Alex Ji

This is important to do, but right now I don't think anything will have changed because it is almost literally a direct port. However after the next step (wrapping the...

@eteq yes, 0.5 sounds good to me! @brechmos-stsci the #210 framework looks good and I will adapt this to match that. `fit_continuum_sigmaclip` sounds good. I realized later it is somewhat...

Yeah, I agree this is a good idea! The only way I have thought to do this is to have spline-specific models and fitters that are meant to work together,...

@eteq I have put in a very basic spline fitter version and a simple test (rebased onto current version). Basically I have just rewrapped scipy's splines into the astropy modeling...

@rosteen it sounds like you have a plan in mind for this, so you should go ahead! I'm happy to contribute in any way that is helpful, but I haven't...

Yeah I am happy to do it, especially if not time sensitive. I can finish this before the end of May. But it would be very helpful to be pointed...

Not at all! On Wed, May 6, 2020 at 10:15 AM Nicholas Earl wrote: > @alexji Do you mind if I rebase and push > directly to your branch with...

It looks like there have been updates to the API for astropy.modeling since I wrote the SplineModel class. I will go through this and fix it up. For now, we...

OK I have fixed up the SplineModel so it passes the test (at least locally, travis is running). For the fitter, I am not sure what methods have to be...

@hcferguson 's common use case of specifying knot number is easy to include. It just involves a helper class to automatically generate evenly spaced knots over the spectrum's dispersion axis...