Stochastic Lie group integrators
Simon J.A. Malham and Anke Wiese
Simon Malham update to 2007-10-16
https://arxiv.org/abs/0704.0022
We present Lie group integrators for nonlinear stochastic differentialequations with non-commutative vector fields whose solution evolves on a smoothfinite dimensional manifold. Given a Lie group action that generates transportalong the manifold, we pull back the stochastic flow on the manifold to the Liegroup via the action, and subsequently pull back the flow to the correspondingLie algebra via the exponential map. We construct an approximation to thestochastic flow in the Lie algebra via closed operations and then push back tothe Lie group and then to the manifold, thus ensuring our approximation lies inthe manifold. We call such schemes stochastic Munthe-Kaas methods after theirdeterministic counterparts. We also present stochastic Lie group integrationschemes based on Castell--Gaines methods. These involve using an underlyingordinary differential integrator to approximate the flow generated by atruncated stochastic exponential Lie series. They become stochastic Lie groupintegrator schemes if we use Munthe-Kaas methods as the underlying ordinarydifferential integrator. Further, we show that some Castell--Gaines methods areuniformly more accurate than the corresponding stochastic Taylor schemes.Lastly we demonstrate our methods by simulating the dynamics of a free rigidbody such as a satellite and an autonomous underwater vehicle both perturbed bytwo independent multiplicative stochastic noise processes.