Symplectic Integrators
Symplectic integrators preserve the geometric phase-space structure of Hamiltonian systems, keeping energy error bounded over long integrations.
Preserving structure, not just accuracy
Hamiltonian systems, from planetary orbits to charged-particle motion, have a geometric structure: their flow preserves phase-space volume and a symplectic two-form. Ordinary integrators chosen only for accuracy ignore this structure and let energy drift systematically over long runs, so a planet spirals into or out of its orbit numerically. Symplectic integrators are designed to preserve the symplectic structure exactly, at the discrete level.
The payoff is qualitative fidelity over very long times. A symplectic method does not conserve energy exactly, but the energy error stays bounded and oscillatory rather than growing, because the method exactly conserves a modified (shadow) Hamiltonian that stays close to the true one.
Backward error analysis
The reason symplectic methods behave so well is explained by backward error analysis: the numerical trajectory is the near-exact trajectory of a slightly perturbed Hamiltonian system. Because that perturbed system is still Hamiltonian, it has its own conserved energy, which the numerics track faithfully, so no secular energy drift appears.
Examples and constraints
The leapfrog and velocity-Verlet schemes are the simplest symplectic integrators; higher-order versions come from symmetric compositions and splitting of the Hamiltonian into kinetic and potential parts. A key constraint is that symplecticity relies on a fixed time step; adaptive step-size control generally breaks it, so specialized techniques are needed to combine adaptivity with structure preservation.
- Preserve the symplectic phase-space structure exactly
- Bounded, non-drifting energy error over long times
- Conserve a modified Hamiltonian near the true one
- Fixed time step usually required to retain symplecticity
Symplectic integration is the standard for long-time orbit following of charged particles in magnetic fields, where accurate confinement statistics depend on avoiding artificial energy drift.