English

Integer-valued o-minimal functions

Logic 2024-04-17 v1 Number Theory

Abstract

We study Ran,exp\mathbb{R}_{\textrm{an},\exp}-definable functions f:RRf:\mathbb{R}\to \mathbb{R} that take integer values at all sufficiently large positive integers. If f(x)=O(2(1+105)x)|f(x)|= O\big(2^{(1+10^{-5})x}\big), then we find polynomials P1,P2P_1, P_2 such that f(x)=P1(x)+P2(x)2xf(x)=P_1(x)+P_2(x)2^x for all sufficiently large xx. Our result parallels classical theorems of P\'olya and Selberg for entire functions and generalizes Wilkie's classification for the case of f(x)=O(Cx)|f(x)|= O(C^x), for some C<2C<2. Let kNk\in \mathbb{N} and γk=j=1k1/j\gamma_k=\sum_{j=1}^{k} 1/j. Extending Wilkie's theorem in a separate direction, we show that if ff is kk-concordant\textit{concordant} and f(x)=O(Cx)|f(x)|= O(C^{x}), for some C<eγk+1C<e^{\gamma_k}+1, then ff must eventually be given by a polynomial. This is an analog of a result by Pila for entire functions.

Keywords

Cite

@article{arxiv.2404.10737,
  title  = {Integer-valued o-minimal functions},
  author = {Neer Bhardwaj and Raymond McCulloch and Nandagopal Ramachandran and Katharine Woo},
  journal= {arXiv preprint arXiv:2404.10737},
  year   = {2024}
}

Comments

14 pages

R2 v1 2026-06-28T15:56:06.993Z