Uniform convergence on subspaces in the von Neumann ergodic theorem with discrete time (Q6114302)

Let \(\mathcal{H}\) be a Hilbert space, and let \(U : \mathcal{H}\to\mathcal{H}\) be a unitary operator. For each vector \(f\in\mathcal{H}\) and all \(n,m\in \mathbb{Z}\) such that \(n > m\), the following expression is called ergodic mean: \[ P_{n,m}f=\sum _{k=m}^{n-1} U^kf. \] Let \(P\) be the orthogonal projection operator on the subspace \(\mathcal{F}_U\) of fixed vectors of the operator \(U\). The von Neumann ergodic theorem asserts the existence of the limit \[\lim\limits_{n-m\rightarrow\infty } P_{n,m}f=f^{*}=Pf\] for each vector \(f\in\mathcal{H}\) in norm of the space \(\mathcal{H}\). \textit{A. G. Kachurovskii} et al. [Sib. Elect. Math. Rep. 20, No. 1, 183--206 (2023)] established conditions that lead to the uniform power-law convergence within vector subspaces, each equipped with its own norm and gave a complete description of all such subspaces for all possible exponents. In the present paper, inspired by the above results, the authors present similar criteria for uniform power-law convergence on subspaces in the case of discrete time, and give a complete description of all such subspaces. The authors provide a spectral condition that ensures uniform convergence with a power-law behavior characterized by an exponent \(\alpha\) in the range of \([0, 2)\) on one-dimensional subspaces within the space \(\mathcal{H}\). Next, the authors present the spectral criterion that indicates the existence of a maximum rate on one-dimensional subspaces. The authors extend the criteria obtained on one-dimensional subspaces to the general case of multidimensional vector subspaces in \(\mathcal{H}\) with their own norms. Also, they describe subspaces having power-law uniform convergence.