Signatures, monopoles and mapping class groups (Q1128277): Difference between revisions

The study of the signature invariant \(\sigma (X)\) of surface bundles \(X \to B\) over surfaces \(B\) has a long history, and can be traced back to the celebrated work of Atiyah and Kodaira where such bundles with nontrivial signatures were first constructed. In this short, well-written article, the author investigates the bounds on the signature \(\sigma(X)\) by an appropriate multiple of the Euler number \(\chi (X)\). Using the recent results of Taubes on the Seiberg-Witten invariant of symplectic 4-manifolds, he proves that for a general aspherical surface bundle without any complex analytic assumption \(2| \sigma (X)| \leq\chi (X)\). However with the complex analytic assumption there is the Miyaoka-Yau inequality \(3|\sigma(X)|\leq \chi(X)\), and in a remark the author stated his belief that this Miyaoka-Yau inequality should hold for a general surface bundle as well. Note a surface bundle is determined by a monodromy representation \(\rho:\pi_1 (B)\to \Gamma_h\), where \(\Gamma_h\) is the mapping class group of the fiber \(F\). An equivalent formulation of the aforementioned result is that if a representation \(\rho: \pi_1(B) \to\text{Sp} (2h,R)\) factors through the mapping class group \(\Gamma_h\), then the first Chern number of the associated flat bundle satisfies the inequality \(|\langle c_1(\rho), [B]\rangle |\leq 1/2 (g-1) (h-1)\), where \(g\) is the genus of the base.