Skip to content

Latest commit

 

History

History
171 lines (117 loc) · 7.05 KB

numfocus-sdg-2020-r3_report.md

File metadata and controls

171 lines (117 loc) · 7.05 KB
geometry output colorlinks
margin=2.5cm
pdf_document
true

Report for the NumFOCUS small development grants program

Abstract

All the work performed during the lasts months regarding the NumFOCUS Small Development Grants Program 2020 3rd round for poliastro is collected and presented within this document. A quick review on the original proposed objectives followed by the problems encountered during their implementation together with the solutions devised are explained. Links to all contributions are provided at the end of the paper.

Review of the original proposal

The main goal of the original proposal was to implement a validation framework for testing all poliastro's capabilities for which no literature examples were available. These capabilities included planetary reference frames transformations and three-dimensional impulsive maneuvers, although orbital elements conversion and Keplerian propagation were also considered.

After a quick study on the possible solutions, validation against similar software was selected as the best approach to solve for the problem. These software were Orekit, GMAT and STK.

The following timeline of deliverables was devised for the whole program:

  • The first two weeks would be devoted to get familiar with previous software and implement simple two-body problem validation cases (orbit element conversion and Keplerian propagation).

  • First month fully devoted to the 3D impulsive maneuvers validation.

  • The goal for the second was to simplify and reformat previous implementation (if required) in order to emulate poliastro's API.

  • During the third month, all the effort would be directed towards the validation of planetary frames transformations.

  • Last month was set with the idea of being used as a time-margin resource in case some of the previous points would take more than expected.

Issues encountered and alternative approaches

During the first weeks of the program, it became evident that some of the original proposed goals would required a different approach to be accomplished:

  • STK implementation. We decided not to implement the validation against this software due to its lack of Linux compatibility. The GMAT and Orekit software were considered enough for the purposes of the validation work.

  • Suppression of poliastro's API emulation. Trying to replicate the logic used by poliastro using the Orekit wrapper introduced an extra layer of complexity within the validation implementation. Therefore, this approach was deprecated in favor of direct comparison between pure Orekit API based tests and poliastro's ones.

  • GMAT not included in CI. Although a solution for including this software within the continuous integration workflow was devised, it was rejected because its excessive complexity. Including GMAT meant downloading more than 300MB every time the CI was triggered and building a Python wrapper to manage GMAT outputs. Therefore, only validation scripts were included as part of the testing process.

Enhancements to the original proposal

In addition to the initial objectives, some others were also implemented:

  • Continuous integration setup. By making use of GitHub's actions, it was possible to introduced a CI workflow within the main poliastro's repository which triggers the whole validation cases every time a new piece of code is merged into the main branch.

Obtained results and benefits

Once all required poliastro's features were validated, the following results arose:

  • An error was detected in the Moon's rotational elements, which which introduced numerical differences of the order of 1E-5. These proved the frame validation logic to be successful.

  • The frame logic inside poliastro's API was proof to be right. This logic could be expanded to the Astropy project, so other projects might benefit from this.

  • Auxiliary frames, such us planetary-fixed non-parallel to ICRF, can be implemented.

  • Groundtrack plots can be now extended to the rest of planets of the Solar System.

Associated pull requests

All the pull requests generated as a result of this work are listed down below, being classified according to the repository in which they originated.

validation repository: https://github.com/poliastro/validation

poliastro repository: https://github.com/poliastro/poliastro

Future work

Regarding planetary frames validation, the rotational elements implemented in Orekit software were taken from the IAU WGCCRE 2009 report while the ones from poliastro do from IAU WGCCRE 2015 one. In issue to the official Orekit's repository was opened to suggest to update their rotational elements, see:

Finally, this validation approach might also be applied to test future poliastro's capabilities. For example, propagation in non-inertial frames can lead to halo orbit's implementation and three-body sub-package enhancements.