An explicit integral polynomial whose splitting field has Galois group \(W(E_8)\) (Q1026983)

It is known that the Weyl group \(W({\mathbf E}_{8})\) of the exceptional algebraic group \({\mathbf E}_{8}\) is a Galois group over \({\mathbb Q}\) [see \textit{T. Shioda}, Proc. Int. Congr. Math., Kyoto/Japan 1990, Vol. I, 473--489 (1991; Zbl 0746.14009) for a survey]. In this article the authors construct an explicit integral polynomial \(P\) whose splitting field over \({\mathbb Q}\) has Galois group isomorphic to \(W({\mathbf E}_{8})\). Their strategy is summarized as follows in the introduction: if \({\mathbf G}/{\mathbb Q}\) is a split connected reductive algebraic group given as a \({\mathbb Q}\)-subgroup of \(\text{GL}(r)\), then the characteristic polynomial of the image in \(\text{GL}(r)\) of a random element is typically isomorphic to \(W({\mathbf G})\). At the beginning of the article, the authors recall several definitions and facts about \({\mathbf E}_{8}\) and \(W({\mathbf E}_{8})\). Next, they consider a semisimple element \(g\) of \({\mathbf E}_{8}({\mathbb Q})\) and \({\mathbf T}\) a maximal torus of \({\mathbf E}_{8}\) that contains \(g\). If \(P(T)\) denotes the polynomial \(\mathrm{det}(T-\text{Ad}(g))/(T-1)^8\) and if it is separable, then the \(\mathrm{Gal}(\bar{\mathbb Q}/{\mathbb Q})\)-action on its roots induces one on the root system \(R({\mathbf T}, {\mathbf E}_{8})\) and an injective homomorphism \[ \mathrm{Gal}(K/{\mathbb Q}) \to W({\mathbf T}, {\mathbf E}_{8}). \] In the third section, the element \(g\) is chosen to be the product of the 16 ``algebraic generators'' of \({\mathbf E}_{8}/{\mathbb Q}\) given by \texttt{Magma}, which the authors see as the result of a random walk. The polynomial \(P(T)\in {\mathbb Z}[T]\) that corresponds to it has degree 240 and is irreducible over \({\mathbb Q}\). Let us denote \(G \subset \mathfrak{S}_{240}\) the Galois group of its splitting field over \({\mathbb Q}\). Looking at the irreducible factors of the polynomial \(P\) modulo 7 and 11, one deduces the existence of two elements, called \(g_{8}\) and \(g_{15}\) in \(G\), with special cycles decomposition. Then a computation with \texttt{Magma} gives the list of the cycle structures of the maximal subgroups of \(W({\mathbf E}_{8})\). None of these contains two elements with the cycle structures of \(g_{8}\) and \(g_{15}\), hence \(G=W({\mathbf E}_{8})\). In section 4, the authors explain how to use the previous construction to get infinitely many linearly disjoint polynomials with Galois group \(W({\mathbf E}_{8})\). The paper ends with two appendices: the first explains how to define the polynomial \(P(T)\) in an intrinsic way (so that it does not depend on a particular implementation in \texttt{Magma}) and the second gives the list of the coefficients of a polynomial \(Q(T)\) in \({\mathbb Z}[T]\) of degree 120 such that \(P(T) = T^{120} Q(T+T^{-1})\). To finish, let us mention that the authors have added a lot of enlightening remarks to give more details and hints on how the proofs could be generalized or made more intrinsic.