Modulated and subsequential ergodic theorems in Hilbert and Banach spaces

The main results of the present paper are the following. Theorem 2.2. Let \(\{\alpha_{n,k}\}_{n>0,k\in \mathbb{Z}}\) be a matrix of complex numbers, such that \(\sum_{k=-\infty}^\infty| \alpha_{n,k}| <\infty\) for every \(n\). Then the following are equivalent. 1. The matrix satisfies the following two conditions: (a) for every \(\Gamma=\{\lambda:| \lambda| =1\}\), \(\lim_{n\to\infty} \sum_{k=-\infty}^\infty \alpha_{n,k}\lambda^k\) exists; (b) \(\sup_{n\geq1}\sup_{| \lambda| =1} | \sum_{k=- \infty }^\infty \alpha_{n,k}\lambda^k| <\infty\). 2. For every unitary operator \(U\) on a Hilbert space \(H\) and every \(x\in H\) the sequence \(\{\sum_{k=-\infty}^\infty \alpha_{n,k}U^kx\}\) converges in norm. 3. For every unitary operator \(U\) on a Hilbert space \(H\) and every \(x\in H\) the sequence \(\{\sum_{k=-\infty}^\infty\alpha_{n,k}U^kx\}\) converges weakly. Taking \(\alpha_{n,k}=a_k/n\) for \(n>0\) and \(0\leq k\leq n-1\) and 0 otherwise, the following corollary follows. Corollary 2.3. Let \(\{a_k\}_{k\geq0}\) be a sequence of complex numbers. Then the following are equivalent. 1.The sequence satisfies the following two conditions: (a) for every \(\lambda\in \Gamma\) there exists \(c(\lambda)\) such that \({1\over n}\sum_{k=0 }^{n-1}a_k\bar\lambda^k\to c(\lambda)\) \((n\to\infty)\); (b) \(\sup_{n \geq1}\sup_{| \lambda| =1} | {1\over n}\sum_{k= 0}^{n-1}a_k\lambda^k| <\infty\). 2. For every unitary operator \(U\) on a Hilbert space \(H\) and every \(x\in H\) the sequence \(\{ \sum_{k=-\infty}^\infty a_kU^kx\}\) converges in norm. 3. For every contraction \(T\) on a Hilbert space \(H\) and every \(x\in H\) the sequence \(\{\sum_{k=-\infty}^\infty a_kT^kx\}\) converges in norm. 4. For every unitary operator \(U\) on a Hilbert space \(H\) and every \(x\in H\) the sequence \(\{\sum_{k=-\infty}^\infty a_kU^kx\}\) converges weakly. When 1. holds, \(c(\lambda)\neq0\) for at most countably many \(\lambda\in\Gamma\), and the limit in 3. has the form \[ \lim_{n\to\infty} {1\over n}\sum_{k=0}^na_kT^kx =\sum_{\lambda\in \Gamma} c(\bar\lambda)E(\lambda,T)x, \] where \(E(\lambda,T)\) is the orthogonal projection on the eigenspace of \(T\) corresponding to the eigenvalue \(\lambda\), and the series converges strongly. In this situation, condition (c) \(\sup_n{1\over n}\sum_{k=0}^{n-1}| a_k| <\infty\) is stronger than the condition (b). The authors give several examples of sequences such as a sequence (Ex. 2.4) satisfying (c) but not (a), (Ex. 2.5) satisfying (a) and (b) but not (c), (Ex. 2.6) satisfying (a) but not (b). Theorem 3.1. Let \(\{\alpha_{n,k}\}\) be a matrix of non-negative numbers, auch that \(\sum_{k=0}^\infty\alpha_{n,k}=1\) for every \(n\). If \(\sum_{k=0}^\infty\alpha_{n,k}\bar\lambda^k\to0\) for every \(\lambda\neq1\) with \(| \lambda| =1\), then for every almost periodic operator \(T\) on a Banach space \(X\) and every contraction \(T\) on a Hilbert space, \(\lim_{n\to\infty} \sum_{k=0}^\infty \alpha_{n,k}T^kx=E(T)x\). From this theorem, the next proposition follows. Proposition 3.2. Let \(\{k_j\}\) be a strictly increasing sequence of positive integers satisfying \(\lim_{n\to\infty}n^{-1}\sum_{j=1}^n\lambda^{k_j}=0\) for every \(\lambda\neq1\) with \(| \lambda| =1\). 1. For every almost periodic operator \(T\) on a Banach space \(X\) and every contraction \(T\) on a Hilbert space, \[ \lim_{n\to\infty} {1\over n}\sum_{j=1}^nT^{k_j}x=E(T)x.\tag{3.2} \] 2. If \(\{k_j\}\) has positive lower density, then (3.2) holds for every weakly almost periodic operator \(T\) on a Banach space \(X\). 3. If \(\{k_j\}\) is thin, then there exists a weakly almost periodic operator \(T\) for which (3.2) does not hold. The authors then give a negative answer to the following question (Proposition 3.3): If \(\{\mu_n\}\) satisfies \(\mu_nf\to M(f)\) pointwise for every almost periodic function \(f\), does then \(\mu_nf\to M(f)\) hold pointwise for every weakly almost periodic functions? In the last section, the results in \S2 are generalized to unitary representations of \(\sigma\)-compact locally compact berian groups (Proposition 4.1 and Theorem 4.2).

Recommendations

Cited in

(20)

This page was built for publication: Modulated and subsequential ergodic theorems in Hilbert and Banach spaces

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4806357)