Exponentially small splitting of separatrices associated to 3D whiskered tori with cubic frequencies (Q2200489)

The authors study the exponentially small splitting of separatrices in a perturbed Hamiltonian system with four degrees of freedom, associated with a three-dimensional whiskered torus having a cubic frequency vector. They start with an integrable Hamiltonian having whiskered tori with a homoclinic whisker or separatrix that consists by coincident stable and unstable whiskers. The measure used for the splitting distance is a splitting function \(\mathcal {M} : \mathbb{T}^3 \rightarrow \mathbb{R}^3\) that is restricted to a section transverse to the unperturbed separatrix. This function provides a vector distance between the whiskers on this section along complementary directions. The maximal splitting distance is then the maximum of \(|\mathcal{M}(\theta)|\) for \(\theta \in \mathbb{T}^3\). The attention is focused on a specific torus with a frequency vector of fast frequencies \(\omega_\epsilon = \frac{\omega}{\sqrt{\epsilon} } \) and \(\omega = (1, \Omega, \tilde{\Omega})\) with a small positive parameter \(\epsilon\). It is assumed that the frequency ratios \(\Omega = \frac{\omega_2}{\omega_1}\) and \(\tilde{\Omega} = \frac{\omega_3}{\omega_1}\) with \(\omega_1 = 1\) generate a complex cubic field. This means that \(\Omega\) is a cubic irrational number whose two conjugates are not real, and \(\tilde{\Omega} = a_0 + a_1 \Omega + a_2 \Omega^2\) with rational \(a_0, a_1\) and \(a_2\) and \(a_2 \neq 0\). An example is the the vector \(\omega = (1, \Omega, \Omega^2)\) where \(\Omega\) is the well-known cubic golden number satisfying \(\Omega^3 = 1 - \Omega\). The authors look at a perturbed Hamiltonian of the form \(H = H_0 + \mu H_1\) where \(\mu\) is small. In general, after perturbation of the Hamiltonian the whiskers no longer coincide. Assuming \(\mu = \epsilon^r\) for some \(r > 0\), this becomes a singular perturbation question and here the splitting is exponentially small with respect to \(\epsilon\). The authors then provide an asymptotic estimate for the maximal splitting distance, and show that this estimate depends strongly on the arithmetic properties of the cubic number \(\Omega\).