Fourfolds of Weil type and the spinor map (Q6173829): Difference between revisions

The article under reviewed consists of an exposition of some recent results proved by \textit{K. G. O'Grady} [Int. Math. Res. Not. 2021, No. 16, 12356--12419 (2021; Zbl 1490.53068)] and \textit{E. Markman} [J. Eur. Math. Soc. (JEMS) 25, No. 1, 231--321 (2023; Zbl 1523.14011)] on certain families of abelian fourfolds admitting specific Hodge classes. Let us describe briefly the construction. Starting with a \(\mathbb{Z}\)-module \(W\) of rank~\(4\), \(V:=W\oplus W^*\) is naturally endowed with a bilinear form of signature \((4,4)\). It is easily shown (Lemma~0.4) that orthogonal complex structures (\textit{i.e.} complex structures that preserve the bilinear form) are in one-to-one correspondence with the set of maximally isotropic subspaces \(Z\subset V_\mathbb{C}\) such that \(V_\mathbb{C}=Z\oplus \overline{Z}\). These subspaces form a \(6\)-dimensional complex manifold that can be realized as a quadric \(Q^+\subset \mathbb{P}S^+_\mathbb{C}\) where \(S^+\) is a lattice of rank~\(8\) endowed with a bilinear form \((\cdot,\cdot)\). The open subset \[ \Omega^+:=\left\{ \ell\in\mathbb{P}S^+_\mathbb{C}\mid (\ell,\ell)=0\ \text{and}\ (\ell,\bar{\ell})>0\right\} \] is then a parameter space for a family of tori \(\mathcal{T}_\ell:=V_\mathbb{C}/(Z_\ell+V)\). A construction (using the Clifford algebra and the spinor representation) enables us to define a class \(c_s\in\bigwedge^4 V\) that can be interpreted as a cohomology class on the torus \(\mathcal{T}_\ell\). The compatibility condition with the complex structure \(Z_\ell\) reads like this: the class \(c_s\in\mathrm{H}^4(\mathcal{T}_\ell,\mathbb{Z})\) is a Hodge class (of type \((2,2)\)) if and only if \((s,\ell)=0\). Playing with another element \(h\), it is possible to construct a family of abelian fourfolds with very interesting properties: this is the content of Theorem~4.6. Theorem. Let \(h\) and \(s\) such that the restriction of \((\cdot,\cdot)\) to the plane \(\langle h,s \rangle\) is positive definite and define \[ \Omega^+_{h,s}:=\left\{ \ell\in\mathbb{P}S^+_\mathbb{C}\mid (\ell,\ell)=(\ell,h)=(\ell,s)=0\ \text{and}\ (\ell,\bar{\ell})>0\right\}. \] This \(4\)-dimensional manifold parametrizes a complete family of polarized abelian fourfold of Weil type and such that the Cayley class does not lie on the line generated by \(\omega_\ell^2\) where \(\omega_\ell\) is the above-mentioned polarization. Being of Weil type means that these abelian fourfolds are endowed with complex multiplication \(K\subset\mathrm{End}(\mathcal{T}_\ell)\) where \(K=\mathbb{Q}(\sqrt{-d})\) where \(d:=(h,h)\cdot(s,s)>0\) (see Section~4.2 for the precise definition). Finally when the \(\mathbb{Z}\)-module \(W\) we started with satisfies \(W=\mathrm{H}^2(A,\mathbb{Z})\) with \(A\) an abelian surface, the whole construction can be interpreted in the framework of moduli spaces of sheaves on this abelian surface as in [\textit{S. Mukai}, Am. J. Math. 117, No. 6, 1627--1644 (1995; Zbl 0871.14025); \textit{K. Yoshioka}, Math. Ann. 321, No. 4, 817--884 (2001; Zbl 1066.14013)]. This makes the link with the works Markman and O'Grady alluded to above. This article gives a \textit{pedestrian} (as said in the abstract) way to get to Markman and O'Grady's results: the notions are elementary and the computations are made as explicit as possible. It is very well written and will be useful for everyone willing to enter this circle of ideas.