On equivalency of martingales and related problems (Q384559)

The paper deals with filtrations of the unit interval generated by families of nested subintervals and their product filtrations (indexed by \(\mathbf{n}=(n_i)_{i=1}^d\in\mathbb{N}^d\)) of the \(d\)-dimensional unit cube. The main results show that martingales with respect to two such filtrations are essentially the same, up to a measure-preserving transformation and passing to a subsequence, and outside an exceptional set \(E_{\mathbf{n}}\) that shrinks to a null set as \(\min n_i\to\infty\). If one of the filtrations is already a subsequence of the other, neither a further subsequence nor an exceptional set is needed. In particular, the a.e. convergence or maximal function estimates along a subsequence of such a filtration is not any ``easier'' than the corresponding questions for the entire filtration. This provides another approach to some results of \textit{A. M. Stokolos} [Colloq. Math. 104, No. 2, 311--315 (2006; Zbl 1085.42013)] and others on ``rare maximal functions''.