Distribution of values of bounded generalized polynomials

The authors study the class \(\text{GP}= \bigcup_{n=1}^\infty \text{GP}_n\) of generalized polynomials, where \(\text{GP}_0\) denotes the ring of polynomial mappings form \(\mathbb Z^d\) (or \(\mathbb R^d\)) to \(\mathbb R\) and \[ \text{GP}_n= \text{GP}_{n-1}\cup \{v+w: v,w\in \text{GP}_{n-1}\}\cup \{vw: v,w\in \text{GP}_{p-1}\}\cup \{[v]: v\in \text{GP}_{n-1}\},\;n\geq 1, \] where \([a]\) denotes the integer part of the vector \(a\) as usual. GP mappings are then generalized polynomials \(u= (u_1,u_2,\dots, u_l)\in\mathbb Z^d\to\mathbb R^l\) with \(u_l\in\text{GP}\). The purpose of this paper is answer the question of whether for a bounded generalized polynomial \(u\), the sequence \(\{u(n)\}_{n\in\mathbb{Z}}\) has some regular behavior. For example, does \[ \lim_{N\to\infty}\,{1\over N} \sum^\infty_{n=1} e^{2\pi iu(n)} \] exist? Results of this type are etablished using properties of nilsystems. Theorem A: (i) For any nilmanifold \(X\), and action \(\phi\) of \(\mathbb{Z}^d\) by nilrotations on \(X\), any piecewise polynomial mapping \(f: X\to\mathbb{R}^l\), and any point \(x\in X\), the mapping \(u(n)= f(\phi(n)x)\), \(n\in\mathbb{Z}^d\), is a \(\text{GP}\) mapping. (ii) For any bounded \(\text{GP}\) mapping \(u: \mathbb{Z}^d\to\mathbb{R}^l\), there exists a nilmanifold \(X\), an ergodic action \(\phi\) of \(\mathbb{Z}^d\) by nilrotations on \(X\), a piecewise polynomial mapping \(f: X\to\mathbb{R}^l\) and a point \(x\in X\) such that \(u(n)= f(\phi(n)x)\), \(n\in\mathbb{Z}^d\). The following gives the basic ideas of the main theorem: Theorem B: Let \(u: \mathbb{Z}^d\to \mathbb{R}^l\) be a bounded \(\text{GP}\) mapping. There exists a bounded piecewise polynomial surface \(S\) such that \(u(n)\in S\) for almost all \(n\in\mathbb{Z}^d\), and the sequence \(\{u(n)\}_{n\in\mathbb{Z}^d}\) is well distributed on \(S\) with respect to \(\mu_S\) (where \(\mu_S\) is the measure on \(S\) which is the image of Lebesgue measure on \(\mathbb{R}^s\) under the map that defines \(S\)). This enables one to give a positive answer to the above question, and also to answer more general questions. A generalization of von Neumann's ergodic theorem is given, and a continuous versions of the above results are presented. There are numerous examples and applications of the theorems worked out in some detail. This paper represents an important contribution to the literature.

Recommendations

Cites work

Cited in

(57)

This page was built for publication: Distribution of values of bounded generalized polynomials

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2373563)