A note on the squarefree density of polynomials
This paper concerns the density of square-free values of a given polynomial in \(s\ge 1\) variables. Let \(P\in \mathbb{Z}[x_1,\dots,x_s]\) be nonzero, with total degree \(\deg{P}\ge 2\). For each prime \(p\), let \(\delta_P(p^2)\) be the proportion of vectors \(x\in [1,p^2]^s\cap \mathbb{Z}^s\) for which \(P(x)\equiv 0\bmod{p^2}\). It is conjectured that (\(\star\)) \N\[\N\frac{\#\{x\in [-B,B]^s\cap \mathbb{Z}^s: \mu(|P(x)|)\ne 0\}} {\#\{x\in [-B,B]^s\cap \mathbb{Z}^s\}} \to \mathfrak{S}_P\N\]\Nas \(B\to \infty\), where \N\[\N\mathfrak{S}_P := \prod_p (1 - \delta_P(p^2)).\N\]\NIn [\textit{J. M. Kowalski} and \textit{R. C. Vaughan}, Acta Arith. 214, 215--233 (2024; Zbl 1560.11141)], it was observed that the infinite product \(\mathfrak{S}_P\), being a limit of a non-increasing sequence of non-negative real numbers, either diverges to \(0\) or converges to a positive real number; and moreover, an upper-bound sieve was used to show that if \(\mathfrak{S}_P=0\), then (\(\ast\)) holds (Theorem~1.3).\N\NThe main result of the present paper (Theorem~1.1) characterizes exactly when \(\mathfrak{S}_P = 0\) occurs: either (a) we have \(\delta_P(p^2) = 1\) for some prime \(p\), or (b) the polynomial \(P\) is divisible by the square of a non-constant polynomial \(L_2\in \mathbb{Z}[x_1,\dots,x_s]\). Interesting corollaries are also given. Corollary~1.2 quantifies the rate of convergence in the asymptotic (\(\star\)), in both cases (a) and (b). Corollary~1.3 uses an upper-bound sieve to show that the true upper square-free density \(\mathfrak{D}_P\) is always at most \(\mathfrak{S}_P\), which implies that \(\mathfrak{D}_P=0\) if either (a) or (b) holds.\N\NThe interested reader is advised to consult this well-written paper for further details, and for references on the rich history of square-free values. The main ingredients in the proof of Theorem~1.1 are multivariate Hensel lifting, the Lang-Weil estimate over the finite fields \(\mathbb{F}_p\), the theory of discriminants (or resultants), and the fact that \N\[\N\prod_p (1-p^{-1}) = 0 < \prod_p (1-p^{-2}).\N\]\NHowever, as remarked in [\textit{Y. G. Zarhin}, ``Comment on `On the squarefree density of polynomials', Preprint, \url{arXiv:2410.09357}], the implication ``\(\mathfrak{S}_P=0\) implies (a) or (b) in Theorem~1.1 is already implicit in Section~6 of [\textit{B. Poonen}, Duke Math. J. 118, No. 2, 353--373 (2003; Zbl 1047.11021)].
This page was built for publication: A note on the squarefree density of polynomials
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7008632)
