A divergence-conforming finite element method for the surface Stokes equation

TL;DR

This paper introduces a divergence-conforming finite element method for solving the surface Stokes equation, addressing challenges like tangential velocity constraints and Killing fields, with proven error bounds and numerical validation.

Contribution

The paper develops a novel surface finite element approach that exactly enforces incompressibility and robustly filters Killing fields, improving numerical solutions for surface Stokes problems.

Findings

Method exactly satisfies incompressibility.

Robustly filters out Killing fields.

Numerical experiments confirm theoretical error estimates.

Abstract

The Stokes equation posed on surfaces is important in some physical models, but its numerical solution poses several challenges not encountered in the corresponding Euclidean setting. These include the fact that the velocity vector should be tangent to the given surface and the possible presence of degenerate modes (Killing fields) in the solution. We analyze a surface finite element method which provides solutions to these challenges. We consider an interior penalty method based on the well-known Brezzi-Douglas-Marini $H({\rm div})$-conforming finite element space. The resulting spaces are tangential to the surface, but require penalization of jumps across element interfaces in order to weakly maintain $H^1$ conformity of the velocity field. In addition our method exactly satisfies the incompressibility constraint in the surface Stokes problem. Secondly, we give a method which robustly filters Killing fields out of the solution. This problem is complicated by the fact that the dimension of the space of Killing fields may change with small perturbations of the surface. We first approximate the Killing fields via a Stokes eigenvalue problem and then give a method which is asymptotically guaranteed to correctly exclude them from the solution. The properties of our method are rigorously established via an error analysis and illustrated via numerical experiments.