Closed kylekrol closed 3 years ago
This actually isn't strictly required as they were implemented inline in the orb::Orbit
class and decoupling the two given the nature of the propagator would actually being very difficult. As such, I'm going to go ahead and close this issue.
Runge-Kutta methods aren't stable for oscillatory system over long periods of time. In the case of orbital dynamics specifically, this may be seen in our orbital radius growing over the course of our simulation.
Moving over the a leapfrog integrator would remedy this issue. See the link below for more details:
@nhz2 had started using these schemes in his orbit propegator implementations and they will be used going forward with the orbit estimator implementation.