General Question Extra tab propagators

[Zatnikitelman]

This is something I've been meaning to ask for a while now, but finally thought of it when I actually had O-F up!
I know that the linear and angular state propagators under Orbiter's Extra tab affect how stable and accurate the simulation is (it says so), but I was wondering if anyone had any tips on making the simulation more accurate by varying the parameters here.
Right now under Linear, I have the Runge-Kutta 2,4,6,8,7th order propagators (in that order descending) and step limits of 0.5, 20, 100, 200s (in that order descending)
I have basically the same under Angular.
Any thoughts on this?

 

[martins]

The Doc/Technotes/dynamics.pdf document has a few graphs displaying some aspects of numerical stability for the different propagators (linear only, not angular). You could use those as an initial guide to determine what order of propagator to use for what time step intervals.

Other than that, experimenting with different settings is a good approach. I would love if somebody did a detailed stability and accuracy analysis and posted the results.

 

[Zatnikitelman]

I'd love to find this out myself as I'd love to be able to have a space station properly propagate its position over time even with using time-warp.
Unfortunately, I don't know C++ yet so I can't write a program that can simply export values and have them compared possibly in a spreadsheet with the standard deviation calculated.
The Dynamics.pdf was an interesting read even if the math was a bit over my head for now. From that document, it looks like instead of simply shooting for accuracy, we'd need to try to find a balance between performance and accuracy thanks due to the complexity of the equations.