search

rasg-affiliates / 21cmSense

Code for calculating the sensitivity of 21cm experiments to the EoR power spectrum
MIT License
7 stars 9 forks source link

Regresion Test #70

Closed steven-murray closed 10 months ago

steven-murray commented 2 years ago

It would be good to have a test where it just runs a full (small) sensitivity calculation and compares against output from @jpober 's original code. We can just store that output in the codebase for testing against.

jpober commented 2 years ago

I'm not sure I can even run my original code anymore, but I agree this is a very good idea. Let me know if you're having any issues making headway on it.

jpober commented 1 year ago

With a lot of handholding (and one critical bug fix), I was able to reproduce the HERA 331 results from HERA Memo 4 (http://reionization.org/wp-content/uploads/2015/05/HERA4_sensecalc.pdf). The following plots show both sensitivity curves on top of each other (you can really only see one set of points because they overlap) and the absolute difference between them. Screenshot 2023-07-05 at 9 28 03 AM Screenshot 2023-07-05 at 9 35 30 AM Unfortunately, I think too many tweaks were necessary to make this a routine test. I had to change the angular diameter distance and H(z) calculations (the old code used the Furlanetto et al. 2006 approximations, not the full ones), the sky noise calculation, and obs_duration, and several other small things. That said, perhaps once the bug fix is merged in, we can define a reference test using the new code for all future runs of the new code to be compared against.