Criteria for the divergence of pairs of Teichmüller geodesics (Q849380)

Let \(S\) be an oriented surface of genus \(g\) with \(n\) punctures and \(3g- 3+ n\geq 1\). Let \(\mathcal T(S)\) denote the Teichmüller space of \(S\) with the Teichmüller metric \(d(\cdot, \cdot)\). A basic question in geometry is to study the long-term behaviour of geodesics. In this paper, the authors study the question of when a pair of geodesic rays \(X_1(t), X_2(t)\), with possibly distinct basepoints, stays bounded distance apart, and when they diverge in the sense that \(d(X_1(t), X_2(t))\to\infty\) as \(t\to\infty\). Teichmüller's theorem implies that a Teichmüller geodesic ray is determined by a quadratic differential \(q\) at the base point and that there are quadratic differentials \(q(t)\) on \(X(t)\) along the ray found by stretching along the horizontal trajectories of \(q\) and contracting along the vertical trajectories. Many cases of the question of divergence of rays are already known. It is a general principle that the asymptotic behavior of the ray is determined by the properties of the vertical foliation of \(q\). The first instance is the case when the quadratic differentials \(q_1,q_2\) defining the geodesic rays \(X_1(t), X_2(t)\) are Strebel differentials. This means that their vertical trajectories are closed and decompose the surface into cylinders. H.~Mazur showed that if the homotopy classes of the cylinders for \(q_1\) coincide with those of \(q_2\), then \(X_1(t)\), \( X_2(t)\) stay bounded distance apart. In particular, this showes that the Teichmüller metric is not negatively curved in the sense of Busemann. A second known case is the case when the vertical foliations of \(q_1\), \( q_2\) are uniquely ergodic. In that case the rays also stay bounded distance apart. The next possibility is that the vertical foliations of \(q_1\), \( q_2\) are topologically equivalent, have a minimal component, but are not uniquely ergodic. (It is well known that for any quadratic differential, the vertical trajectories decompose the surface into cylinders and subsurfaces in which every trajectory is dense). In that case, in each minimal component, there exist a finite number of mutually singular ergodic measures, and any transverse measure is a convex combination of ergodic measures. N.~Ivanov showed that if the transverse measures of \(q_1\), \( q_2\) in these minimal components are absolutely continuous with respect to each other, then the rays stay bounded distance apart. In this paper, the authors prove the converse. Let \(q_2\), \( q_2\) be quadratic differentials on \(X_1\) and \(X_2\) with vertical foliations \(\left[F_{q_1}^\nu,|dx_1|\right]\) and \(\left[F_{q_2}^\nu,|dx_2|\right]\) and determining rays \(X_1(t), X_2(t)\). The main result of the paper is then {Theorem A.} Suppose \(F_{q_1}^\nu\) and \(F_{q_2}^\nu\) are topologically equivalent. Suppose there is a minimal component \(\Omega\) of the foliations \(F_{q_1}^\nu\) with ergodic measures \(\nu_1,\dots, \nu_p\) such that, restricted to \(\Omega\), \(|dx_1| =\sum_{i=1}^p a_i\nu_i\), \(|dx_2| =\sum_{i=1}^p b_i\nu_i\), and there is some index \(i\) such that either \(a_i = 0\) and \(b_i > 0\) or \(a_i > 0\) and \(b_i = 0\). Then the rays \(X_1(t)\) and \(X_2(t)\) diverge. In particular, this holds when the transverse measures are distinct ergodic measures. The last possibility is that the vertical foliations of \(q_1\), \( q_2\) are not topologically equivalent. If the geometric intersection of the vertical foliations is nonzero, then the rays diverge. The authors prove {Theorem B.} Suppose \(q_1\), \(q_2\) are quadratic differentials such that the vertical foliations \(\left[F_{q_1}^\nu,|dx_1|\right]\) and \(\left[F_{q_2}^\nu,|dx_2|\right]\) are not topologically equivalent, but \[ i\Big(\left[F_{q_1}^\nu,|dx_1|\right], \left[F_{q_2}^\nu,|dx_2|\right]\Big)=0. \] Then the rays \(X_1(t)\) and \(X_2(t)\) diverge. These theorems together with the previously known results completely answer the question of divergence of rays. The authors also prove {Theorem C.} Let \(\nu_1\,\dots, \nu_p\) be a maximal collection of ergodic measures for a minimal foliation \([F,\mu]\). Then there is a sequence of multicurves \(\gamma_n =\{\gamma_n^1,\dots,\gamma_n^k\}\) such that \(\gamma_n^j\to[F,\nu_j]\) in the projective space of measured foliations. In other words, any two topologically equivalent measured foliations can be ap\-pro\-xi\-ma\-ted by a sequence of multicurves, with possibly different weights. This result settles a question asked by Moon Duchin.