Lie Group Variational Integrators for the Full Body Problem

Taeyoung Lee, Melvin Leok, N. Harris McClamroch

Abstract: We develop the equations of motion for full body models that describe the dynamics of rigid bodies, acting under their mutual gravity. The equations are derived using a variational approach where variations are defined on the Lie group of rigid body configurations. Both continuous equations of motion and variational integrators are developed in Lagrangian and Hamiltonian forms, and the reduction from the inertial frame to a relative coordinate system is also carried out. The Lie group variational integrators are shown to be symplectic, to preserve conserved quantities, and to guarantee exact evolution on the configuration space. One of these variational integrators is used to simulate the dynamics of two rigid dumbbell bodies.

1. Introduction

In classical celestial mechanics, planets and asteroids are frequently treated as point masses. However, for systems like binary asteroids or spacecraft orbiting non-spherical bodies, the orientation (attitude) of the body significantly affects its trajectory, and vice versa. This coupled translational and rotational dynamics is known as the full body problem.

Simulating these systems accurately over long periods is notoriously difficult. Standard numerical methods, such as Runge-Kutta, often suffer from two critical failures:

  1. Geometric Drift: Rotation matrices must remain orthogonal ($R^T R = I$). Numerical errors cause these matrices to "drift" over time, leading to configurations that no longer represent valid rigid bodies.
  2. Energy and Momentum Loss: General-purpose integrators do not inherently respect the conservation laws of physics, leading to non-physical energy dissipation or momentum gain in long-term simulations.

2. Methodology

The authors propose a robust framework based on Lie Group Variational Integrators. The core philosophy is to discretize the physical principle itself—Hamilton’s Principle of least action—rather than discretizing the resulting continuous differential equations.

2.1 Systematic Framework

The paper systematically derives eight distinct types of equations of motion, categorized by time (continuous vs. discrete), coordinates (inertial vs. relative), and formulation (Lagrangian vs. Hamiltonian). This comprehensive approach ensures that the dynamics can be analyzed in the most appropriate frame for a given problem.

Organizational framework of the derived equations of motion, showing the categorization by time, coordinates, and formulation.

2.2 Variational Derivation on Lie Groups

The configuration space is defined as $SE(3)$, the semi-direct product of translations ($\mathbb{R}^3$) and rotations ($SO(3)$). By defining variations using the Lie algebra ($\mathfrak{so}(3)$), the authors ensure the system never leaves the rotation manifold. As illustrated in the comparison below, the discrete equations inherit the symmetries of the original physical system, making them automatically symplectic and structure-preserving.

Comparison of the derivation process for continuous and discrete variational mechanics.

3. Experimental Evaluation

The proposed methods were evaluated using a Rigid Dumbbell Model. This model consists of two equal rigid spheres connected by a massless rod, creating a non-spherical mass distribution that highlights the complex coupling between rotation and translation under mutual gravity.

4. Key Results

The Lie group variational integrator was compared against a standard 4th-order Runge-Kutta (RK4) method, yielding significant improvements in physical fidelity:

  • Energy Conservation: In simulations of two dumbbells, the variational integrator maintained a maximum energy deviation of only $2.69 \times 10^{-7}$, while the RK4 method exhibited significant drift ($1.12 \times 10^{-2}$).
  • Orthogonality: The error in the rotation matrix ($||I - R^T R||$) for the variational integrator remained near machine precision ($2.86 \times 10^{-13}$), whereas the RK4 method drifted to $2.24 \times 10^{-2}$, eventually rendering the simulation physically invalid.
  • Robustness: The integrator successfully captured rapid energy interchanges during close encounters between bodies, a scenario where standard explicit methods typically fail.

5. Conclusion and Future Work

By discretizing the Action Integral rather than the equations of motion, the authors have developed a framework that maintains the geometric and physical integrity of the full body problem. While the current integrators are second-order accurate and require implicit updates for rotation, they provide a foundation for long-term, high-fidelity simulations in celestial mechanics. Future work may involve scaling these methods to N-body systems and implementing higher-order symmetric composition methods to improve computational efficiency.

Summary updated

Abstract

We develop the equations of motion for full body models that describe the dynamics of rigid bodies, acting under their mutual gravity. The equations are derived using a variational approach where variations are defined on the Lie group of rigid body configurations. Both continuous equations of motion and variational integrators are developed in Lagrangian and Hamiltonian forms, and the reduction from the inertial frame to a relative coordinate system is also carried out. The Lie group variational integrators are shown to be symplectic, to preserve conserved quantities, and to guarantee exact evolution on the configuration space. One of these variational integrators is used to simulate the dynamics of two rigid dumbbell bodies.
BibTeX
@inproceedings{lee2007lie,
  author = {Taeyoung Lee and Melvin Leok and N. Harris McClamroch},
  title = {{Lie Group Variational Integrators for the Full Body Problem}},
  booktitle = {Computer},
  year = {2007},
  doi = {10.1016/j.cma.2007.01.017},
}
Figure