Finite rigid sets in curve complexes of nonorientable surfaces (Q2181574)

In this paper, the authors extend a previous result of Aramayona-Leininger about the simplicial structure of the curve complex of an orientable surface [\textit{J. Aramayona} and \textit{C. J. Leininger}, J. Topol. Anal. 5, No. 2, 183--203 (2013; Zbl 1277.57017)] to the nonorientable case. Ilbira-Korkmaz's result is as follows: if \(S_{g,n}\) is the nonorientable surface of genus \(g\) with \(n\) holes, then there is a finite subcomplex of its curve complex that is rigid when \(g+n\ne 4\). The proof is moreover explicit and constructive, locating a specified rigid set in each of these cases. The paper is written in a pleasant style, with nicely labelled pictures that helpfully guide the reader through the constructions and arguments. The \textit{curve complex} \(\mathcal{C}(S)\) of a surface \(S\), introduced by \textit{W. J. Harvey} [Ann. Math. Stud. 97, 245--251 (1981; Zbl 0461.30036)], is a simplicial complex whose vertices are isotopy classes of simple closed curves embedded on the surface, and with simplices that span vertices representing curves that can be realized pairwise disjointly. The curve complex has grown to become an indispensable tool for analyzing the geometry and structure of mapping class groups, Teichmüller spaces, and hyperbolic 3-manifolds [\textit{B. H. Bowditch}, Invent. Math. 171, No. 2, 281--300 (2008; Zbl 1185.57011); \textit{J. F. Brock} et al., Ann. Math. (2) 176, No. 1, 1--149 (2012; Zbl 1253.57009); \textit{J. L. Harer}, Invent. Math. 84, 157--176 (1986; Zbl 0592.57009); \textit{H. A. Masur} and \textit{Y. N. Minsky}, ibid. 138, No. 1, 103--149 (1999; Zbl 0941.32012); \textit{K. Rafi}, Geom. Funct. Anal. 17, No. 3, 936--959 (2007; Zbl 1129.30031)]. Most germane to the subject of this paper, Ivanov showed that when \(S\) is orientable, apart from some low-complexity exceptions that can be computed explicitly, the automorphism group of \(\mathcal{C}(S)\) is isomorphic to \(\mathrm{Mod}(S)\), the mapping class group of \(S\) [\textit{N. V. Ivanov}, Int. Math. Res. Not. 1997, No. 14, 651--666 (1997; Zbl 0890.57018)]. The second author extended this to the nonorientable case in a joint work with \textit{F. Atalan} [Groups Geom. Dyn. 8, No. 1, 39--68 (2014; Zbl 1306.57013)]. A map between simplicial complexes is \textit{locally injective} when its restriction to the star of a vertex is injective; equivalently, the map induces a graph homomorphism on the \(1\)-skeleton. A subcomplex \(Y\subset X\) is \textit{rigid} if any locally injective map \(Y\to X\) is given by the restriction of a global automorphism of \(X\) to the subcomplex \(Y\). Shackleton showed that the entire complex \(\mathcal{C}(S)\) is rigid [\textit{K. J. Shackleton}, Pac. J. Math. 230, No. 1, 217--232 (2007; Zbl 1165.57017)] (cf. [\textit{E. Irmak} and \textit{J. D. McCarthy}, Turk. J. Math. 34, No. 3, 339--354 (2010; Zbl 1206.57018); \textit{E. Irmak}, Algebr. Geom. Topol. 9, No. 4, 2055--2077 (2009; Zbl 1203.57008)]). Subsequently, in response to a question of Lars Louder, Aramayona-Leininger produced finite rigid subcomplexes of \(\mathcal{C}(S)\) [\textit{J. Aramayona} and \textit{C. J. Leininger}, loc. cit.]