-
Type:
Task
-
Status: Done (View Workflow)
-
Priority:
Normal
-
Resolution: Done
-
Affects Version/s: None
-
Fix Version/s: None
-
Component/s: None
-
Labels:
-
Epic Link:
In PIPE2D-920, we made what we call "6k fluxmodeldata", which contained ~ 6,000 spectra from AMBRE synthetic spectra. We now have to make "60k fluxmodeldata": ~60,000 spectra with their parameters on the finer grid that Yamashita-san's plan requires.
1. We have to extend the format of broadband/photometries.fits so that its user can distinguish the spectra in the 6k package from the other interpolated spectra.
2. We have to make a 60k package. We may include in it a file that indicates the hyperparameters used in interpolating the spectra.
3. We have to modify makeFluxModelInterpolator.py so that it will be aware of the new format of broadband/photometries.fits, and it will make interpolator.pickle only from the 6k spectra. I think we should not include interpolator.pickle in the package but each user should run makeFluxModelInterpolator.py to make it, because the pickle depends on scipy's version.
I will make the first version of 60k package with the hyperparameters already known to work reasonably. I think I can find better hyperparameters by estimating the goodness of hyperparameters with cross validation, but that will be another issue.
Could you review this PR (drp_stella)?
Merged. Thanks for the review.
Full package is available here: https://hscdata.mtk.nao.ac.jp/hsc_bin_dist/pfs/fluxmodeldata-ambre-20220518-full.tar.xz
Change of appearance of the interpolated flux with respect to log(g).
Change of appearance of the interpolated flux with respect to Teff.
Judging merely from these two animations, I guess the interpolation was successful.
If you simply replace the small package with this one, drp_stella will probably suffer from short of memory. I have yet to revise drp_stella before closing this issue.