On the Mumford-Tate conjecture for hyperkähler varieties (Q2148095)

This paper is concerned with a partial solution of the Mumford-Tate conjecture for hyperkähler varieties. Let \(k\subset \mathbb{C}\) be a finitely generated field and \(\bar{k}\subset \mathbb{C}\) its algebraic closure, consider a smooth projective veriety \(X\) over \(k\). Given any prime number \(l\), Artin's comparison theorem gives an isomorphism of \(\mathbb{Q}_l\)-vector spaces \[H^i_B(X(\mathbb{C}),\mathbb{Q})\otimes_\mathbb{Q}\mathbb{Q}_l\cong H^i_{\textit{é}t} (X(\bar{k}) ,\mathbb{Q}_l)\] for any \(i\), where \(H^i_B(X(\mathbb{C}),\mathbb{Q})\) is the singular cohomology of \(X(\mathbb{C})\) and \(H^i_{\textit{é}t} (X(\bar{k}) ,\mathbb{Q}_l)\) is the étale cohomology of \(X(\bar{k})\) with coefficients in \(\mathbb{Q}_l\). The singular cohomology comes with a Hodge structure, while the étale cohomology has a natural Galois representation, moreover both data are encoded by the corresponding Tannakian fundamental groups. The Mumford-Tate conjecture (Conjecture \(2.1\)) predicts that the two groups are identified by the Artin's correspondence. The statement is referred as Mumford-Tate conjecture in codimension \(j\), when it holds for \(i=2j\). The conjecture has a motivic version (Conjecture \(2.2\)) which implies the classical formulation. From now on we only consider complex projective manifolds. Recall that a hyperkähler manifold \(X\) is a connected, simply connected, compact Kähler manifold with a nowhere degenerate holomorphic \(2\)-form which spans \(H^{0,2}(X)\). The author proves the Mumford-Tate conjecture in any codimension for hyperkähler manifolds of \(K3^{[m]}\)-type, \(OG10\)-type and all their self-products (Theorem \(1.1\)). The result is proven using André motives, in fact a more general statement is shown: if \(X\) is hyperkähler with \(b_2(X)>3\) and for all \(i\geq 0\), the component of degree \(2i\) of the André motive of \(X\) is abelian, then the Mumford-Tate conjecture holds in any codimension for \(X\) (Theorem \(1.2\)). It has to be mentioned that the abelianity of the André motives of deformation types \(K3^{[m]}\), \(\mathrm{Kum}_m\) and \(OG6\) has been established by \textit{A. Soldatenkov} [``Deformation principle and André motives of projective hyperkähler manifolds'', Preprint, \url{arXiv:1904.11320}]. Later, \textit{S. Floccari} et al. [Commun. Contemp. Math. 23, No. 4, Article ID 2050034, 50 p. (2021; Zbl 1458.14016)] proved the abelianity of those motives for the \(OG10\)-type, the last known deformation class. Combining these results with Theorem \(1.2\) proves the conjecture true for all known examples, but the proof of abelianity relies on deformations to specific examples, while the proof of Theorem \(1.1\) does not involve any deformation or abelianity of motives. For the last observation, the techniques used in the proof of Theorem \(1.1\) might be refined for a complete proof of the conjecture for hyperkähler manifolds. Last comment on other works towards the solution, the conjecture was proved to be true in codimension \(1\) for projective hyperkähler manifolds with \(b_2(X)>3\) by André. The validity of the conjecture in codimension \(1\) is indeed used in the proof of Theorem \(1.2\). We now give a short outline of the proofs of the above results. Consider the algebra \[ H_B^+(X)=\bigoplus_i H_B^{2i}(X,\mathbb{Q}),\] it carries a Hodge structure with associated Mumford-Tate group \(MT^+(X)\) and the action of a \(\mathbb{Q}\)-Lie algebra \(\mathfrak{g}_{tot}(X)\) described by \textit{M. Verbitsky} [Geom. Funct. Anal. 6, No. 4, 601--611 (1996; Zbl 0861.53069)] and \textit{E. Looijenga} and \textit{V. A. Lunts} [Invent. Math. 129, No. 2, 361--412 (1997; Zbl 0890.53030)]. Moreover, \(H_B^+(X)\) is the realization of a motive whose motivic Galois group is denoted by \(G^+_{mot}(X)\). The author proves that \(MT^+(X)\) is a direct factor of \(G^+_{mot}(X)\), under the assumption \(b_2(X)>3\) (using the validity of the conjecture in codimension \(1\)). Another key step proved in the article is that the finiteness of the index of \(MT^+(X)\) in \(G^+_{mot}(X)\) implies the conjecture in all codimensions. The two groups coincide for abelian motives (a general result by \textit{Y. André} [Publ. Math., Inst. Hautes Étud. Sci. 83, 5--49 (1996; Zbl 0874.14010)]) and from this Theorem \(1.2\) follows. In order to prove Theorem \(1.1\), the author obtains the finiteness of the complement of \(MT^+(X)\) using its interplay with the algebra action of \(\mathfrak{g}_{tot}(X)\). The representation has a detailed description for \(K3^{[m]}\)-type (given by \textit{E. Markman} [J. Algebr. Geom. 17, No. 1, 29--99 (2008; Zbl 1185.14015)]) and \(OG10\)-type (given by \textit{M. Green} et al. [Math. Ann. 382, No. 3--4, 1517--1590 (2022; Zbl 1497.53095)]). The same conclusions cannot be deduced for \(\mathrm{Kum}_m\) and \(OG6\)-types, because the \(\mathfrak{g}_{tot}(X)\)-representation in cohomology has a different behavior than the other cases.