Absolutely trianalytic tori in the generalized Kummer variety (Q291774)

The title is misleading: the main theorem here about absolutely trianalytic tori in the generalised Kummer variety is that there aren't any. A subvariety of a hyperkähler manifold \(X\) is absolutely trianalytic if it is analytic with respect to any of the complex structures on \(X\). Such a variety \(Z\) does not have to be normal but its normalisation \(\tilde Z\) is smooth and comes with a hyperkähler structure. So it has a torus factor and some irreducible symplectic factors, and this paper considers the case where only the torus factor is present. In that case, if \(X\) is the generalised Kummer variety associated with the torus \(T\), one may think of \(Z\) as embedded in the Hilbert scheme (or Douady space) \(T^{[n]}\) and compose that embedding with the map to the symmetric product \(T^{(n)}\). Pulling back to \(T^n\) one obtains a union of irreducible components each of which is a subtorus of \(T^n\). However, since trianalyticity is preserved under deformations, we may as well assume that \(T\) is general; it follows that the components are tori and, like \(T\), have Picard rank zero. It follows from this that \(Z\) does not meet the exceptional divisor of \(T^{[n]}\to T^{(n)}\), and then a computation of the volume of \(Z\) with respect to the Kähler forms gives a contradiction.