Option to use CBVs and better priors #63
Option to use CBVs and better priors #63christinahedges merged 69 commits intoSSDataLab:masterSSDataLab/psfmachine:masterfrom
Conversation
Co-authored-by: Jorge Martínez-Palomera <jorgemarpa@users.noreply.github.com>
Co-authored-by: Jorge Martínez-Palomera <jorgemarpa@users.noreply.github.com>
…o machine-perturbation
Co-authored-by: Christina Hedges <14965634+christinahedges@users.noreply.github.com>
…o perturbation-cbv
…o perturbation-cbv
|
Using smooth CBVs instead of PCA components and adjusting the priors for the perturbation model improved the deviation from the x-axis: sum of the
Data points of the 1-1 relation are sources with mismatching mean light curves before and after the time model and sources with spiky features (or added variability) after applying the time model.
After narrowing the priors the sources with bad time models are fixed. |
christinahedges
left a comment
There was a problem hiding this comment.
This is 99.999%, I will merge when @jorgemarpa checks that the segments keyword is the desired behavior.
Two small changes/new features:
machine.build_time_model()acceptsother_vectorsas argument to add other type of model components from user-level, e.g. using CBVs as regressors.PerturbationMatrix. The prior follows ~ N(1, 0,5) which enforces "small" deviations from the mean model. This accounts for the mismatch in mean flux we saw before between PSF and PSF-NOVA and reduces the chances to get outlier data points.TODO:
test_perturbation_matrix3d()Requires #60 to be merged first.