. This is somewhat surprising as \(\mathcal{C}(S)\) is locally-infinite, and one has the feeling that there is plenty of room inside \(\mathcal{C}(S)\) to move finite subcomplexes around without inducing the rigid action on \(\mathcal{C}(S)\) specified by a mapping class. Aramayona-Leininger's rigid subcomplexes are quite attractive. For instance, in response to [loc. cit., Question 2], Birman-Broaddus-Menasco showed that, when the surface \(S\) has genus zero, the rigid subcomplexes of Aramayona-Leininger generate the homology of \(\mathcal{C}(S)\) as a module over \(\mathrm{Mod}(S)\) [\textit{J. Birman} et al., J. Topol. Anal. 7, No. 1, 47--71 (2015; Zbl 1308.57009)]. Later, Gaster-Greene-Vlamis used Aramayona-Leininger's rigid set to produce lower bounds for the chromatic number of \(\mathcal{C}(S)\) [\textit{J. Gaster} et al., Forum Math. Sigma 6, Paper No. e17, 42 p. (2018; Zbl 1405.57008)]. In the wake of Aramayona-Leininger's work, a flurry of different examples of rigid subcomplexes for \(\mathcal{C}(S)\), and some of its close relatives (notably the pants complex and the arc complex), were produced [\textit{J. Aramayona} and \textit{C. J. Leininger}, Pac. J. Math. 282, No. 2, 257--283 (2016; Zbl 1342.57015); \textit{E. Irmak}, ``Exhausting curve complexes by finite rigid sets on nonorientable surfaces'', Preprint, \url{arXiv:1906.09913}; \textit{R. Maungchang}, Topology Appl. 237, 37--52 (2018; Zbl 1387.57034); \textit{J. Hernández Hernández} et al., ``Finite rigid subgraphs of pants graphs'', Preprint, \url{arXiv:1907.12734}; \textit{E. Shinkle}, ``Finite rigid sets in arc complexes'', Preprint, \url{arXiv:1909.08762}; \textit{B. Szepietowski}, ``A note on the curve complex of the 3-holed projective plane'' Preprint, \url{arXiv:1907.09042}]. The present work naturally fits into this context, exploring rigidity in \(\mathcal{C}(S)\) in the nonorientable setting. There are some subtle features of the curve complex in the nonorientable setting that may be unfamiliar to readers who have not encountered it before. For instance, in contrast to the orientable setting, the curve complex of a nonorientable surface is not pure -- there are maximal simplices of different dimensions. Such differences are exploited in the present work, and are helpfully explained clearly and concisely in an introductory section. There are two natural next steps suggested by this paper, and subsequently both have been investigated by other authors. First, in analogy with Aramayona-Leininger's construction of an exhaustion of \(\mathcal{C}(S)\) by finite rigid sets in the orientable setting, Irmak constructed such an exhaustion in the nonorientable setting when \(g+n\ne 4\) [\textit{J. Hernández Hernández} et al., loc. cit.]. Second, Szepietowski has constructed an exhaustion by finite rigid sets for the nonorientable surface with \(g=1\) and \(n=3\) [\textit{B. Szepietowski}, loc. cit.]. This means that the remaining cases for which rigid sets have yet to be explicitly written down are \((g,n)\in\{(2,2),(3,1),(4,0)\}\) in the nonorientable setting. (Note that the case \((g,n)=(0,4)\) is in fact the orientable four-holed sphere, whose curve graph is either totally disconnected or the Farey graph, depending on the choices one makes for how to define the curve complex in low-complexity; rigidity for either is not hard to work out.) All of these investigations of combinatorial rigidity in simplicial complexes associated to surfaces are fascinating, and there are other related questions that can be explored. For instance, what are the rigid subcomplexes of \(\mathcal{C}(S)\) (or the arc complex, pants complex, etc.)? Given a subcomplex, how can one tell whether it is rigid? Which finite subcomplexes have rigid expansions that exhaust \(\mathcal{C}(S)\) [\textit{J. Aramayona} and \textit{C. J. Leininger}, loc. cit.; \textit{J. Hernández Hernández}, Glasg. Math. J. 61, No. 1, 195--230 (2019; Zbl 1481.57035)]?