Closed taylorbell57 closed 6 months ago
Attention: Patch coverage is 57.14286%
with 3 lines
in your changes are missing coverage. Please review.
Project coverage is 58.49%. Comparing base (
ee4570c
) to head (6629d15
).
Files | Patch % | Lines |
---|---|---|
src/eureka/S1_detector_processing/s1_process.py | 50.00% | 2 Missing :warning: |
src/eureka/S5_lightcurve_fitting/fitters.py | 0.00% | 1 Missing :warning: |
:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.
@kevin218, just had to deal with a merge conflict, and now this is good to go again
This gives us access to the EMI correction step for MIRI data in Stage 1 (introduced in
jwst
v1.13.4) but requires us to usePython>=3.10
. After initial testing on my end, the change in Python version seems to have no consequences, but someone with an Apple Silicon computer should probably run a check. I've also added an extra testing suite: now the code will be tested with and without the PyMC3 optional install which is important as that option continues to fall behind because of that hard upper-limit on thejwst
version caused bytheano
's deprecation and the resultingnumpy
upper limit.I also added bug fixes in a couple places. Fixed an incorrect self-import style used in S1, fixed the chi-squared calculation in the case of no masked points (in which case
~lc.flux.mask
just givesTrue
which gets treated as a1
which in turn gives a deeply negative reduced chi-squared value), and added the ability to continue an emcee chain with a subset of the parameters that had been used before.