A uniform central limit theorem for set-indexed partial-sum processes with finite variance (Q1077805)

Let \(\{X_ j:\) \(j\in {\mathbb{Z}}^ d_+\}\) be an array of i.i.d. real valued random variables (r.v.) indexed by the d-dimensional positive integer lattice with \(EX_ j=0\), \(EX^ 2_ j=1\). For any bounded \(B\in {\mathfrak B}^ d\), the class of all Borel sets in \(R^ d\), define \(S(B)=\sum_{j\in B}X_ j\). Let \({\mathfrak A}\) be a family of Borel subsets of the unit cube \(I^ d=[0,1]^ d\) and define the normalized partial-sum process \(S_ n=\{S_ n(A):\quad A\in {\mathfrak A}\}\) by \(S_ n(A)=n^{-d/2}S(nA)\) where \(nA=\{nx:\quad x\in A\}.\) If for the metric \(\rho (A,B)=| A\Delta B|\) the metric entropy with inclusion \(N_ 1(\epsilon,{\mathfrak A},\rho)\) satisfies \[ \int^{1}_{0}(\epsilon^{-1}\log N_ 1(\epsilon,{\mathfrak A},\rho))^{1/2}d\epsilon <\infty, \] then an appropriately smoothed version of the partial-sum process converges weakly to the Brownian process indexed by \({\mathfrak A}\).