资源描述:
《Statistical Inequality英文学习材料》由会员上传分享,免费在线阅读,更多相关内容在教育资源-天天文库。
1、LectureNotes21ProbabilityInequalitiesInequalitiesareusefulforboundingquantitiesthatmightotherwisebehardtocompute.Theywillalsobeusedinthetheoryofconvergence.Theorem1(TheGaussianTailInequality)LetXN(0;1).Then2e 2=2P(jXj>):IfX1;:::;XnN(0;1)then1 n2=2P(jXnj>)pe:nProof
2、.ThedensityofXis(x)=(2) 1=2e x2=2.Hence,Z1Z11P(X>)=(s)dss(s)dsZ1 2=210()e= (s)ds=:Bysymmetry,2e 2=2P(jXj>):NowletX;:::;XN(0;1).ThenX=n 1PnXN(0;1=n).Thus,X=dn 1=2Z1nni=1inwhereZN(0;1)andp12 1=2p n=2P(jXnj>)=P(njZj>)=P(jZj>n)e:n1Theorem2(Markov
3、'sinequality)LetXbeanon-negativerandomvariableandsupposethatE(X)exists.Foranyt>0,E(X)P(X>t):(1)tProof.SinceX>0,Z1ZtZ1E(X)=xp(x)dx=xp(x)dx+xp(x)dxZ00Zt11xp(x)dxtp(x)dx=tP(X>t):ttTheorem3(Chebyshev'sinequality)Let=E(X)and2=Var(X).Then,21P(jX jt)andP(jZjk)(2)t2k2wh
4、ereZ=(X )=.Inparticular,P(jZj>2)1=4andP(jZj>3)1=9.Proof.WeuseMarkov'sinequalitytoconcludethatE(X )2222P(jX jt)=P(jX jt)=:t2t2Thesecondpartfollowsbysettingt=k. 1PnIfX1;:::;XnBernoulli(p)thenandXn=ni=1XiThen,Var(Xn)=Var(X1)=n=p(1 p)=nandVar(Xn)p(1 p)1P(jXn pj>)
5、=2n24n2sincep(1 p)1forallp.42Hoeding'sInequalityHoeding'sinequalityissimilarinspirittoMarkov'sinequalitybutitisasharperinequality.Webeginwiththefollowingimportantresult.Lemma4SuppposethatE(X)=0andthataXb.ThenE(etX)et2(b a)2=8:2Recallthatafunctiongisconvexifforeac
6、hx;yandeach2[0;1],g(x+(1 )y)g(x)+(1 )g(y):Proof.SinceaXb,wecanwriteXasaconvexcombinationofaandb,namely,X=b+(1 )awhere=(X a)=(b a).Bytheconvexityofthefunctiony!etywehavetXtbtaX atbb Xtaee+(1 )e=e+e:b ab aTakeexpectationsofbothsidesandusethefactthatE(X)=0togettXatbbta
7、g(u)Ee e+e=e(3)b ab awhereu=t(b a),g(u)=
u+log(1 +
eu)and= a=(b a).Notethat000g(0)=g(0)=0.Also,g(u)1=4forallu>0.ByTaylor'stheorem,thereisa2(0;u)suchthatu2u2u2t2(b a)200000g(u)=g(0)+ug(0)+g()=g()=:2288Hence,EetXeg(u)et2(b a)2=8:Next,weneedtouseCherno'smethod.Lemma
8、5LetXbearandomvariable.Then ttXP(X>)infeE(e):t0Proof.Foranyt>0,XtXt ttXP(X>