Relations in the tautological ring of \({\mathcal M}_g\) (Q2566607)

The paper studies various relations in the tautological ring of the moduli space \({\mathcal M}_g\) of (non-singular) genus-\(g\) curves. As a consequence, the author proves the Faber conjecture on the generators of the ring. In more details, let \(\psi\in A^1({\mathcal M}_g)\) be the first Chern class of the relative dualizing sheaf. Define \(k_a:=\pi_*(\psi^{a+1})\in A^a({\mathcal M}_g)\). The tautological ring \(R^*({\mathcal M}_g)\) is the \(\mathbb Q\)-subring of \(A^*({\mathcal M}_g)\), generated by the \(k\)-classes. The author proves that this ring is in fact generated by a subset: \(k_1,\dots,k_{[\frac{g}{3}]}\) of classes (thus significantly improving the well-known result by Mumford). This fact is proved by giving explicit relations in the tautological ring. Rather surprisingly, the relations have one simple generating function, an exponential of a linear combination of the \(k_a\) classes. The relations arise from the following geometric idea. One considers the locus of genus-\(g\) curves that can be written as the degree \(d\) cover of \(\mathbb P^1\). Let \(w_{d,g}\) be the Chow class of this locus. Then the class \(k_{b-1}w_{d,g}\) can be computed in various ways (by playing with the marked points). Comparison of the results gives the relations among the classes. As discussed in the appendix, these relations can also be obtained from the relations in the tautological ring of fibred product of \({\mathcal E}_g\) with itself. For a very simple and readable introduction to this area cf. \textit{R. Vakil} [Notices Am. Math. Soc. 50, No. 6, 647--658 (2003; Zbl 1082.14033)].