A Novel Formulation of Point Vortex Dynamics on the Sphere: Geometrical and Numerical Aspects

Joris Vankerschaver, Melvin Leok

Introduction

Simulating the motion of point vortices—singularities in a fluid's vorticity—is a foundational problem in hydrodynamics. These models are critical for understanding complex phenomena ranging from atmospheric patterns on Earth to the behavior of quantum fluids. However, when these vortices are constrained to a spherical geometry, numerical simulation faces significant hurdles. Standard numerical methods, such as the fourth-order Runge-Kutta (RK4) scheme, often suffer from "numerical drift," where conserved physical quantities like energy and angular momentum gradually erode over long integration periods.

To address this, researchers typically turn to variational integrators. These methods discretize the system's Lagrangian rather than the equations of motion directly, naturally preserving the symplectic structure of the physics and ensuring superior long-term stability. However, a fundamental topological obstruction exists: a linear Lagrangian cannot be defined directly on a compact manifold like the 2-sphere ($S^2$). This paper presents a novel solution by "lifting" the dynamics into a higher-dimensional space where these obstructions vanish.

The Hopf Lift: Bypassing Topological Obstructions

The core innovation of this research is the use of the Hopf fibration to map the physics of fluid vortices from the 2D sphere ($S^2$) to the 3D sphere ($S^3$).

1. Lifting to $S^3$

While the symplectic form on $S^2$ is not "exact" (meaning it cannot be written as the derivative of a global potential), the 3-sphere $S^3$ does not share this limitation. By mapping the positions of $N$ vortices on the unit 2-sphere to the 3-sphere, the authors bypass the topological barrier that prevents the construction of a standard Lagrangian.

2. Lie Group Identification

The authors identify $S^3$ with the Lie group $SU(2)$ (special unitary matrices). This identification allows the application of Lie group dynamics, providing a robust algebraic framework for the system. They construct a "Hopf-lifted" linear Lagrangian on $(S^3)^N$. Although this introduces "gauge freedom"—extra degrees of freedom that do not change the physical position on the original 2-sphere—the resulting Euler-Lagrange equations project exactly back onto the known physical equations for point vortices on $S^2$.

Numerical Methodology

By applying the midpoint rule to this new lifted Lagrangian, the authors derive a discrete-time variational integrator. For $S^1$-invariant Hamiltonians, which describe point vortex systems, this simplifies into an implicit midpoint method on $S^3$. This approach is:

  • Symplectic: It preserves the underlying geometric structure of the phase space.
  • Second-Order Accurate: It maintains a consistent level of local error.
  • Constraint-Preserving: It maintains the unit-length constraint and the vortex moment to machine precision without requiring expensive re-projection steps.

Benchmarking and Results

The algorithm was evaluated using four classic numerical benchmarks in fluid dynamics: stable vortex rings, the spherical von Kármán vortex street, self-similar 3-vortex collapse, and large ensembles simulating Kelvin-Helmholtz instabilities.

Long-Term Stability and Energy Conservation

In simulations of the von Kármán vortex street, the Hopf integrator demonstrated bounded energy oscillations. In contrast, non-symplectic methods like RK4 exhibited a clear linear drift in energy over time, eventually leading to unphysical results.

Moment Preservation

The integrator preserved the "vortex moment" (the spherical equivalent of angular momentum) to machine precision ($10^{-15}$). While the standard midpoint method on $S^2$ also performs well here, RK4 showed significant drift, highlighting the advantage of the variational approach for maintaining physical symmetries (Noether’s Theorem).

Handling Chaos and Singularities

In the Kelvin-Helmholtz instability simulation, the Hopf integrator maintained the stability of the vortex ring for twice as long ($t=30$) as non-symplectic integrators ($t=15$). Furthermore, during the near-collapse of three vortices—a highly sensitive numerical event—the geometric integrator recovered the original energy and moment values after the event, whereas RK4 suffered permanent, irreversible "jumps" in these invariants.

Discussion and Future Outlook

The research demonstrates that adding redundancy—the extra dimension of the "lift"—actually simplifies the mathematical treatment of spherical dynamics. The "gauge freedom" inherent in the $S^3$ formulation allows the integrator to remain on the manifold naturally.

While the current formulation is second-order accurate and requires solving nonlinear equations at each step (due to its implicit nature), it offers a significant leap in reliability for long-term simulations. The authors suggest that this "lifting" technique could be generalized to other surfaces using quantum line bundles or applied to other partial differential equations, such as the Korteweg-de Vries (KdV) or nonlinear Schrödinger equations, opening new avenues for geometric numerical integration in complex physical systems.

Summary updated

Abstract

In this paper, we present a novel Lagrangian formulation of the equations of motion for point vortices on the unit 2-sphere. We show first that no linear Lagrangian formulation exists directly on the 2-sphere but that a Lagrangian may be constructed by pulling back the dynamics to the 3-sphere by means of the Hopf fibration. We then use the isomorphism of the 3-sphere with the Lie group SU(2) to derive a variational Lie group integrator for point vortices which is symplectic, second-order, and preserves the unit-length constraint. At the end of the paper, we compare our integrator with classical fourth-order Runge--Kutta, the second-order midpoint method, and a standard Lie group Munthe-Kaas method.
BibTeX
@article{vankerschaver2014novel,
  author = {Joris Vankerschaver and Melvin Leok},
  title = {{A Novel Formulation of Point Vortex Dynamics on the Sphere: Geometrical and Numerical Aspects}},
  journal = {J. Nonlinear sci},
  year = {2014},
  doi = {10.1007/s00332-013-9182-5},
}
Figure