search

kevin218 / Eureka

Eureka! is a data reduction and analysis pipeline intended for time-series observations with JWST.
https://eurekadocs.readthedocs.io/
MIT License
58 stars 45 forks source link

Fixing the S5 GP Model #557

Closed taylorbell57 closed 10 months ago

taylorbell57 commented 1 year ago

There were numerous bugs with the current GP model; this PR brings the GP model back to working order for individual fits and makes it so that the GP model can also be used for shared or multwhite fits.

Resolves #555

The GPModel.py file was long neglected and had numerous bugs in it, some of which were from updates to other models. I've significantly cleaned up the code and changed the fitted GP variables. So far I've tested this with success on a single photometric lightcurve as well as a multwhite fit of two photometric lightcurves. It can be a bit tricky to get the optimizer to converge when doign a multwhite fit, but that's to be expected and the underlying code works (successfully run with lsq).

The only outstanding thing to be done before this PR is ready for review is updating the documentation pages and ECF/EPF tempates to reflect the changes to the fitted GP parameters.

taylorbell57 commented 12 months ago

FYI, I'll try to find time today or tomorrow to get the documentation pages and ECF/EPF templates updated before we merge this into main

taylorbell57 commented 11 months ago

FYI I found a few bugs with the GP code that I want to patch before this PR gets merged. Working on those right now

taylorbell57 commented 10 months ago

@evamariaa, if you're happy with my two responses above can you mark them resolved and then I'll merge this PR. But do respond to them if you think anything further needs to be done

evamariaa commented 10 months ago

@evamariaa, if you're happy with my two responses above can you mark them resolved and then I'll merge this PR. But do respond to them if you think anything further needs to be done

yes, go ahead! great work!