Basmajian-type inequalities for maximal representations (Q2213945)

The authors prove several inequalities for locally symmetric space associated to an appropriate symplectic representation \(\rho: \pi_1(\Sigma) \to \operatorname{Sp}(n, \mathbb{R})\) of the fundamental group of a surface \(\Sigma\) which generalize the Basmajian identity between the length of the boundary and the length of orthogeodesics in the Teichmüller theory of module spaces of hyperbolic surfaces. Let \(\Sigma\) be a hyperbolic surface with nonempty geodesic boundary \(\partial \Sigma\), and let \(O(\Sigma)\) denote the set of unoriented orthogeodesics of \(\Sigma\) (i.e., the set of geodesics with endpoints on the boundary and orthogonal to it). Then the length of the boundary is related to lengths of orthogeodesics as follows: \[\ell(\partial \Sigma) = 4\sum_{\alpha \in O(\Sigma)} \log \coth (\ell(\alpha)/2).\] It is known that the Teichmüller space is parametrized by maximal symplectic representations \(\pi_1(\Sigma) \to \operatorname{PSp}(1, \mathbb{R})\), that is injective representations which maximize the Euler class. Similarly, an injective representation \(\rho : \pi_1(\Sigma) \to \Gamma = \rho(\pi_1(\Sigma)) \subset \operatorname{Sp}(n, \mathbb{R})\) is called maximal if it maximizes the Toledo class which is a generalization of the Euler class. Any such representation defines the quotient \( M= \Gamma\backslash \mathcal{X}\) of the Hermitian symmetric space \(\mathcal{X} = \operatorname{Sp}(n,\mathbb{R})/\operatorname{U}(n)\) of the admissible complex structures in the symplectic space \((\mathbb{R}^{2n}, \omega )\), which is identified with the Ziegel domain \(\mathcal{X} = S^2\mathbb{R}^n + i \mathcal{P}\) associated with the cone \(\mathcal{P}\) of positively defined matrices in \(S^2\mathbb{R}^n\). The manifold \(M\) is a smooth locally symmetric space. The authors define three \(\operatorname{Sp}(2n,\mathbb{R})\)-invariant distances on the symmetric space \(\mathcal{X}\): the determinant Finsler distance \(d_F\), the Riemannian distance \(d_R\) and a Weyl chamber-valued distance \(d_W\), used in the formulation of the inequalities. Let \( d_F(\gamma)\), \(d_R(\gamma)\), \(d_W(\gamma)\) be the length of a path in \(\mathcal(X)\) w.r.t. these distances. Using these distances, the authors define three lengths of an element \(\gamma \in \Gamma\) as the infimum of the length between \(X \in \mathcal{X}\) and its shift \(\gamma(X)\). The boundary \(\partial \mathcal{X}\) contains a unique closed \(\operatorname{Sp}(n,\mathbb{R})\)-orbit \(L(\mathbb{R}^{2n})\) which is identified with the space of Lagrangian subspaces in \(\mathbb{R}^{2n}\). Two transversal Lagrangian subspaces \(a, b \in L(\mathbb{R}^{2n})\) define a unique maximal-rank totally geodesic submanifold \(Y_{a,b}\), called \(R\)-tube, which is considered as a generalized geodesic, joining the boundary points \(a,b\). Generalizing the notion of orthogeodesics, the authors define the notion of orthotube and its length and prove some inequalities, which are a generalization of the Basmajian identity.