Filtrations of numerically flat Higgs bundles and curve semistable Higgs bundles on Calabi-Yau varieties (Q6168280)

In this paper, the authors consider Higgs bundles satisfying a notion of numerical flatness (H-nflatness) that was introduced in [\textit{U. Bruzzo} and \textit{D. Hernández Ruipérez}, Differ. Geom. Appl. 24, No. 4, 403--416 (2006; Zbl 1108.14028); \textit{U. Bruzzo} and \textit{B. G. Otero}, Adv. Math. 226, No. 4, 3655--3676 (2011; Zbl 1263.14019)], and show that they have Jordan-Hölder filtrations whose quotients are stable, locally free and H-nflat. The main result of this paper is to show that simply connected Calabi-Yau varieties of all dimensions are Higgs varieties, generalizing [\textit{U. Bruzzo} et al., Asian J. Math. 23, No. 6, 905--918 (2019; Zbl 1440.14201)] where this was proved for \(K3\) surfaces. The result is applied to show that curve semistable Higgs bundles on simply connected Calabi-Yau varieties have vanishing discriminant. This will be accomplished by proving an equivalent version of the following conjecture: a curve semistable Higgs bundle has vanishing discriminant, which is stated in terms of a class of Higgs bundles that have been called Higgs numerically flat (for short, H-nflat); the main technical points are two. First the authors prove that H-nflat Higgs bundles have a particular kind of filtration, and then they apply a result given in [\textit{I. Biswas} et al., Asian J. Math. 20, No. 5, 989--1000 (2016; Zbl 1362.32014)] which states that on a simply connected Calabi-Yau variety a polystable Higgs bundle always has a vanishing Higgs field. Then the conjecture follows from the ordinary vector bundle case. This paper is organized as follows: Section 1 is an introduction to the subject. Section 2 and Section 3 are devoted to H-nflat Higgs bundles and filtering H-nflat Higgs bundles, respectively. Section 4 deals with the conjecture for Calabi-Yau varieties. Section 5 is devoted to Higgs varieties and special metrics. In this section the authors show, as another application of the theorem on the filtrations of H-nflat bundles, that on a Higgs variety H-nflat bundles are \(1\)-H-nflat, and actually that this property characterizes Higgs varieties (thus the coincidence of the notions of H-nflatness and \(1\)-H-nflatness is yet another form of the conjecture above). An immediate but nontrivial byproduct is that the notions of numerical flatness and \(1\)-numerical flatness for ordinary vector bundles coincide.