Moduli spaces of Riemann surfaces as Hurwitz spaces (Q6135871)

Let \(\mathfrak M_{g,1}\) be the moduli space of Riemann surfaces of genus \(g\ge 0\) with one boundary component (or, equivalently, one marked and directed point). Using the theory of generalised Hurwitz spaces [\textit{A. Bianchi}, ``Partially multiplicative quandles and simplicial Hurwitz spaces'', Preprint, \url{arXiv:2106.09425}; Geom. Dedicata 217, No. 5, Paper No. 84, 56 p. (2023; Zbl 1522.18017); ``Deloopings of Hurwitz spaces'', Preprint, \url{arXiv:2107.13081}], the present work establishes a combinatorial model for \(\mathfrak M_{g,1}\), and use it to give an alternative proof of the Mumford conjecture on the stable rational cohomology ring of moduli spaces [\textit{D. Mumford}, Prog. Math. 36, 271--328 (1983; Zbl 0554.14008)], which was originally proved by \textit{I. Madsen} and \textit{M. Weiss} [Ann. Math. (2) 165, No. 3, 843--941 (2007; Zbl 1156.14021)]. This is achieved in several steps: 1. The author constructs a space \(\bar{\mathcal{O}}_{g,1}[d]\), whose elements are \(d\)-fold branched covers \(S\to\mathbb C\) such that the compactified covering \(\bar S\to \mathbb C\mathrm P^1\) is a smooth Riemann surface of genus \(g\). This space naturally extends the classical Hurwitz spaces and admits a forgetful map \(\bar{\mathcal{O}}_{g,1}[d]\to \mathfrak M_{g,1}\). The first main result of the paper states that for \(2g\le d\), this map is a homotopy equivalence. 2. The space \(\bar{\mathcal{O}}_{g,1}[d]\) has a simplicial description: For the partially multiplicative quandle \(\mathfrak S_d^{\text{geo}}\) (where only products of so-called `geodesic pairs' are defined), the simplicial Hurwitz space \(\mathrm{Hur}^\Delta(\mathfrak S_d^{\text{geo}})\) as defined in the author's previous works [see references above] turns out to have a component that is both a topological manifold and homeomorphic to \(\bar{\mathcal{O}}_{g,1}[d]\). 3. The author considers, for \(d\ge 0\), the relative Hurwitz-Ran space \(\mathrm{Hur}_+(\mathcal R,\partial\mathcal R;\mathfrak S_d^{\text{geo}},\mathfrak S_d)\) and defines \(\mathbb B_\infty\) as the colimit of (one component of) them where \(d\to\infty\). Then the previous two results enable us to identify the stable homology of \(\mathfrak M_{g,1}\) with the homology of \(\Omega^2_0\mathbb B_\infty\), i.e. of one component of \(\Omega^2\mathbb B_\infty\) 4. By a standard argument in rational homotopy theory, the rational cohomology ring of \(\Omega^2_0\mathbb B_\infty\) can be identified with the polynomial algebra \(\mathbb Q[x_1,x_2,\dotsc]\); this altogether reproves the Mumford conjecture. Even more is true: It turns out that the space \(\mathbb B_\infty\) is equivalent to the infinite loop space \(\Omega^{\infty-2}\mathrm{MTSO}(2)\) appearing in the theorem of Madsen and Weiss; this is shown in the last section of the paper.