Computing Stokes flows in periodic channels via rational approximation
Yidan Xue
TL;DR
The paper develops a fast, accurate framework for 2D Stokes flows in periodic channels by representing Goursat functions with trigonometric rational bases and placing poles via AAA-LS in a conformal boundary map. The method enforces 2π-periodicity in the basis, solves a well-conditioned least-squares system to obtain the Goursat coefficients, and uses Vandermonde–Arnoldi orthogonalization to stabilize the basis. It demonstrates Poiseuille and Couette problems with analytic and polygonal boundaries, achieving 6-digit accuracy in under a second and enabling efficient computation of unsteady particle trajectories and mixing via Poincaré maps. The work highlights the practical impact for rapid, high-fidelity simulations of periodic Stokes flows and discusses extensions to related periodic Laplace/Stokes problems and multiply connected domains.
Abstract
Rational approximation has proven to be a powerful method for solving two-dimensional (2D) fluid problems. At small Reynolds numbers, 2D Stokes flows can be represented by two analytic functions, known as Goursat functions. Xue, Waters and Trefethen [SIAM J. Sci. Comput., 46 (2024), pp. A1214-A1234] recently introduced the LARS algorithm (Lightning-AAA Rational Stokes) for computing 2D Stokes flows in general domains by approximating the Goursat functions using rational functions. In this paper, we introduce a new algorithm for computing 2D Stokes flows in periodic channels using trigonometric rational functions, with poles placed via the AAA-LS algorithm [Costa and Trefethen, European Congr. Math., 2023] in a conformal map of the domain boundary. We apply the algorithm to Poiseuille and Couette problems between various periodic channel geometries, where solutions are computed to at least 6-digit accuracy in less than 1 second. The applicability of the algorithm is highlighted in the computation of the dynamics of fluid particles in unsteady Couette flows.
