied Probability and Statistics,2000,16(2):133-144. [9] CHRISTOFIDES T C,VAGGELATOU E.A connection between supermodular ordering and positive/negative association[J].Journal of Multivariate Analysis,2004,88:138-151. [10] WANG X J,DENG X,ZHENG L L,et al.Complete convergence for arrays of rowwise negatively superadditive dependent random variables and its applications[J].Statistic:A Journal of Theoretical and Applied Statistics,2014,48(4):834-850. [11] SHEN Y,WANG X J,YANG W Z,et al.Almost sure convergence theorem and strong stability for weighted sums of NSD random variables[J].Acta Mathematica Sinica:English Series,2013,29(4):743-756. [12] WANG X J,HU S H,SHEN Y,et al.Some new results for weakly dependent random variable sequences [J].Chinese Journal of Applied Probability and Statistics,2010,26(6):637-648. (責任編輯 張淑艷) Strong convergence properties of weighted sums of negatively superadditive dependent random variables ZHANG Yu,SHEN Aiting (School of Mathematical Sciences, Anhui University, Hefei 230601, China) Negatively superadditive dependent(NSD) random variables include independent random variables and negatively associated(NA) random variables as special cases. In this paper, by using the three series theorem of NSD random variables and the truncated method of random variables, some strong convergence properties of weighted sums of NSD random variables are established under some much weaker conditions. The results obtained in the paper generalize the corresponding ones of independent random variables and NA random variables. weighted sum; negatively superadditive dependent(NSD) random variable; convergence property; three series theorem of NSD; truncated method of random variable 2015-05-19; 2015-07-31 安徽省自然科學基金資助項目(1308085QA03) 張 玉(1989-),男,安徽合肥人,安徽大學碩士生; 沈愛婷(1979-),女,安徽合肥人,博士,安徽大學副教授,碩士生導師. 10.3969/j.issn.1003-5060.2016.10.028 O211.4 A 1003-5060(2016)10-1437-04