Reduction of manifolds with semi-negative holomorphic sectional curvature (Q1627766)

Let \(X\) be a complex projective manifold such that its canonical bundle \(K_X = \det \Omega_X\) is nef. The abundance conjecture claims that the canonical bundle is actually semi-ample, so some multiple defines a morphism \(X \rightarrow Y\), the Iitaka fibration. If this morphism is not birational, i.e. if \(X\) is not of general type, the relative tangent sheaf \(T_{X/Y}\) defines some distinguished directions in the tangent bundle \(T_X\). Assume now that \(X\) has semi-negative holomorphic sectional curvature \(H\), then it is interesting to try to recover these distinguished tangent directions from the curvature tensor. In [J. Differ. Geom. 104, No. 3, 419--441 (2016; Zbl 1432.53095)] the authors introduced the notion of flat tangent directions as a candidate for the distinguished tangent directions defined by the Iitaka fibration. In this paper they construct an example to show that in general this is not the case and define the more refined notion of truly flat tangent directions. The generic codimension of the vector space of truly flat directions defines a differential geometric invariant \(r^-_{tf}\), the authors prove that \[ n(X) \geq r^-_{tf} \geq \nu(X), \] where \(n(X)\) is the nef dimension and \(\nu(X)\) is the numerical dimension of \(X\). If the abundance conjecture holds, then \(n(X)=\nu(X)\), so the truly flat directions identify the correct subspaces. As an application the authors prove a decomposition theorem for manifolds with semi-negative holomorphic sectional curvature in terms of the nef dimension.