Effective and quantitative forms of Hilbert's irreducibility theorem: polynomial specialization bounds and complete arithmetic description of exceptional sets

Hilbert's irreducibility theorem (1892) guarantees that an irreducible two-variable polynomial $F(T,Y)\in\mathbb{Z}[T,Y]$ admits infinitely many integers $t_0$ for which $F(t_0,Y)$ remains irreducible over $\mathbb{Q}$ with its Galois group preserved, and it underlies the founding reduction of the inverse Galois problem to rational function fields. The theorem is an existence statement from the pre-algorithmic era: it neither produces an admissible specialization effectively nor describes the exceptional set. This record concerns the research problem of making the theorem effective, quantitative, and arithmetically complete. A century of progress has resolved the density strand (exceptional sets have density zero, with sharp $B^{1/2}$ counting rates) and produced effective bounds polynomial in the logarithmic height but exponential in the degree. Three precise cores remain open: (1) whether the least admissible specialization admits a bound polynomial in both the degree and the logarithmic height — posed explicitly by Dèbes and Walkowiak in 2008 and still open, with all current bounds (Paredes and Sasyk 2024; Cluckers, Dèbes, Hendel, Nguyen and Vermeulen 2025) exponential in the degree; (2) the complete arithmetic determination of the exceptional set for arbitrary explicit input, which the Krumm–Sutherland reduction transforms into determining rational points on an explicit finite list of curves of unbounded genus, a task with no known general algorithm; (3) explicit unconditional dependence on the base number field, without assuming the Generalized Riemann Hypothesis (GRH).

Background

The theorem. Let $F(T,Y)\in\mathbb{Z}[T,Y]$ be irreducible as a two-variable polynomial with $\deg_Y F\ge 1$, and let $G$ be the Galois group of $F$ over the rational function field $\mathbb{Q}(T)$, i.e. the automorphism group of its splitting field over $\mathbb{Q}(T)$. A specialization is the substitution of a rational number $t_0$ for the parameter $T$, giving a one-variable polynomial $F(t_0,Y)\in\mathbb{Q}[Y]$; the integer specializations are the case $t_0\in\mathbb{Z}$. Hilbert (1892) proved that infinitely many integers $t_0$ — in fact all but an arithmetically small exceptional set — leave $F(t_0,Y)$ irreducible over $\mathbb{Q}$, preserve the number of irreducible factors in the multivariate and algebraic-coefficient variants, and preserve the full Galois group: $\mathrm{Gal}(F(t_0,Y))\cong G$. Since realizing a finite group over $\mathbb{Q}(T)$ is comparatively easy (the generic degree-$n$ polynomial has group $S_n$), the theorem reduces realizing groups over $\mathbb{Q}$ to realizing them over $\mathbb{Q}(T)$ — the founding reduction of inverse Galois theory, which Hilbert used to realize the symmetric and alternating groups of every degree over any number field and to construct number fields with no proper subfields.

The exceptional set. The complement of the good specializations, $E(F)=\{t\in\mathbb{Q} : F(t,Y)\ \text{reducible over}\ \mathbb{Q}\ \text{or}\ \mathrm{Gal}(F(t,Y))\not\cong G\}$, is the exceptional set, a subset of the rational numbers; write $E_{\mathbb{Z}}(F)=E(F)\cap\mathbb{Z}$ for its integer part, the object that is counted in quantitative statements. It can be infinite: for $F=Y^2-T$ the bad specializations are exactly the rational squares, so $|E_{\mathbb{Z}}(F)\cap[-B,B]|$ grows like $B^{1/2}$, showing the natural counting rate $B^{1/2}$ cannot be improved in general. The theorem implies $E(F)$ is contained in a thin set: a finite union of images of non-invertible rational maps plus a finite set, the standard notion encoding arithmetically small.

Effectivity and quantitative bounds. Hilbert's proof — a combinatorial 'cube' lemma on iterated substitutions together with analytic perturbation arguments on algebraic functions — is non-effective: it yields neither a procedure that produces an admissible $t_0$ with bounded effort, nor explicit numerical bounds. Writing $d$ for the total degree of $F$ (the maximum of the total degrees of its monomials in $T$ and $Y$) and $H$ for its height (the maximum absolute value of its coefficients), the modern program asks for explicit upper bounds on $|E_{\mathbb{Z}}(F)\cap[-B,B]|$ in terms of $d$, $\log H$, and $B$ — via the determinant method of Bombieri and Pila for integral points on curves and its successors, and the large and larger sieves (sieve techniques from analytic number theory) — and for an effective least admissible value: an explicit function $P(d,\log H)$ such that some integer $t_0$ with $|t_0|\le P$ is admissible, where admissible means $F(t_0,Y)$ irreducible over $\mathbb{Q}$ with $\mathrm{Gal}(F(t_0,Y))\cong G$. 'Effective' here means explicitly bounded in the input invariants, since a known bound immediately yields a terminating brute-force search.

