Questions about Nyx future plans and needs

[Ryan-D-Gast]

Hello, I have two questions.

1. After looking at the source code it looks like for the propagator IVP ODE methods of the Runge Kutta variety. It looks like the solvers are integrated directly and not standalone ODE solvers, i.e. not agnostic to the problem. I could be wrong I found it difficult to discern reading the source code. Would a Rust implementation of scipy's solve_ivp be useful for future developments in the library? For my own personal projects I've translated DOP853 and other simpler solvers like Adams and BDF into rust and built a private package and testing framework around it for my own projects. It looks like for differential equation solvers you guys are all set but I just wanted to check and see if there is a need. Also is LSODE a commonly used or desired solver in the Astrodynamics world? I plan on working on it next.

2. Do you guys have plans for a Desktop application? I've been messing around with frameworks liked Tauri and trying out more web focused frameworks all seem like they could be promising for this kind of application. If so would nyx capabilities, if a sufficient desktop frontend was done, be able to compete with applications like Free-Flyer?

For both of these questions I am unsure how I would get involved, though I would like to, so any recommendations on that front would be appreciated. Thank you.

[ChristopherRabotin]

Hi Ryan,

Thanks for your questions.

Yes, the ODE methods are decently tightly integrate into the code base. It's possible to provide other dynamics by implementing the Dynamics trait. Before Nyx, I developed another astrodynamics framework in Go where the ODE solver was generic and its separate library: that turned out to be a bad design because of the difficulty to pass derivative relevant data that isn't part of the state (e.g. planetary positions).

I'm not too familiar with other solvers apart from the Runge Kutta methods. If these lead to very similar results but provide some other advantage (e.g. less memory usage, or less steps), I would be happy to review such an implementation. Nyx is used in-flight so a lot of validation work was needed, notably comparing the results of GMAT to those of Nyx. That's the main reason why Nyx has the same solvers as GMAT (plus a few more).

Yes, Nyx definitely has the capability to compete with Free Flyer and STK/ODTK. Developing a desktop application with Nyx is definitely on my mind and that of other companies. The AGPL license however sets strict open-source rules on how one can use Nyx in so-called derivative works, notably requiring that any application that interacts with Nyx also be made open-source when such an application is distributed to third-parties (i.e. there is no limitation on in-house use of software built on top of Nyx, only once that software is distributed to other people or companies, regardless of the price). Other AGPL software typically provides a commercial license for commercial users.