Mean-bounded operators and mean ergodic theorems (Q1056879)

From MaRDI portal





scientific article; zbMATH DE number 3895691
Language Label Description Also known as
default for all languages
No label defined
    English
    Mean-bounded operators and mean ergodic theorems
    scientific article; zbMATH DE number 3895691

      Statements

      Mean-bounded operators and mean ergodic theorems (English)
      0 references
      1985
      0 references
      A linear operator T on a Banach space B is said to be ''mean-bounded'' if \(\sup_{n\geq 1}n^{-1}\| I+...+T^{n-1}\| <+\infty \). It was conjectured by A. Brunel that any mean-bounded positive operator on \(L_ p\) \((1<p<\infty)\) verifies the mean and the pointwise ergodic theorem, that is \(\lim_{n}n^{-1}(I+...+T^{n-1})f\) exists in norm and a.e. for any \(f\in L_ p.\) The norm convergence is not a consequence of the well-known mean ergodic theorems of Yosida-Kakutani, since it is not obvious that \(T^ n/n\to 0.\) We first prove that the resolent of T is well-defined and is strongly convergent if B is reflexive. Then, using a Tauberian theorem of J. Karamata and the crucial hypothesis T positive, we obtain the Cesàro convergence in norm. The pointwise convergence problem is still open. We only note that the resolvent is pointwise convergent on a dense subset of \(L_ p\).
      0 references
      mean-bounded positive operator
      0 references
      the mean and the pointwise ergodic theorem
      0 references
      resolent
      0 references
      Tauberian theorem
      0 references
      Cesàro convergence in norm
      0 references
      0 references

      Identifiers