On the maximal ergodic theorem for certain subsets of the integers (Q1101568)

From MaRDI portal
scientific article
Language Label Description Also known as
English
On the maximal ergodic theorem for certain subsets of the integers
scientific article

    Statements

    On the maximal ergodic theorem for certain subsets of the integers (English)
    0 references
    0 references
    1988
    0 references
    Giving an affirmative answer to a problem considered by \textit{A. Bellow} [Lect. Notes Math. 945, 429-431 (1982)] and \textit{H. Furstenberg} (Proc. Durham Conf., June 1982), the author proves the following deep and interesting result: Let (X,\(\mu\),T) be a dynamical system then \(\frac{1}{n}\sum_{k\leq n}T^{k^ 2}f\) converges almost surely for any \(f\in L^ 2(X,\mu)\), more generally \(k^ 2\) can be replaced by an arbitrary polynomial function p(k) with integer coefficients. The problem is reduced to the proof of an inequality \(\| M_ nf\|_ 2\leq C\| f\|_ 2,\quad f\in L^ 2(X,\mu),\) where \(M_ nf\) is the ``maximal function'' \(\sup_{j\leq n}| (\sum_{k\leq j}T^{p(k)})/card\{k: p(k)\leq j\}|.\) This inequality can be proved by showing the according inequality for the special system (\({\mathbb{Z}},\lambda,T)\), \(\lambda\) the counting measure, T the shift. This can be done by Fourier transform methods and careful estimates of exponential sums (Gauss sums if \(p(k)=k^ 2)\), estimates from \textit{A. Sárközy}'s paper [Acta Math. Acad. Sci. Hung. 31, 125- 149 (1978; Zbl 0387.10033)] are used, the case \(p(k)=k^ t\) is associated with the Waring problem. The method of major arcs is of fundamental importance, based on \textit{I. M. Vinogradov} [The method of trigonometric sums in the theory of numbers (1954; Zbl 0055.275; Russian original 1947; Zbl 0041.370)] and \textit{R. C. Vaughan} [The Hardy- Littlewood method (1981; Zbl 0455.10034)]. As consequences one obtains results on uniform distribution, e.g. \(\frac{1}{n}\sum_{k\leq n}f(x+m^ t\alpha)\to \int f(x)dx\) almost surely for \(\alpha\not\in {\mathbb{Q}}\) and \(f\in L^{\infty}({\mathbb{R}}/{\mathbb{Z}})\) or \(\frac{1}{n}\sum_{m\leq n}f(2^{mt}x)\to \int f(x)dx\) a.s. and \(f\in L^{\infty}({\mathbb{R}}/{\mathbb{Z}})\), generalizing the Riesz-Raikov result for \(t=1.\) Furthermore the author obtains according results for commuting transformations and pointwise ergodic theorems for random sets for \(f\in L^ p\), \(p>1\).
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    maximal ergodic theorem
    0 references
    maximal function
    0 references
    exponential sums
    0 references
    Gauss sums
    0 references
    Waring problem
    0 references
    uniform distribution
    0 references
    pointwise ergodic theorems
    0 references
    0 references
    0 references
    0 references