g-band plateau at 400-500 days

[JTerwel]

The model, in its current version, has a constant magnitude between from roughly 400 to 500 days. Besides this, the light curve does not decline in a linear fashion as would be expected.

The reason for the the non-linear decay in mag space is easy: The model is coded to decrease linearly in flux space at late-times. This was needed to avoid negative regions in the interpolated spectra, causing wildly unpredictable light curves.

The root cause of all this seems (to me) to be that the individual data points of a given spectrum may be negative due to unlisted uncertainties. This causes unpredictable behaviour when interpolating in both the time and wavelength directions.

A possible solution, or at least something worth a try, is trying to smoothen each spectrum enough to eliminate this issue, and then to resample the smoothened spectrum. These smoothened, resampled spectra are then used to make the model. If this works properly (i.e. the way I envision it) it should also decrease / remove the need for the linear time interpolation at late phases.

[JTerwel]

This is definitely something worth trying, and I will do so when I find the time for it. (That may take a while though)