From exceptional sets to curves. Galois theory reduces $E(F)$ to finitely many one-parameter resolvents: for each conjugacy class of maximal subgroups $M$ of $G$ one can compute a polynomial $f_M(T,X)$ such that any exceptional $t$ is a first coordinate of a rational point on one of the curves $f_M(T,X)=0$, up to an explicitly computable finite set (Krumm 2016; Krumm and Sutherland 2021). Determining rational points on curves is elementary for genus $0$, governed by the Mordell–Weil group for genus $1$, and fundamentally hard for genus $\ge 2$, where Faltings' theorem gives finiteness ineffectively. This is the precise obstruction standing between the known reduction and a complete description of $E(F)$.

Problem Statement

Give an effective, quantitative, and arithmetically complete form of Hilbert's irreducibility theorem for an explicitly given irreducible polynomial $F(T,Y)\in\mathbb{Z}[T,Y]$ with $\deg_Y F\ge 1$, with Galois group $G$ over $\mathbb{Q}(T)$; core 3 extends the objective to the number-field variant. Throughout, $d$ denotes the total degree of $F$ (the maximum of the total degrees of its monomials in $T$ and $Y$), $H$ its height (the maximum absolute value of its coefficients), and $E(F)\subseteq\mathbb{Q}$ its exceptional set as defined in the background. The precise open cores are:

(1) Least admissible specialization. Prove or refute: there exists a polynomial $P$ such that for every irreducible $F\in\mathbb{Z}[T,Y]$ with $\deg_Y F\ge 1$, of total degree $d$ and height $H$, some integer $t_0$ with $|t_0|\le P(d,\log H)$ makes $F(t_0,Y)$ irreducible over $\mathbb{Q}$ with $\mathrm{Gal}(F(t_0,Y))\cong G$. This prove-or-refute question was posed explicitly by Dèbes and Walkowiak (2008); this record fixes $d$ to be the total degree, in agreement with the $\deg(P)$ convention used by Dèbes and Walkowiak and with the sharpest current bounds (Cluckers et al. 2025). All current effective bounds are polynomial in $\log H$ but exponential in $d$.

(2) Complete arithmetic description of the exceptional set. For arbitrary explicit $F$, determine $E(F)$ completely: decide whether it is finite and, if so, compute it exactly; if infinite, produce a finite arithmetic description (e.g. value sets of finitely many explicit parametrizations) from which the elements of $E(F)$ of bounded height — in particular $E(F)\cap\mathbb{Z}\cap[-B,B]$ — can be effectively enumerated for every $B$. The Krumm–Sutherland reduction shows this is equivalent, in general, to determining the rational points on an explicit finite list of curves of unbounded genus, for which no general algorithm is known.

(3) Unconditional field dependence. Give effective bounds whose constants depend explicitly and unconditionally — without assuming the Generalized Riemann Hypothesis — on the base number field $K$ when the theorem is stated over $K$ (that is, for irreducible $F\in K[T,Y]$ with coefficients in the number field $K$, with the Galois group taken over $K(T)$).

A rigorous resolution of any single core, in the stated generality, constitutes the scientific objective tracked by this record.

