Closed
Description
Gammapy version
In log-log plots, like in SEDs, if you plot a pure the power-law component of a composite spectral_model that contains an absorber (e.g. when fitting something with EBL) I would expect to recover a perfect bow-tie shape. Instead, in I get something like this:
I think that is a buggy representation of that model. I guess this arises from converting from linear-space to log-spaced, but it probably should be handled and 'simetrized' on the fly by gammapy.
For the coding-sprint ?