Dynamics of the N-fold Pendulum in the framework of Lie Group Integrators

Elena Celledoni, Ergys Çokaj, Andrea Leone, Davide Murari, Brynjulf OwrenarXiv 2021 · arXiv

Abstract

This paper explores the simulation of complex dynamics in an $N$-fold 3D pendulum—a chain of $N$ connected spherical pendulums—using Lie group integrators. By employing Runge–Kutta–Munthe–Kaas (RKMK) methods with adaptive time-stepping, the authors demonstrate an efficient approach to maintaining geometric constraints, such as keeping pendulums on a spherical manifold, as system complexity increases.

Introduction

Simulating multi-body mechanical systems, such as a chain of pendulums, presents significant challenges because the system's state space is non-Euclidean. Each link in an $N$-fold pendulum is constrained to move on the surface of a sphere ($S^2$). Standard numerical solvers often suffer from "drift," where numerical errors lead to unphysical results, such as changes in the lengths of the pendulum rods.

Schematic of an N-fold 3D pendulum system.

To address this, Lie group integrators are used to evolve solutions directly on the manifold. However, implementing these efficiently for complex multi-body systems requires careful handling of the tangent bundle (positions and velocities) and the development of robust variable step-size algorithms to manage increasingly "stiff" dynamics as more links are added.

Methodology

The core of the proposed approach is the Runge–Kutta–Munthe–Kaas (RKMK) method. This technique transforms differential equations on a manifold $\mathcal{M}$ into equivalent equations on a Lie algebra $\mathfrak{g}$, which is a flat vector space. Integration is performed in this flat space, and the result is projected back onto the manifold using an "exponential map," ensuring that geometric constraints are satisfied by construction.

Key technical contributions include:

  • Geometric Framework: Proving that the tangent bundle $TQ$ of a homogeneous manifold $Q$ is also a homogeneous manifold, allowing the entire phase space of a Lagrangian system to be treated within the Lie group framework.
  • Variable Step Size: Integrating an embedded Runge-Kutta pair—specifically the Dormand–Prince 5(4) method—within the RKMK framework. This allows the solver to automatically adjust the time step based on the complexity of the motion.
  • N-fold Modeling: Formulating Euler-Lagrange equations for $N$ connected 3D pendulums compatible with infinitesimal generators of Lie group actions.

Key Results and Discussion

The performance of the RKMK(5,4) method was evaluated against constant step size versions and standard MATLAB solvers like ode45.

Performance metrics of the RKMK(5,4) solver: (Left) Accuracy vs. number of pendulums; (Right) Adaptive step size behavior over time.

As shown in Figure 4, the variable step size RKMK(5,4) method significantly outperformed fixed-step alternatives. The adaptive solver successfully shrinks the time step during periods of violent motion and expands it when the dynamics are smoother. This "wise" allocation of computational effort leads to higher accuracy for a given number of total steps compared to fixed-step methods.

Furthermore, the RKMK methods preserve the unit length of each link to machine precision without the need for external projection steps. The authors also noted that as the number of pendulums ($N$) increases, the dynamics become more intricate, making adaptive stepping essential for efficiency. This structural modeling approach suggests that a chain of many pendulums is a viable way to approximate continuous flexible structures like beams.

Conclusion

This work demonstrates that Lie group integrators, particularly adaptive RKMK methods, provide a robust framework for simulating $N$-fold pendulums while strictly adhering to geometric constraints. Future work will focus on incorporating friction and complex link interactions, as well as exploring the continuum limit where $N$ approaches infinity to model flexible structures like rods or strings.

Summary updated

Abstract

Since their introduction, Lie group integrators have become a method of choice in many application areas. Various formulations of these integrators exist, and in this work we focus on Runge--Kutta--Munthe--Kaas methods. First, we briefly introduce this class of integrators, considering some of the practical aspects of their implementation, such as adaptive time stepping. We then present some mathematical background that allows us to apply them to some families of Lagrangian mechanical systems. We conclude with an application to a nontrivial mechanical system: the N-fold 3D pendulum.
BibTeX
@misc{celledoni2021dynamics,
  author = {Elena Celledoni and Ergys Çokaj and Andrea Leone and Davide Murari and Brynjulf Owren},
  title = {{Dynamics of the N-fold Pendulum in the framework of Lie Group Integrators}},
  year = {2021},
  eprint = {2109.12325},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2109.12325},
}
Figure