Adding Higher Fidelity Propagators

[hunterqueb]

What are the current plans for adding higher fidelity propagators?
Python is in desperate need for a high quality astrodynamics library, and the only thing stopping me from adopting AstroForge into my workflow is support for additional propagators. I think the main issues with the current propagator implementation needs to be addressed:

1. Moving away from python implementations of ode solvers
2. Implementation of different integrators

In regards to (1), scipy's solve_ivp is slow, and moving to a pipeline that includes calling C implementations of RK45 would greatly increase performance in applications where the propagator can be called millions of times. For (2), higher order methods like RK89 would be a nice feature when propagating objects with more complex force models.

[jvarey]

I whole heartedly agree with you on all fronts.

Python is in desperate need for a high quality astrodynamics library, and the only thing stopping me from adopting AstroForge into my workflow is support for additional propagators.

I would love for AstroForge to be a great solution for high-quality astrodynamics in python, and it would be great to extend what we have re: propagators.

Moving away from python implementations of ode solvers

Most definitely. I have some hope of implementing the low-level solver in Rust, but that is a ways off from being ready.

Implementation of different integrators

This would also be great. Although before doing a bunch of work, I would want to see what the relative benefit is from using a higher order numerical solver vs using more accurate force models. An extreme example- you wouldn't want to throw the two-body equations of motion into an RK89 solver with 10^-12 km absolute and relative tolerance.

This is going to come down to resources, but I view the propagators as a pretty valuable part of the package, so I'm in favor.