On the reducibility behavior of Thue polynomials (Q518065)

The study of irreducible polynomials on at least two variables after specialization of some variables in a field \(K\) is an important subject. In particular, the Hilbert irreducibility theorem states that if \(f(t, x)\in \mathbb{Q}[t, x]\) is an irreducible polynomial over \(\mathbb{Q}(t)\) and with \(\deg_{x}(f) \geq 2\), then \(f(t_0, x)\) remains irreducible for infinitely many specializations \(t_0 \in \mathbb{Z}\) of the variable \(t.\) It is also known that this theorem holds for any finite extension of \(\mathbb{Q}\) (number field). The paper under review provides a special case of this subject. More precisely, let \(H(t, x) \in \mathbb{Q}[t, x]\) be a homogeneous polynomial of total degree \(n\), not divisible by \(t\), and not a proper power over \(\overline{\mathbb{Q}}\). Then if \(n\not= 2, 4\), the polynomial \(H(t_0, x)\) is irreducible for all but finitely many \(t_0\in \mathbb{Z}\). This was conjectured in the paper [\textit{P. Müller}, Ann. Inst. Fourier 52, No. 4, 983--1015 (2002; Zbl 1014.12002), Conjecture 4.10] and proven for odd degree \(n.\) These results are not true for number fields as pointed in the paper [loc. cit., Remark, page 1000]. The proof of the main result (in the paper under review) uses the classification of primitive permutation groups containing an \(n\)-cycle. In particular if \(G \leq {S_n}\) be a finite permutation group generated by a cyclic transitive subgroup and a normal transitive subgroup, then \(G\) does not contain any element of order larger than \(n\).