Computation of Stable and Unstable Manifolds of hyperbolic Trajectories in Two-dimensional, Aperiodically Time-Dependent Vector Fields

In this paper, we develop two accurate and fast algorithms for the computation of the stable and unstable manifolds of hyperbolic trajectories of two-dimensional, aperiodically time-dependent vector fields. First we develop a benchmark method in which all the trajectories composing the manifold are integrated from the neighborhood of the hyperbolic trajectory. This choice, although very accurate, is not fast and has limited usage. A faster and more powerful algorithm requires the insertion of new points in the manifold as it evolves in time. Its numerical implementation requires a criterion for determining when to insert those points in the manifold, and an interpolation method for determining where to insert them. We compare four different point insertion criteria and four different interpolation methods. We discuss the computational requirements of all of these methods. We find two of the four point insertion criteria to be accurate and robust. One is a variant of a criterion originally proposed by Hobson. The other is a slight variant of a method due to Dritschel and Ambaum arising from their studies of contour dynamics. The preferred interpolation method is also due to Dritschel. These methods are then applied to the computation of the stable and unstable manifolds of the hyperbolic trajectories of several aperiodically time-dependent variants of the Duffing equation. (C) 2003 Elsevier B.V. All rights reserved.