Building hyperbolic metrics suited to closed curves and applications to lifting simply (Q1707245)

Let \(\mathcal{S}\) be a surface of finite type and negative Euler characteristic, and let \(\gamma\) be a closed curve on \(\mathcal{S}\), assumed not to be simple; that is, \(\gamma\) has positive self-intersection number \(i(\gamma, \gamma)\). A celebrated result of [\textit{P. Scott}, J. Lond. Math. Soc., II. Ser. 17, 555--565 (1978; Zbl 0412.57006)] guarantees that there is a finite degree cover of \(\mathcal{S}\) where \(\gamma\) lifts to a simple closed curve; when this happens, it is said that \(\gamma\) \textit{lifts simply}. The degree \(\mathrm{deg}(\gamma)\) is defined to be the minimum degree of a cover where \(\gamma\) lifts simply. The questions addressed in the present paper are concerned with how \(\mathrm{deg}(\gamma)\) grows as \(\gamma\) gets more complicated, which could be interpreted to mean either ``has more self-intersections'' or ``grows longer.'' That is, how does \(f_\mathcal{S}(k) = \max{ \{ \mathrm{deg}(\gamma) \mid i(\gamma, \gamma) \leq k \} }\) grow as \(k\) grows? Given a metric \(\rho\) on \(\mathcal{S}\) and a length \(L\), if \(\ell_\rho(\gamma)\) denotes the length of \(\gamma\) in the metric \(\rho\), how does \(f_\rho(L) = \max{ \{ \mathrm{deg}(\gamma) \mid \ell_\rho(\gamma) \leq L \} }\) grow as \(L\) grows? The main result of the present paper is that \(f_\mathcal{S}\) and \(f_\rho\) are related, in the following sense: given \(\gamma\) with \(i(\gamma,\gamma) \leq k\), one may construct a metric \(\rho\) on \(\mathcal{S}\) such that \(\ell_\rho(\gamma) \leq C_3 \sqrt{k}\), where \(C_3\) is a constant depending on \(\mathcal{S}\), but not on \(\gamma\). This result is applied to give upper bounds for \(f_\mathcal{S}(k)\) and \(f_\rho(L)\), complementing lower bounds given by the second author [\textit{J. Gaster}, ``Lifting curves simply'', Int. Math. Res. Not., Vol. 2016, Issue 18, 5559--5568 (2016)]. Specifically, when \(\mathcal{S}\) does not have cusps, it is shown that there is a constant \(C_4\) depending only on \(\mathcal{S}\) such that \(f_\mathcal{S}(k) \leq C_4 \, k\). This complements an existing linear lower bound to show that \(f_\mathcal{S}(k)\) grows linearly. When \(\mathcal{S}\) has cusps, the upper bound obtained is that there exist \(L_0 > 0\), and a constant \(C_2\), such that for any \(L \geq L_0\), we have \(f_\rho(L) \leq C_2 \, L \, e^{L/2}\). The best known lower bound at present is \(C_1 \, e^{L/2} \leq f_\rho(L)\) for a constant \(C_1\). The method used to prove the main result, based on work of the fourth author [\textit{J. Sapir}, ``Orbits of non-simple closed curves on a surface'', Preprint, \url{arXiv:1602.09093}], involves choosing a ``well-balanced'' pants decomposition, with control over how often \(\gamma\) intersects the cuffs. Adding seams of the pants creates a graph in \(\mathcal{S}\), in which a cycle can be chosen which is freely homotopic to \(\gamma\). By using Dehn twists, the combinatorics of the graph are optimized, so that the metric \(\rho\) may be constructed by judiciously choosing lengths for the pants curves.