astrophysics

We could look into using the diskcache package to cache MAST search operations:

https://pypi.org/project/diskcache/

The package is a disk-based key-value store which makes it easy to specify an expiration time for each entry. This would allow us to e.g. cache search queries for 24 hours, which would significantly speed up searches across different Python processes, whilst largely avoiding

It would be useful to keep track of how many retries a simulation did in the amr_diag.out file, which already records the coarse timestep info

The python implementation of leapfrog does not make use of the fact that drifts can be combined when not storing output, causing it to be slower than necessary. Comparing to the C implementation, it also appears to not be as precise.

We would like to make the BlackBodySimpleSource.create_packets() method (in tardis/montecarlo/packet_source.py) unit-sensitive. That is, we would like to take in quantities with units and return quantities with units.

Additional context
A potential thing to think about here is how this method is used in the code, and having that be unit-sensitive as well. Remembering that this class ca

🎯 Todo

For the parsers in io.hitran, only groups 1-2 and classes 1, 4-6 are implemented.
Notably, group 6 and class 3 are missing which are needed for linear diatomic molecules.
But this is surprising to me as I never had issues with this... am I missing something?

How they should be parsed can be found in the HITRAN2004 paper:
https://hitran.org/media/refs/HITRAN-2004.pdf
tables 3 & 4

As per comment here: https://github.com/StingraySoftware/stingray/pull/610/files#r782942772

The following seminar work of Hernquist details a light profile called the Hernquist profile:

https://ui.adsabs.harvard.edu/abs/1990ApJ...356..359H/abstract

The intensity calculation for a spherical model is given by equations (32-35):

It would be good to implement this light

It was noted in the review for our JOSS paper that the keyword argument noise for the function redshifts_from_comoving_density has a misleading name. See the review comment here. They suggested renaming to fixed_N. My own suggestion would be shot_noise (or perhaps poisson). Since this would require a break