Complete \(f\)-moment convergence for Sung's type weighted sums and its application to the EV regression models (Q2066523)

Let \(\left\{ S_{n},n\geq 1\right\} \) denote a sequence of random variables, \( \left\{ c_{n},n\geq 1\right\} \) a sequence of positive numbers, and \(f\) an increasing function with \(f(0)=0\). The sequence \(\left\{ S_{n},n\geq 1\right\} \) converges \(f\)-moment completely, if \(\sum_{n=1}^{\infty }c_{n}Ef(\left\{ \left\vert S_{n}\right\vert -\varepsilon \right\} _{+})<\infty \) for all \(\varepsilon >0\). Special cases are \(f(t)=t^{q}\), \(q>0\) with \(c_{n}=1\). In the main results, the authors consider a sequence of the form \( S_{n}=\max_{1\leq k\leq n}\left\vert \sum_{i=1}^{k}a_{ni}X_{i}\right\vert /n^{\alpha }\) and \(c_{n}=n^{\alpha p-2}\) and state conditions on the \(X_{i}\) and \(a_{ni}\) and the parameters \(\alpha ,p\) to ensure that \(\left\{ S_{n},n\geq 1\right\} \) converges \(f\)-moment completely. In the second part of the paper the authors apply the results in the simple linear errors-in-variables model and consider both a theoretical result and a numerical simulation.