Top tautological group of \(\mathcal{M}_{g,n}\) (Q502154)

In the earliest studies of the tautological rings of moduli spaces of curves (Mumford, Faber), the focus was on the tautological rings of the spaces \(M_g\). It has since turned out to be fruitful to also consider (partial) compactifications of these spaces, and to allow for the possibility of marked points. The most well studied are the tautological rings of the spaces \(\bar M_{g,n}\) (stable \(n\)-pointed genus \(g\) curves), \(M_{g,n}^{ct}\) (curves of compact type) and \(M_{g,n}^{rt}\) (curves with rational tails). One reason to focus on these three spaces in particular is that Looijenga's theorem generalizes to these cases: the top tautological group is always one-dimensional. Moreover, expressions have been proven which allow one to express any top degree class in the tautological ring as a multiple of a chosen generator of the socle. See e.g. the survey of \textit{C. Faber} [in: Moduli spaces of Riemann surfaces. Providence, RI: American Mathematical Society (AMS); Princeton, NJ: Institute for Advanced Study (IAS). 197--219 (2013; Zbl 1275.14026)]. The paper under review considers instead the spaces \(M_{g,n}\) of \textit{smooth} \(n\)-pointed genus \(g\) curves. The main theorem is that the top tautological group \(R^{g-1}(M_{g,n})\) is a vector space of dimension \(n\). This result had originally been conjectured by Zvonkine. Vanishing above this degree is a special case of Graber-Vakil's Theorem \(\star\). The easier half of the theorem is to exhibit \(n\) linearly independent classes (this is done by an intersection number computation). The hard part is to prove that the dimension is \textit{at most} \(n\), which requires a rather complicated argument and a study of double ramification cycles. Moreover, the authors give explicit formulas which allow one to express any monomial of degree \(g-1\) in the tautological ring as a linear combination of the \(n\) chosen generators of \(R^{g-1}(M_{g,n})\).