Forward AutodIff Compatibility

[jmurphy6895]

I changed the internal time to be handled as seconds past JD_J2000 instead of DateTime to allow for automatic differentiation compatibility and changed some of the return types to handle more general inputs, namely ForwardDiff's Duals. Reverse Autodiff is blocked by not knowing how to handle the DCM struct at the moment but will be handled in a future PR.

The previous DateTime interfaces are still present they just internally convert the DateTime to the seconds past JD_J2000 and call the newer function.

[ronisbr]

Hi @jmurphy6895 !

Aweome PR, thank you very much! I just added some general comments, check, please, if they apply.

[codecov]

Welcome to Codecov 🎉

Once you merge this PR into your default branch, you're all set! Codecov will compare coverage reports and display results in all future pull requests.

Thanks for integrating Codecov - We've got you covered ☂️

[ronisbr]

Hi @jmurphy6895,

Can you please check if the failures are only related to test packages that cannot be installed in Julia 1.6? If so, can you modify the test script to ignore those tests in Julia 1.6? I plan to drop support for Julia 1.6 but it will take sometime to update all packages in SatelliteToolbox.jl ecosystem.

[jmurphy6895]

Hey @ronisbr,

I don't see your comments on the PR, are they on a different branch or maybe not pushed?

As for the failure, yeah it looks like it's because of Julia 1.6:
LoadError: Unsatisfiable requirements detected for package FiniteDiff
...
restricted by julia compatibility requirements to versions: 2.0.0-2.22.0 or uninstalled — no versions left

I added a conditional to not run on v1.6

[ronisbr]

Thanks @jmurphy6895 ! I forgot to push the comments :)

[jmurphy6895]

I see why it's still failing, it's not the test it's the .toml compat, I'll loosen it and see if the tests still pass

[jmurphy6895]

I'll get to these changes after work today

[jmurphy6895]

W.r.t. to the failing CI, it looks like the DifferentiationInterface will not work with Julia 1.6 for FiniteDIff (Julia 1.6 requires <=v2.22 and DifferentiationInterface requires >=2.23)

I don't explicitly need the DifferentiationInterface package to test this, it will just make testing all the different AD backends very easy in the future. I'll defer to you on what you'd like to happen here

[ronisbr]

Hi @jmurphy6895 !

Let's ignore those errors. As soon as we merge this commit, I turn off the tests for Julia 1.6. The last thing (sorry, I missed it before): can you please return the version to 0.1.5? I prefer to have a separate commit with the version bump. After that, I will merge this PR and tag the new release :)

[jmurphy6895]

@ronisbr Done!

[ronisbr]

Awesome! Let's the CI run and I will merge this! Thank you very much for the PR @jmurphy6895 !

[ronisbr]

Hi @jmurphy6895 !

I did some modifications:

1. Fixed precompilation warnings.
2. Improved docstrings (making them less verbose).
3. Most important: fixed J2000.0 epoch definition. Your PR revealed a long-lasting bug between the code and the documentation. When we pass no time information, we fallback to J2000.0 epoch, which is 2000-01-01T12:00:00 instead of 2000-01-01T00:00:00. Hence, I modified everything to provide the coefficients at the correct definition of J2000.0 epoch.

[ronisbr]

Can you please test if I broke something related to autodiff? If not, I will tag the new release.

[jmurphy6895]

LGTM, one step closer to a fully differentiable propagator!