Closed wbalmer closed 7 months ago
Yes I agree that would be interesting to test! You meant the option to fit fsed for each cloud species separately and global Kzz?
Yes, exactly! Doing so would allow me to reproduce exactly ZJ's retrieval tested on AF Lep b in his 2023 paper, but including some of the nice species only features, like fitting the GP for the covariance matrix estimation.
I have implemented the global_fsed
parameter in commit 055af59e3598dc9d970e268b1e255884e2a7ae24. You can set it to False
for retrieving the fsed
parameter for each cloud species individually. For each species, it will use the same fsed
range that is specified in bounds
.
I did some brief testing but please let me know if this works fine!
I ran some tests overnight and it appears to work great! I'll let you know how the converged results look.
Hi Tomas! In petitRADTRANS.retrieval.cloud_cond, you can set a global fsed and Kzz, or a fsed for each cloud species and a global fsed. How difficult would this be to implement in species? I think it would be useful, given that different clouds might condense or be lofted at different rates, resulting in different exponential decays in opacity.