References

1. David Hilbert (1892). Ueber die Irreducibilität ganzer rationaler Functionen mit ganzzahligen Coefficienten. Journal für die reine und angewandte Mathematik 110, 104–129. DOI: 10.1515/crll.1892.110.104. URL: https://doi.org/10.1515/crll.1892.110.104
2. Karl Dörge (1926). Zum Hilbertschen Irreduzibilitätssatz. Mathematische Annalen 95, 84–97. DOI: 10.1007/BF01206597. URL: https://doi.org/10.1007/BF01206597
3. Michael Fried (1974). On Hilbert's Irreducibility Theorem. Journal of Number Theory 6(3), 211–231. DOI: 10.1016/0022-314X(74)90015-8. URL: https://doi.org/10.1016/0022-314X(74)90015-8
4. Stephen D. Cohen (1981). The Distribution of Galois Groups and Hilbert's Irreducibility Theorem. Proceedings of the London Mathematical Society (3) 43(2), 227–250. DOI: 10.1112/plms/s3-43.2.227. URL: https://doi.org/10.1112/plms/s3-43.2.227
5. Erich Kaltofen (1985). Effective Hilbert irreducibility. Information and Control 66(3), 123–137. DOI: 10.1016/S0019-9958(85)80056-5. URL: https://doi.org/10.1016/S0019-9958(85)80056-5
6. Andrzej Schinzel and Umberto Zannier (1995). The least admissible value of the parameter in Hilbert's Irreducibility Theorem. Acta Arithmetica 69(3), 293–302. DOI: 10.4064/AA-69-3-293-302. URL: https://doi.org/10.4064/AA-69-3-293-302
7. Peter Müller (2002). Finiteness results for Hilbert's irreducibility theorem. Annales de l'Institut Fourier 52(4), 983–1015. DOI: 10.5802/aif.1907. arXiv:math/0109071. URL: https://doi.org/10.5802/aif.1907
8. Yann Walkowiak (2005). Théorème d'irréductibilité de Hilbert effectif. Acta Arithmetica 116(4), 343–362. DOI: 10.4064/aa116-4-3. URL: https://doi.org/10.4064/aa116-4-3
9. Pierre Dèbes and Yann Walkowiak (2008). Bounds for Hilbert's Irreducibility Theorem. Pure and Applied Mathematics Quarterly 4(4), 1059–1083. DOI: 10.4310/pamq.2008.v4.n4.a4. URL: https://doi.org/10.4310/pamq.2008.v4.n4.a4
10. David Zywina (2010). Hilbert's irreducibility theorem and the larger sieve. Preprint, arXiv:1011.6465. URL: https://arxiv.org/abs/1011.6465
11. Umberto Zannier (2010). Hilbert irreducibility above algebraic groups. Duke Mathematical Journal 153(2), 397–425. DOI: 10.1215/00127094-2010-027. URL: https://doi.org/10.1215/00127094-2010-027
12. Abel Castillo and Rainer Dietmann (2017). On Hilbert's irreducibility theorem. Acta Arithmetica 180(1), 1–14. DOI: 10.4064/aa8380-2-2017. arXiv:1602.00314. URL: https://doi.org/10.4064/aa8380-2-2017
13. David Krumm (2016). Explicit Hilbert Irreducibility. Preprint, arXiv:1610.03528. URL: https://arxiv.org/abs/1610.03528
14. David Krumm and Nicole Sutherland (2021). Galois groups over rational function fields and explicit Hilbert irreducibility. Journal of Symbolic Computation 103, 108–126. DOI: 10.1016/j.jsc.2019.11.003. arXiv:1708.04932. URL: https://doi.org/10.1016/j.jsc.2019.11.003
15. Lior Bary-Soroker and Alexei Entin (2021). Explicit Hilbert's irreducibility theorem in function fields. In: Contemporary Mathematics 767, 125–134. DOI: 10.1090/conm/767/15402. URL: https://doi.org/10.1090/conm/767/15402
16. Marcelo Paredes and Román Sasyk (2024). Effective Hilbert's Irreducibility Theorem for global fields. Israel Journal of Mathematics 261(2), 851–877 (issue of June 2024; published online 18 December 2023). DOI: 10.1007/s11856-023-2604-7. arXiv:2202.10420. URL: https://doi.org/10.1007/s11856-023-2604-7
17. Theresa C. Anderson, Ayla Gafni, Robert J. Lemke Oliver, David Lowry-Duda, George Shakan and Ruixiang Zhang (2023). Quantitative Hilbert Irreducibility and Almost Prime Values of Polynomial Discriminants. International Mathematics Research Notices 2023(3), 2188–2214. DOI: 10.1093/imrn/rnab296. URL: https://doi.org/10.1093/imrn/rnab296
18. Lior Bary-Soroker and Daniele Garzoni (2023). Hilbert's Irreducibility Theorem via Random Walks. International Mathematics Research Notices 2023(14), 12512–12537. DOI: 10.1093/imrn/rnac188. URL: https://doi.org/10.1093/imrn/rnac188
19. Gal Binyamini, Raf Cluckers and Dmitry Novikov (2024). Bounds for Rational Points on Algebraic Curves, Optimal in the Degree, and Dimension Growth. International Mathematics Research Notices 2024(11), 9256–9265. DOI: 10.1093/imrn/rnae034. URL: https://doi.org/10.1093/imrn/rnae034
20. Raf Cluckers, Pierre Dèbes, Yotam I. Hendel, Kien Huu Nguyen and Floris Vermeulen (2025). Improvements on dimension growth results and effective Hilbert's irreducibility theorem. Forum of Mathematics, Sigma 13, e153. DOI: 10.1017/fms.2025.10096. arXiv:2311.16871. URL: https://doi.org/10.1017/fms.2025.10096
21. Yuki Ishihara and Kazuhiro Yokoyama (2025). Effective Hilbert's Irreducibility Theorem for Primary Ideals. In: Computer Algebra in Scientific Computing (CASC 2025), Lecture Notes in Computer Science 16235, 134–153, Springer. DOI: 10.1007/978-3-032-09645-6_8. URL: https://doi.org/10.1007/978-3-032-09645-6_8
22. Giulio Bresciani (2025). A Higher Dimensional Hilbert Irreducibility Theorem. American Journal of Mathematics 147(3), 779–794. DOI: 10.1353/ajm.2025.a961346. arXiv:2101.01090. URL: https://doi.org/10.1353/ajm.2025.a961346