Elliptic actions on Teichmüller space (Q2319839)

Let \(S\) be a surface of finite topological type, let \(\mathcal{MCG}(S)\) be the mapping class group of \(S\), that acts on the Teichmüller space \(\mathcal{T}(S)\). The solution of the Nielsen Realization Problem -- see [\textit{S. P. Kerckhoff}, Ann. Math. (2) 117, 235--265 (1983; Zbl 0528.57008)] for the case of a closed surface -- asserts that any finite subgroup \(H<\mathcal{MCG}(S)\) has a non-empty set of fixed points \(\text{Fix}(H)\). This is true for the action of a finite-group on any simply-connected complete manifold that is negatively curved, by an argument of Cartan that uses the notion of a {barycenter} of an orbit -- see for example, Chapter I, Theorem 13.5 of [\textit{S. Helgason}, Differential geometry, Lie groups, and symmetric spaces, reprint with corrections of the 1978 original (2001; Zbl 0993.53002)]. However, the Teichmüller metric \(d_T\) on \(\mathcal{T}(S)\) is \textit{not} negatively curved, and the Weil-Petersson metric \(d_{WP}\) on \(\mathcal{T}(S)\), though negatively curved, is \textit{incomplete}, which makes Kerckhoff's result striking. This paper studies the structure of the set of \textit{\(R\)-almost fixed points} in \(\mathcal{T}(S)\), that are moved by the action of the finite subgroup \(H\) by a distance bounded by \(R>0\) in the Teichmüller metric: \[ \text{Fix}_R^T(H) = \{X \in \mathcal{T}(S)\mid \text{diam}_T(H\cdot X) <R\}. \] The main result of the paper is that \(\text{Fix}_R^T(H)\) is \(R^\prime\)-close to the fixed-point set \(\text{Fix}(H)\), for some \(R^\prime>0\) that only depends on \(R\) and the surface \(S\) (see Theorem 1.2 or 5.6). A key ingredient of the proof is a coarse distance formula in the augmented marking complex \(\mathcal{AM}(S)\), that the author had shown is a quasi-isometric model of \((\mathcal{T}(S), d_T)\), in previous work -- see [J. Lond. Math. Soc., II. Ser. 94, No. 3, 933--969 (2016; Zbl 1360.30038)]. The proof also uses extensions of technical results arising from the ``hierarchy machinery'' in [\textit{H. A. Masur} and \textit{Y. N. Minsky}, Geom. Funct. Anal. 10, No. 4, 902--974 (2000; Zbl 0972.32011)], and [\textit{J. Tao}, ibid. 23, No. 1, 415--466 (2013; Zbl 1286.57020)], to the setting of \(\mathcal{AM}(S)\). The work of J. Tao, together with the main result, is used to prove the existence of ``coarse barycenters'' of the orbit of any finite-order element of \(\mathcal{MCG}(S)\) (Theorem 1.3 or 6.1). Finally, in contrast to these results reminiscent of negative curvature, the author gives examples where \(\text{Fix}_R^T(H)\) is \textit{not} quasiconvex (for some \(R>0\), some surface \(S\) and finite subgroup \(H<\mathcal{MCG}(S)\)) by adapting constructions in [\textit{K. Rafi}, Geom. Topol. 18, No. 5, 3025--3053 (2014; Zbl 1314.30082)]. The author concludes with a discussion of how these examples are, in fact, expected to be part of a common phenomenon. The paper is very well-written, and the technical arguments are well motivated.