Polynomial actions of rings of integers of global fields and quasirandomness of Paley-type graphs
Ethan Ackelsberg, Vitaly Bergelson
Published: 2025/9/22
Abstract
The goal of this paper is to undertake an in-depth study of the phenomenon behind the Furstenberg--S\'ark\"ozy theorem, which, in its modern form due to Kamae and Mend\`es-France, states that if $E$ is a set of integers with positive density and $P$ is an intersective polynomial, then there are distinct elements $x, y \in E$ such that $x - y = P(n)$ for some some $n$. In this paper, we identify an algebraic framework (rings of integers of global fields) for Furstenberg--S\'ark\"ozy-type theorems. One of our main results establishes necessary and sufficient conditions for a polynomial to satisfy the Furstenberg--S\'ark\"ozy theorem over the ring of integers of a global field, providing an extension of the result of Kamae and Mend\`es-France. The Furstenberg--S\'ark\"ozy phenomenon goes beyond infinite rings and has interesting additional aspects in finite rings. As an example, classical exponential sum estimates can be used to show that large subsets of finite fields contain the asymptotically ``correct'' number of pairs $(x,y)$ whose difference is a square. In previous work, the class of polynomials satisfying this strong form of the Furstenberg--S\'ark\"ozy theorem over finite fields was classified. In the present paper, we establish asymptotic results characterizing sequences of finite principal ideal rings that produce ``correct'' statistics in the Furstenberg--S\'ark\"ozy theorem and show that these families are much more general than finite fields. As an application of our enhanced forms of the Furstenberg--S\'ark\"ozy theorem over finite rings, we produce new families of examples of quasirandom graphs of algebraic origin. The production of these new examples hinges on a two-way connection between asymptotic total ergodicity -- the phenomenon responsible for enhanced versions of the Furstenberg--S\'ark\"ozy theorem over finite fields and rings -- and quasirandomness.