A note on the uniform laws for dependent processes via coupling (Q5950024)

The author studies coupling of a \(\widetilde\varphi\)-mixing process \((X_n)_{n\geq 1}\) by a sequence of independent random variables with the same marginals. Let \({\mathcal F}_k^l\) denote the \(\sigma\)-field generated by the random variables \(X_k,\dots,X_l\) and define \[ \widetilde\varphi_k=\sup\Bigl\{ \biggl|P\bigl(B\mid A\bigr)-P(B) \biggr|:A\in {\mathcal F}_1^m,\;B\in{\mathcal F}^{m+k}_{m+k}, m\geq 1\Bigr\}. \] The process is called \(\widetilde\varphi\)-mixing, if \(\widetilde\varphi_k\to 0\) as \(k\in\infty\). The main coupling result states that a sequence of independent random variables \((\widetilde X_n)_{n\geq 1}\) with the same marginals exists in such a way that \[ P\left( \bigcap^m_{j=0} \{X_{i_j} \neq\widetilde X_{i_j}\} \right)\leq \widetilde \varphi^m_1 \] for all choices of indices \(1=i_0<i_1 <\cdots <i_m\). As an application, the author shows that a \(\widetilde\varphi\)-mixing sequence satisfies a uniform law of large numbers with respect to a class of functions \({\mathcal F}\) if and only if the same holds for a sequence of i.i.d. random variables with the same marginals.