A Note on the Construction of Explicit Symplectic Integrators for Schwarzschild Spacetimes

TL;DR

This paper develops and compares explicit symplectic integrators for simulating particle motion in Schwarzschild spacetimes, finding that three-part splitting with optimized fourth-order methods offers superior accuracy and efficiency.

Contribution

It introduces a three-part Hamiltonian splitting method and demonstrates that optimized fourth-order symplectic integrators outperform existing algorithms in accuracy for Schwarzschild spacetime simulations.

Findings

Three-part splitting yields the best accuracy among tested methods.

Optimized fourth-order Runge-Kutta-Nyström integrators outperform Yoshida algorithms.

Sixth-order integrators do not significantly improve accuracy over fourth-order ones.

Abstract

In recent publications, the construction of explicit symplectic integrators for Schwarzschild and Kerr type spacetimes is based on splitting and composition methods for numerical integrations of Hamiltonians or time-transformed Hamiltonians associated with these spacetimes. Such splittings are not unique but have various choices. A Hamiltonian describing the motion of charged particles around the Schwarzschild black hole with an external magnetic field can be separated into three, four and five explicitly integrable parts. It is shown through numerical tests of regular and chaotic orbits that the three-part splitting method is the best one of the three Hamiltonian splitting methods in accuracy. In the three-part splitting, optimized fourth-order partitioned Runge-Kutta and Runge-Kutta-Nystr\"{o}m explicit symplectic integrators exhibit the best accuracies. In fact, they are several orders of magnitude better than the fourth-order Yoshida algorithms for appropriate time steps. The former algorithms need small additional computational cost compared with the latter ones. Optimized sixth-order partitioned Runge-Kutta and Runge-Kutta-Nystr\"{o}m explicit symplectic integrators have no dramatic advantages over the optimized fourth-order ones in accuracies during long-term integrations due to roundoff errors. The idea finding the integrators with the best performance is also suitable for Hamiltonians or time-transformed Hamiltonians of other curved spacetimes including the Kerr type spacetimes. When the numbers of explicitly integrable splitting sub-Hamiltonians are as small as possible, such splitting Hamiltonian methods would bring better accuracies. In this case, the optimized fourth-order partitioned Runge-Kutta and Runge-Kutta-Nystr\"{o}m methods are worth recommending.