Exoplanet reference data

[nichollsh]

PROTEUS uses exoplanet reference data to make certain plots. These are obtained from OSF, and based on the DACE database (https://dace.unige.ch/exoplanets).

This means that the population is not updated as new exoplanets are discovered.

We could consider using the new ExoAtlas library to obtain data on exoplanet populations instead. It's installable via pip. https://github.com/zkbt/exoatlas/

This will become increasingly relevant as we move towards larger ensembles and 'retrievals' with PROTEUS (#436, FormingWorlds/PROTEUS#424)

[timlichtenberg]

Agree, though DACE is in principle also updated from time to time. There are no perfect choices, so if you think this new ExoAtlas is a good way to go, then fine for me.

[nichollsh]

On this note: do we know what the future holds for DACE? I understand that PlanetS is ending. If DACE is no longer available, we will need to revisit our spectral file pipeline and potentially run HELIOS_K ourselves to generate opacities.

[timlichtenberg]

I would be very surprised if Geneva is just dropping the website tbh.

[nichollsh]

I have used this exoatlas library in the InferAGNI tool. Would be valuable to incorporate into PROTEUS for comparison with exoplanet populations. E.g. when doing bayesian inference, or with upcoming large projects more broadly.

https://github.com/nichollsh/InferAGNI/blob/main/src/inferagni/planets.py

[timlichtenberg]

This would presumable go into the data repo? (Coming after the current interior-refactor)

[nichollsh]

It could do.

However, using this exoatlas library actually means that we do not need to host the data ourselves. The library obtains data from the NASA exoplanet archive directly. Perhaps this could be seen as a reason not to use it, since then we have less control over how the data are obtained.

What do you think? We could instead write our own version of exoatlas within the larger fwl-io repo?

[timlichtenberg]

It would download a data file though, right?

[nichollsh]

It would download a data file though, right?

It does download data, yes. You can specify where this goes by setting an environment variable.

[timlichtenberg]

Fine for me if you want to implement it rifght now in PROTEUS; afterwards I think this type of data downloading should be handled by the fwl-io module. But that can be done during the development of the module itself.