Open settwi opened 6 months ago
Attention: Patch coverage is 50.29586%
with 84 lines
in your changes are missing coverage. Please review.
:exclamation: No coverage uploaded for pull request base (
main@2deabc0
). Click here to learn what that means.
:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.
@samaloney @KriSun95 @DanRyanIrish @hayesla tear this apart plz
PR Description
A crack at an updated "modular" fitting interface. Notebook included demonstrating fitting a "line" of "X-ray" data. No real tests written yet but that's on the agenda, if we want to continue with any of this.
I am curious as to what folks think!
Here's a copy/paste from the "prelude" to that Jupyter notebook:
Modular modeling framework example
This notebook shows how a fitting framework could be made, in the context of issue #81 and issue #126.
sunkit-spex
has (or will have?) a modular interface. First we define the data interface. For now it is only X-ray data but we can add more data types.Then we define the fitting interface. A
PhotonModel
represents any kind of model which represents photons which get counted. In the future we can add different models, likeIntegratedFluxModels
if the need arises; there is not a need to tie us to a base class right now.The first example tackled here is fitting data to a line. More complex models may be implemented however the user or developers see fit by modifying the
perform_fit
method in subsequent classes. The first class implemented is a Monte Carlo $\chi^2$ fitter.A lot of ideas here are similar to pr #130. I think we should pick and choose what we like from both and move from there.