WorksMathematics
High-Order Area-Preserving Runge–Kutta Methods for Surface Diffusion Flow of Curves
Abstract
We propose a class of high-order area-preserving Runge–Kutta (RK) methods for the surface diffusion flow of curves, a curvature-driven geometric flow that decreases the perimeter while conserving the enclosed area. By coupling the parametric finite element approximation of the classical BGN formulation with symplectic RK time discretisation, we rigorously prove that the fully discrete scheme unconditionally conserves the enclosed area of curves. This conservation arises solely from the intrinsic symplectic structure of the RK methods, without introducing any additional constraints, auxiliary variables, or Lagrange multipliers to enforce structure preservation. To attain high temporal accuracy, we adopt -stage Gauss RK methods, which are symplectic and yield order (e.g., ). The resulting fully implicit, strongly nonlinear system is solved efficiently via an equivalent discrete formulation combined with a monolithic damped Newton iteration. Extensive numerical tests on various evolving curves validate the theoretical findings, confirming exact area preservation, high-order temporal convergence, and favorable mesh quality, while the Newton iteration count rapidly decreases to fewer than 8 per step, demonstrating the robustness of the tailored solver.