Logarithm laws for equilibrium states in negative curvature (Q739561)

Let \(M\) be a complete connected Riemannian manifold with pinched sectional curvature at most \(-1\), and let \((g^t)_{t\in \mathbb{R}}\) be its geodesic flow. The authors consider a closed geodesic \(D_0\) in \(M\), and want to study the spiraling of geodesic lines around \(D_0\). Let \(F:T^1M\to \mathbb{R}\) be a potential (i.e, a Hölder-continuous function). Let \(\mathcal{M}\) be the set of probability measures \(m\) on \(T^1M\) invariant under the geodesic flow, for which the negative part of \(F\) is \(m\)-integrable, and let \(h_m(g^1)\) be the entropy of the geodesic flow with respect to \(m\). The pressure of the potential \(F\) is \[ P=P(F)=\sup_{m\in\mathcal{M}}\Big [h_m(g^1)+\int_{T^1M}Fdm\Big ]. \] The following result relates measure-theoretical invariants of a Gibbs measure \(m_F\) and the Patterson densities \(\mu_x^F\). Theorem 1.1. If \(m_F\) is finite and \(F\) is \(m_F\)-integrable, and if the sectional curvature of \(M\) has bounded first-order derivatives, then the Hausdorff dimension of the Patterson measure \(\mu_x^F\)(with respect to the Gromov-Bourdon visual distance on \(\partial_\infty\tilde M\)) is equal to the metric entropy of the Gibbs measure \(m_F\) (for the geodesic flow). Suppose the length of \(D_0\) is \(l_0\) and \(v_0\in T^1M\) is tangent to \(D_0\). Let \[ P_0=P(F_{|T^1D_0})=\frac{\max\{\int_0^{l_0}F(g^tv_0)dt,\int_0^{l_0}F(g^t(-v_0))dt\}}{l_0}. \] If \(m_F\) is finite, then \(P_0<P\). Let \(\epsilon_0>0\) and let \(\psi:[0,+\infty[ \to[0,+\infty[\) be a Lipschiz map, and \(E(\psi)\) be the set of \((\epsilon_0,\psi)\)-Liouville vectors around \(D_0\), that is, the set of \(v\in T^1M\) such that there exists a sequence \((t_n)_{n\in\mathbb{N}}\) in \([0,+\infty[\) converging to \(+\infty\) such that for every \(t\in [t_n,t_n+\psi(t_n)]\), the footpoint of \(g^tv_0\) belongs to the \(\epsilon_0\)-neighbourhood of \(D_0\). The main result of this paper is the following. Theorem 1.2. Assume that \(M\) is compact. If the integral \(\int_0^{+\infty}e^{\psi(t)(P_0-P)}dt\) diverges (resp. converges) then the \(m_F\)-almost every (resp. no) point of \(T^1M\) belongs to \(E(\psi)\). There are two corollaries in this paper, one of which is the following, and another is about its arithmetic application. Corollary 1.3. Under the assumptions of Theorem 1.2, for \(m_F\)-almost every \(v\in T^1M\), we have \[ \lim\sup_{t\to+\infty}\frac{\mathfrak{p}(v,t)}{\log t}=\frac{1}{P-P_0}, \] where \(\mathfrak{p}:T^1M\times\mathbb{R} \to[0,+\infty]\) is the penetration map.