Fast Construction of Forward Flow Maps using Eulerian Based Interpolation Schemes

We propose a modification to a recently developed Eulerian interpolation scheme for constructing the flow map for autonomous, periodic and aperiodic dynamical systems. We show that the proposed methods significantly improve the computational efficiency when a large number of flow maps are needed, yet retain second-order accuracy as in the original approach. The idea is to pre-compute and to store a carefully selected set of intermediate flow maps. When the initial and the final time of a required flow map are known, our proposed methods can simply load these pre-processed short-time flow maps for flow map construction. Numerical examples are included to validate our theoretical prediction, and demonstrate the effectiveness of these proposed Eulerian interpolation schemes as a simple numerical tool for other applications such as the finite-time Lyapunov exponent in determining the Lagrangian coherent structure of dynamical systems and the geometrical optics problems.