A CLT for weighted time-dependent uniform empirical processes (Q473519)

Let \(E\) be a parameter space and \(l^\infty({\mathcal F})\) be the class of bounded real-valued functions on \({\mathcal F}\). Put \[ {\mathcal C}= \{C_{s,y}: s\in E,\,y\in\mathbb{R}\}, \] where \(C_{s,y}= \{z\in D(E): Z(s)\leq y\}\) and \(D(E)\) is a collection of real-valued functions on \(E\) such that \(\operatorname{P}(Y(\cdot)\in D(E))= 1\). Identifying a point \((s, y)\in E\times\mathbb{R}\) with \(C_{s,y}\), one uses \(l^\infty(E\times\mathbb{R})\) instead of \(l^\infty(\mathcal{C})\). \textit{J. Kuelbs} et al. [Ann. Probab. 41, No. 2, 785--816 (2013; Zbl 1287.60034)] established sufficient conditions for the time dependent empirical process based on \[ \{I_{\{Y_t\leq y\}}- \operatorname{P}(Y_t\leq y): t\in E,\;y\in\mathbb{R}\} \] to satisfy the central limit theorem in \(l^\infty(E\times\mathbb{R})\). In this paper, referring to the above results, the author obtains, among other results, a sufficient condition for the weak convergence of time dependent weighted uniform empirical processes. More precisely, let \(\{X (t), X_1(t), X_2(t),\dots\}\) \((t\in E)\) be i.i.d.\ uniform processes, i.e., for each \(t\in E\), \(X(t)\) and the \(X_i(t)\)'s are uniformly distributed on \((0,1)\). Then, for some weight function \(w(x)> 0\) \((0< x< 1)\), the empirical process based on \[ \{w(x)(I_{\{X_t\leq x\}}-x): t\in E,\;x\in[0,1]\} \] converges weakly in \(l^\infty(E\times[0, 1])\). As an illustration of the main result, an example is given.