The number of fiberings of a surface bundle over a surface (Q1748438)

Three-manifolds fibering as a surface bundle over the circle often do so non-uniquely. In [\textit{W. P. Thurston}, Mem. Am. Math. Soc. 339, 99--130 (1986; Zbl 0585.57006)], Thurston gave a precise description in terms of the homology of the manifold, and established a dichotomy: a fibered 3-manifold either admits a unique surface bundle structure, or else admits infinitely many, necessarily with fibers of unbounded genera. The paper under review investigates the phenomenon of non-unique surface bundle structures on 4-manifolds. The starting point for the work is a theorem of \textit{F. E. A. Johnson} [Arch. Math. 73, No. 2, 81--89 (1999; Zbl 1003.20036)], which establishes that a 4-manifold \(E\) with Euler characteristic \(\chi(E) \neq 0\) has only finitely many structures as a surface bundle over a surface. In [Algebr. Geom. Topol. 15, No. 6, 3613--3652 (2015; Zbl 1334.57029)], the reviewer developed certain techniques for constraining the number of surface bundle structures on a given 4-manifold. These are formulated in terms of the multiplicative structure of \(H^*(E)\). These ideas are expanded on and improved in the paper under review. There are several classes of 4-manifolds known to admit multiple surface bundle structures. Both unbranched and branched covers of products of surfaces can exhibit this behavior, leading to at least two distinct bundle structures inherited from the projection onto either factor. There are also examples admitting four or more structures, as constructed by the reviewer in [Geom. Topol. 19, No. 5, 2901--2923 (2015; Zbl 1329.57033)]. The author studies examples from each of these classes and proves that the ``expected'' number of fiberings (e.g. two in the case of a (branched) cover of a product) is in fact sharp, so that there are no ``hidden'' fiberings.