Rosenthal-type inequalities for the maximum of partial sums of stationary processes and examples (Q1951694)

The aim of the paper is to prove new Rosenthal-type inequalities for stationary sequences under dependence assumptions that are more general than the various standard ones. The main feature is that the introduction of the projection of the partial sums to zero instead of sums of ``immediate'' projections; that is, weighted sums of terms \(\|E_0(S_k^2)\|_{(p/2)}\) instead of sums of terms such as \(E_{k-1}(X_k^2)\). The point is that \(\|E_0(S_n^2)\|_{(p/2)}\), frequently is reasonably close to the variance of the partial sums. An essential ingredient in the proofs is a new Doob-type maximal inequality for martingales. Various applications are given.