On the polynomial equivalence of subsets \(E\) and \(f(E)\) of \(\mathbb{Z}\) (Q1818248)

Let \(D\) be an integral domain with quotient field \(K\) and \(E\) a subset of \(D\). Then the ring \(\text{Int}(E,D)\) of integer-valued polynomials on \(E\) is the set of polynomials \(f\) over \(K\) such that \(f(E)\subseteq D\). If \(E=D\) the notation \(\text{Int}(D)\) is used. Two subsets \(E_1\) and \(E_2\) of \(D\) are polynomially equivalent if \(\text{Int}(E_1,D)=\text{Int}(E_2,D)\). The first main result describes the situation when \(E_1,E_2\) are subsets of a Dedekind domain with the finite norm property (abbreviated FNP). In such rings \(E_1\) and \(E_2\) are polynomially equivalent if and only if \(E_1\) and \(E_2\) have coinciding images in the factor rings \(D/Q\) for each non-zero prime-power ideal \(Q\) in \(D\). Then the relation of this condition to a closely related result established by \textit{D. L. McQuillan} [J. Number Theory 39, No. 3, 245-250 (1991; Zbl 0739.13009)] is discussed. The third section is devoted to the polynomial equivalence of \(E\) and \(f(E)\) when \(E\) is finite (despite the fact that the section is entitled `The case where \(E\) is infinite'). An argument of \textit{McQuillan} shows that \(E\) and \(f(E)\) are polynomially equivalent if and only if \(f\) induces a permutation on \(E\). In the main result of this section equivalent conditions are proved for the case when for each finite subset \(E\) of an integral domain \(D\), all permutations of \(E\) are induced by elements of \(\text{Int}(D)\). It is also shown that Dedekind domains with FNP satisfy them. The fourth section is devoted to subsets \(E\) of \(\mathbb Z\), the ring of integers, which are prime-power complete, what means that \(E\) contains a complete set of residues modulo \(q\) for each prime power \(q\in{\mathbb Z}\). It is proved here, among others, that if \(E\) and \(f(E)\) are polynomially equivalent then \(f(t)=a\pm t\) for some \(a\in{\mathbb Z}\). Section 5 of the paper explores the same question for infinite sets which lead to a similar conclusion. Specialization of the proved results shows that if \(E\) is the infinite sets of primes, or the set of factorials, or of powers of a fixed integer \(>1\), then \(t\) is the only polynomial \(f\in\text{Int}({\mathbb Z})\) such that \(E\) and \(f(E)\) are polynomially equivalent.