Inducing maps between Gromov boundaries (Q727557)

The authors introduce radial functions on metric spaces which induce continuous maps on their Gromov boundaries. Let \(X\) and \(Y\) be metric spaces. A function \(f: X\to Y\) is called large-scale Lipschitz if there exist \(\ell_{1}, \mu_{1} > 0\) so that \(d(f(x), f(y))\leq \ell_{1}\cdot d(x, y)+\mu_{1}\) , for every \(x, y\in X\). A large-scale Lipschitz function \(f: X\to Y\) of Gromov hyperbolic spaces is called visual if \((x_{n}, y_{n})_{a}\to \infty \,\Longrightarrow\,(f(x_{n}), f(y_{n}))_{f(a)}\to \infty\) for some \(a\in X\), where \((x, y)_{a}\) denotes the Gromov product with respect to \(a\in X\). Visual functions induce continuous maps on the boundaries. Theorem 2.8. Suppose \(f: X\to Y\) is a large-scale Lipschitz function of proper hyperbolic spaces. If \(f: X\to Y\) is visual, then \(f\) induces a function \(\tilde{f}:X\cup \partial X\to Y\cup \partial Y\) such that \(\tilde{f}(\partial X)\subset \partial Y\) and \(\tilde{f}\) is continuous at all points of \(\partial X\). Conversely, if \(f\) induces a function \(\tilde{f}:X\cup \partial X\to Y\cup \partial Y\) such that \(\tilde{f}(\partial X)\subset \partial Y\) and \(\tilde{f}\) is continuous at all points of \(\partial X\), then \(f\) is visual. Here, \(\partial X\) is the (sequential) boundary of a proper, hyperbolic geodesic space defined as \[ \partial X = \{ (x_{n})\mid (x_{n})\text{ sequence in }X, \liminf (x_{i}, x_{j})_{a} = \infty\}/{\sim} \] where the equivalence of sequences is given by \[ (x_{n}) \sim (y_{n})\,\Longleftrightarrow\, \liminf (x_{i}, y_{j})_{a} = \infty. \] Let \(X\) and \(Y\) be metric spaces. A function \(f: X\to Y\) is called (\(\ell_{2}, \mu_{2}\))-radial with respect to \(x_{0}\in X\) if it is large-scale Lipschitz and there exist \(\ell_{2}, \mu_{2} >0\) so that for every finite geodesic \(\gamma :[0, M]\to X\), \(\gamma (0) = x_{0}\) the following condition holds: \(\ell_{2}\cdot d(x, y)-\mu_{2}\leq d(f(x), f(y))\), for every \(x, y\in \mathrm{Im}\gamma \). A hyperbolic space \(X\) is called visual if there is a constant \(D>0\) (depending on the base point \(a\) of \(X\)) such that for each \(x\in X\) there is an infinite geodesic ray \(\xi \) at \(a\) such that \((x, \xi )_{a} > d(x, a)-D\). A sufficient condition for a function \(f: X\to Y\), where \(X\) is visual, to be radial is given in Proposition 3.6: A metric \(d\) on \(\partial X\) is called visual if there are constants \(K, C>1\) such that \(\frac{K^{-(p, q)_{a}}}{C}\leq d(p, q)\leq C\cdot K^{-(p, q)_{a}}\). A function \(f:X\to Y\) of metric spaces is called Hölder if there are constants \(A, B\) such that \(d(f(x), f(y))\leq B\cdot d(x, y)^{A}\), for every \(x, y\in X\). Suppose \(X\) and \(Y\) are proper Gromov hyperbolic spaces whose boundaries are equipped with visual metrics. If \(g : \partial X \to \partial Y\) is a Hölder map and \(X\) is visual, then there is \(M>0\) such that each radial extension \(f:X\to Y\) of \(g\) with parameter \(A\leq M\) is a radial function. A converse of the above theorem is the following: Theorem 5.5. Suppose that \(X\) and \(Y\) are proper geodesic hyperbolic spaces. If \(f:X\to Y\) is radical, then \(f\) induces a Hölder map between boundaries \(\tilde{f}:\partial X\to \partial Y\) when equipped with visual metrics. A function \(f:X\to Y\) is called coarsely surjective if \(f(X)\) is dense in \(Y\) in the sense: There exists a parameter \(S \geq 0\) with the property that for every \(y\in Y\) there exists an \(x\in X\) such that \(d(f(x), y)\leq S\). There is an interplay of the coarse surjectivity of a function \(f:X\to Y\) and the surjectivity of the induced function \(\tilde{f}:\partial X\to \partial Y\); see Theorems 6.3 and 6.4 in the paper. A function is called \(n\)-to-\(1\) if its point-inverses contain at most \(n\) points. The authors examine conditions under which a function \(f:X\to Y\) and the induced \(\tilde{f}:\partial X\to \partial Y\) are \(n\)-to-\(1\); see Theorems 7.1, 7.4, 9.2 and 9.3 in the paper. As a consequence the following theorem is proved Theorem 7.5. Suppose that \(X\) and \(Y\) are proper \(\delta\)-hyperbolic spaces and \(f:X\to Y\) is a radial function. If \(f\) is coarsely \((n+1)\)-to-\(1\) and coarsely surjective then \(\dim (\partial Y) \leq \dim (\partial X)+n\). Moreover, if \(X\) and \(Y\) are hyperbolic groups then \(\operatorname{asdim} (Y) \leq \operatorname{asdim} (X)+n\). This improves a result in [\textit{T. Miyata} and the second author, Fundam. Math. 223, No. 1, 83--97 (2013; Zbl 1288.54025)].