Astrodynamics and the Julia language

My research is almost entirely based on writing code, so I end up spending quite a bit of time working with different programming languages, developing code for all sorts of simulations, piping different tools, etc. I don’t necessarily believe that Julia is the best language for all purposes - but I do believe that it is a great language, particularly for research in optimization theory/astrodynamics (comme par hasard!).

Research coding requires a lot of trial and error, especially since I tend to get new ideas or discover pitfalls while I implement things. Occasionally, I want to be able to have lazy syntax (e.g. a cheeky double for-loop) without having to pay (too much) with performance. And for these purposes, Julia is great. Of course, there are also the well-discussed Julia perks - its speed (thank you JIT), ease of parallelization (<3), and capability to write mathematical notation-like code, to name a few.

That is not to say Julia is good for everything; there are situations where things just have to be implemented in C/C++; if I need to pipe multiple software, I think python is still easier to work with; finally, there are some (I am sure many more, but at least some that I occasionally work with) that are best explored with other languages, namely: machine learning (beyond some simple regression-style NN tasks), global optimization (the pagmo/pygmo libraries are just too good, and I personally feel that the Julia metaheuristics options are not as advanced).

Anyway, for (research in) Astrodynamics, I do think Julia can be an attractive option!

• dynamic language that is fast! (numba/cython is nice, but can be limiting)
• DifferentialEquations.jl is rich in capabilities (type of algorithms, customizability, parallelism paradigms…)
• there is a SPICE wrapper - SPICE.jl
• JuMP.jl provides good support for solving LP/MILP/QP/CP etc. within a modeling language setting

Meanwhile, drawbacks that I personally felt using Julia is:

• Lack of widely-used black-box NLP solver (or wrapper to solver like SNOPT/IPOPT) where variables can be passed as a vector of floats rather than defining as a “variable” type (i.e. not a modeling language)

Below is a few packages for astrodynamics routines that I have used as building blocks to some of my research projects.

Packages

Available:

• AstrodynamicsBase.jl - basic astrodynamics routines for coordinate transformations, handling two-body orbital elements, defining astrodynamics constants, solving Kepler’s problem, etc.
• Lambert.jl - native julia Lambert solver, implementing Dario Izzo’s revised algorithm. Also implements the MGA-1DSM model for high-thrust, gravity-assist interplanetary trajectory optimization.
• R3BP.jl - routines for restricted three-body problems (CR3BP, BCR4BP)
• FullEphemerisPropagator.jl - full-ephemeris spacecraft state propagation library, building on top of DifferentialEquations.jl.

Coming soon!

• [Navigation.jl]
• [QLaw.jl]

Posts on Julia x Astrodynamics

• https://discourse.julialang.org/t/about-the-astro-space-category/5036