Adaptive finite difference methods for the Willmore flow: mesh redistribution algorithm and tangential velocity approach

TL;DR

This paper introduces two adaptive finite difference methods for simulating Willmore flow, utilizing monitor functions for dynamic mesh adaptation and incorporating energy stability, leading to accurate and efficient geometric evolution modeling.

Contribution

The paper presents novel adaptive finite difference schemes for Willmore flow, including a mesh redistribution strategy and a tangential velocity approach with energy stability guarantees.

Findings

Accurately captures complex interface geometries.

Demonstrates robustness and computational efficiency.

Provides energy-stable numerical algorithms.

Abstract

We develop two adaptive finite difference methods for the numerical simulation of the Willmore flow, employing the kth-order backward differentiation formula (BDFk) for time discretization, together with monitor functions for dynamic mesh adaptation along evolving interfaces. The first approach is based on a weighted arc-length equidistribution strategy driven by a monitor function to adaptively redistribute grid points. An adaptive monitor selection mechanism, constructed from the curvature and its variation, enhances spatial resolution in regions of strong geometric complexity while preserving mesh regularity. The second approach eliminates explicit reparameterization by incorporating a tangential velocity into the Willmore flow, with mesh redistribution inherently embedded in the geometric evolution. We further develop an energy-stable correction algorithm for the second method to guarantee discrete energy stability at the theoretical level. In both approaches, the monitor function serves as the core component of the adaptive framework, encoding essential geometric information -- such as curvature and curvature variation -- to guide mesh refinement and redistribution. Extensive numerical experiments demonstrate that the proposed BDFk-based adaptive schemes accurately capture the geometric evolution of the Willmore flow and exhibit excellent robustness and computational efficiency for problems involving complex interface geometries.