Complete \(q\)-order moment convergence of moving average processes under \(\varphi \)-mixing assumptions. (Q742901): Difference between revisions

The authors prove the complete \(q\)-order moment convergence of the maxima and suprema of partial sums of moving average processes based on sequences of \(\varphi \)-mixing random variables. In particular, it is assumed that \(\{Y_i, i\in \mathbb {Z}\}\) is a doubly infinite sequence of independent and identically distributed random variables such that the coefficients \[ \varphi (m)=\sup _{k\geq 1}\sup \{| {P}(B| A)-P(B)|: A\in \mathcal {F}_{-\infty }^k,\, P(A)\neq 0, \,B \in \mathcal {F}_{k+m}^{\infty }\} \] satisfy \(\varphi (m)\to 0\) as \(m\to \infty \), where \(\mathcal {F}_{n}^m=\sigma \{Y_i:n\leq i\leq m\}\) for any \(n\leq m\), \(n,m\in \mathbb {Z}\). Such variables are sometimes also referred to as uniformly mixing or \(\phi \)-mixing. Furthermore, let \(\{a_i\}_{i\in \mathbb {Z}}\) be an absolutely summable sequence of real numbers. The moving average process \(\{X_n:n\in \mathbb {N}\}\) is defined as \[ X_n=\sum _{i=-\infty }^{\infty } a_i Y_{i+n}\quad \text{for } n\geq 1. \] The authors provide complete \(q\)-order moment convergence for the maxima and suprema of the partial sums \(S_n=\sum _{k=1}^n X_k\) under suitable conditions. In particular, they give sequences \(\{a_n\}\) and \(\{b_n\}\) such that \[ \sum _{n=1}^{\infty } a_n E\{b_n^{-1} | Z_n| -\varepsilon \}_{+}^q<\infty \] for all \(\varepsilon >0\), where \(x_{+}=\max \{0,x\}\) and \(Z_n\) is either \(\max _{1\leq k\leq n} S_k\) or \(\sup _{k\geq n} S_k\). The results of the paper complement and extend some of the results previously published in the literature. They might find applications in sequential analysis of time series.