Gushel-Mukai varieties with many symmetries and an explicit irrational Gushel-Mukai threefold (Q2162461)

In this paper the authors construct an explicit smooth Fano complex threefold with Picard number \(1\), index \(1\) and degree \(10\) known as Gushel-Mukai threefold, and they prove it is not rational by showing it has a faithful \(\mathrm{PSL}(2,\mathbb{F}_{11})\)-action. \textit{A. Beauville} [Ann. Sci. Éc. Norm. Supér. (4) 10, 309--391 (1977; Zbl 0368.14018)] proved that a general Fano threefold with Picard number \(1\), index \(1\) and degree 10 (also known as a Gushel-Mukai threefold) is irrational, but no smooth examples are known. Here they construct a complete 2-dimensional family of such examples (Corollary 5.3) including one such threefold defined over \(\mathbb{Q}\) by explicit equations (Corollary 5.4). The starting point of the paper is an EPW-sextic with a faithful \(\mathrm{PSL}(2,\mathbb{F}_{11})\)-action discovered by Mongardi in 2013. Starting from this famous sextic hypersurface \(Y_{\mathbb{A}} \subset \mathbb{P}^5\) with a faithful action of a simple group \(\mathbb{G}:= \mathrm{PSL}(2, \mathbb{F}_{11})\) they prove that \(\Aut(Y_{\mathbb{A}})=\mathbb{G}\) (Proposition 4.1) and this is the only quasi smooth EPW sextic with an automorphism of order \(11\) (Theorem 4.2). Along the way they construct Gushel-Mukai varieties of various dimensions with large (finite) automorphism groups. By [\textit{O. Debarre} and \textit{A. Kuznetsov}, Épijournal de Géom. Algébr., EPIGA 4, Article 19, 45 p. (2020; Zbl 1457.14092)] the intermediate Jacobians of Gushel-Mukai varieties obtained from the sextic \(Y_{\mathbb{A}}\) are all isomorphic to a principally polarized abelian variety \((\mathbb{J}, \theta)\). They use the Clemens-Griffiths [\textit{C. H. Clemens} and \textit{P. A. Griffiths}, Ann. Math. (2) 95, 281--356 (1972; Zbl 0214.48302)] criterion to prove that the GM threefolds that they construct from \(Y_{\mathbb{A}}\) are not rational. To apply the criterion it is enough to prove that their (common) intermediate Jacobian \((\mathbb{J}, \theta)\) is not a product of Jacobians of curves and to get this, they use the fact that \((\mathbb{J}, \theta)\) has too many automorphisms (because of the \(\mathbb{G}\)-action). This is how they produce a complete 2-dimensional family of irrational GM threefolds, all mutually birationally isomorphic. Moreover another important object that the authors study in the paper is the 10-dimensional principally polarized abelian variety \((\mathbb{J}, \theta)\). They get that this abelian variety is indecomposable and isogenous to the product of ten copies of an elliptic curve (Proposition 5.1 and 5.8) and they conjecture that this variety is isomorphic to an explicit 10 dimensional principally polarized abelian variety (see the construction in Proposition C.3). Since this situation is quite similar to the one of the Klein cubic threefold \(W \subset \mathbb{P}^4\), they call the sextic \(Y_{\mathbb{A}}\), the Klein EPW sextic. Finally it is remarkable to recall that the proofs heavily use the construction by \textit{K. G. O'Grady} [Mich. Math. J. 62, No. 1, 143--184 (2013; Zbl 1276.14008)] of canonical double covers of quasi-smooth EPW sextics called double EPW sextics [\textit{O. Debarre} and \textit{A. Kuznetsov}, Math. Ann. 378, No. 3--4, 1435--1469 (2020; Zbl 1468.14027)].