Intersection of curves on surfaces and their applications to mapping class groups (Q5962656): Difference between revisions

Let \(S\) be a compact oriented surface \(S\), and \(\ast_0,\ast_1\in\partial{S}\) points on the boundary. The set of homotopy classes of loops on \(S\) from \(\ast_0\) to \(\ast_1\) is denoted by \(\Pi S(\ast_0,\ast_1)\), and the set of homotopy classes of free loops on \(S\) by \(\hat{\pi}'(S)\). The authors define a \(\mathbb{Q}\)-linear map \[ \mu:\mathbb{Q}\Pi S(\ast_0,\ast_1)\to\mathbb{Q}\Pi S(\ast_0,\ast_1)\otimes\mathbb{Q}\hat{\pi}'(S), \] by using the self intersection of paths. This map is closely related to the Goldman-Turaev Lie bialgebra \(\mathbb{Q}\hat{\pi}'(S)\) and another map \[ \sigma:\mathbb{Q}\hat{\pi}'(S)\otimes\mathbb{Q}\Pi S(\ast_0,\ast_1)\to\mathbb{Q}\Pi S(\ast_0,\ast_1), \] defined by the authors in their previous paper [Quantum Topol. 5, No. 3, 347--423 (2014; Zbl 1361.57027)]. By investigating the properties of \(\mu\) the authors arrive at the notion of a bimodule of a Lie bialgebra and they prove that \(\mathbb{Q}\Pi S(\ast_0,\ast_1)\) is a \(\mathbb{Q}\hat{\pi}'(S)\)-bimodule with respect to \(\sigma\) and \(\mu\). Then they discuss applications of this structure to the study of the mapping class group \(\mathcal{M}(S,\partial S)\) of \(S\). With each unoriented loop \(C\) on \(S\) one can associate the generalized Dehn twist \(t_C\), first introduced by the second author in [J. Topol. Anal. 5, No. 3, 271--295 (2013; Zbl 1282.57025)], as an element of a certain enlargement of \(\mathcal{M}(S,\partial S)\). When \(C\) is simple, \(t_C\) is the usual Dehn twist along \(C\). A natural question is whether \(t_C\) is realized by a diffeomorphism, i.e. whether \(t_C\) is in \(\mathcal{M}(S,\partial S)\). The authors provide a criterion for the realizability in terms of \(\mu\), and they use it to construct new examples of generalized Dehn twists not realized by diffeomorphisms. In [``Grupoid-theoretical methods in the mapping class groups of surfaces'', Preprint, \url{arXiv:1109.6479}] the authors defined the geometric Johnson homomorphism \(\tau\) as a generalization of the classical construction of the Johnson homomorphisms to arbitrary compact oriented surfaces with non-empty boundary. The target of \(\tau\) is a certain Lie subalgebra of the completion of \(\mathbb{Q}\hat{\pi}'(S)\). In the present paper they provide a constraint on the image of \(\tau\) by showing that it is contained in the kernel of the Turaev cobracket. This can be seen as a generalization, for surfaces with non-connected boundary, of \textit{S. Morita}'s results [Duke Math. J. 70, No. 3, 699--726 (1993; Zbl 0801.57011)].