Recurrence and vectors escaping to infinity for Lipschitz operators (Q6063619)

Let \((M,d)\) be a \textit{pointed metric space}, namely, a pair consisting of a metric space \((M,d)\) and a distinguished point \(0_M\in M\), and \(\mathrm{Lip}_0(M)\) denote the set of Lipschitz functions \(f:M\to \mathbb R\) which vanish at \(0_M\). The space \(\mathrm{Lip}_0(M)\) equipped with the norm \(\mathrm{Lip}(\cdot)\) is a dual Banach space. The map \[ \delta:M\to \mathrm{Lip}_0(M)^*,\quad \delta(\varphi)(x):=\phi(x),\quad x\in M,\ \varphi\in\mathrm{Lip}_0(M), \] is a nonlinear isometry: \(\|\delta(x)-\delta(y)\|=d(x,y)\) for all \(x,y\in M\). Then the \emph{Lipschitz-free} space of \(M\) is \[ \mathcal F(M):=\overline{\mathrm{span}}\left(\delta(x):\, x\in M\right). \] the closure being taken with respect to the norm topology on \(\mathrm{Lip}_0 (M)^*\). It is known [\textit{N. Weaver}, Lipschitz algebras. 2nd edition. Hackensack, NJ: World Scientific (2018; Zbl 1419.46001) ] that, if \((M,d)\) and \((N,\rho)\) are pointed metric spaces and \(f\in \mathrm{Lip}_0(M,N)\) (namely \(f:M\to N\) is a Lipschitz function such that \(f(0_M)=0_N\)), then there exists a unique bounded linear operator \(\widehat{\cdot}:\mathcal F(M)\to \mathcal F(N)\) such that \[ \delta_N(f(x)) = \widehat f(\delta_M(x)),\qquad \|\widehat f\|=\mathrm{Lip}(f). \] Note that, if \(M=N\), then \(\widehat f\in\mathcal L(\mathcal F(M))\). If \(f\in\mathrm{Lip}_0(M,M)\) let \(R_f\) denote the recurrent points of \(f\), and \(A_f\) denote the set of points in \(M\) whose \(f\)-orbit escapes to infinity (with respect to \(0_M\)). An operator \(T\in\mathcal L(\mathcal F(M))\) is called \textit{wild} if \(X=A_T\cup R_T\) and \(\mathrm{int}(R_T)\neq \emptyset\neq \mathrm{A_T}\). (It follows that \(A_T\) is nonempty but not dense; see [\textit{G. T. Prăjitură}, Theta Ser. Adv. Math. 12, 145--154 (2010; Zbl 1216.47008)]/) One of the main results of this paper is that there does \textit{not} exist a pointed metric space \((M,d)\) and \(f\in\mathrm{Lip}_0(M,M)\) such that \(\widehat{f}\in\mathcal L(\mathcal F(M))\) is wild. The authors achieve this result by studying the set of recurrent vectors \(R_{\widehat f}\), and the set of vectors escaping to infinity \(A_{\widehat f}\). For example, the authors prove that, if \((M,d)\) is complete and \(\mathrm{int}(R_{\widehat f})\neq \emptyset\), then \(R_{f}=M\) and \(R_{\widehat f}\) is dense in \(\mathcal F(M)\). Finally, again for complete \((M,d)\), the authors characterize the maps \(f\in\mathrm{Lip}_0(M,M)\) for which \(\widehat f\) is \textit{rigid} in the sense of [\textit{G. T. Prăjitură}, Theta Ser. Adv. Math. 12, 145--154 (2010; Zbl 1216.47008)].