2504.03851
p-adic root separation and the discriminant of integer polynomials
Victor Beresnevich, Bethany Dixon
correcthigh confidence
- Category
- Not specified
- Journal tier
- Strong Field
- Processed
- Sep 28, 2025, 12:56 AM
- arXiv Links
- Abstract ↗PDF ↗
Audit review
The paper proves that for 0 ≤ ν ≤ n−1 there are ≫ Q^{n+1 − (n+2)/n · ν} irreducible degree-n polynomials of height ≍ Q with |D(P)|_p ≪ Q^{−2ν} (Theorem 3.1, formula (3.7); proof in §9) . The candidate solution claims a strictly stronger exponent, ≫ Q^{n+1 − (n+1)/(2(n−1)) · ν}, obtained by imposing high-order p-adic Taylor-vanishing at a point and counting lattice solutions. This contradicts the best-known cubic upper bound #D3,p(Q,ν) ≪ Q^{4 − 5/3·ν + ε} for all polynomials (hence for the irreducible subset), so the candidate lower bound cannot be correct for n=3 (0 ≤ ν ≤ 2) . The paper’s argument is coherent and (modulo a minor typo where Dn,p appears in place of Dirr in (9.8)) completes the proof; the model’s counting step overestimates the number of admissible polynomials.
Referee report (LaTeX)
\textbf{Recommendation:} minor revisions \textbf{Journal Tier:} strong field \textbf{Justification:} The work generalizes influential real/Archimedean techniques to the p-adic setting and proves a lower bound that is consistent with (and in the cubic case essentially matches) the best known upper bounds. The proof is carefully structured around a quantitative non-divergence lemma that yields many irreducible polynomials with controlled derivatives, leading to discriminant and counting results. Aside from a small notational slip in the covering argument, the paper appears correct and clearly written.