Thanks for the explanation! It actually seems like it is working to me. IF you look at your first sweep iteration at 5 GHz and you place an optical time domain visualizer before and after the fiber you can look at how the chirp changes across the intensity profile.
You’ll notice in the attached image that on the left there is the initial predistorted pulses with an almost linear negative chirp over the pulse. On the right you have the post fiber signal (I disabled nonlinearity, PMD, third order dispersion and attenuation). The pulses are very compresses and the chirp is still negative but much smaller.
There seems to still be a problem with the input signal as the chirp is not linear over the entire power of the pulses.
Regards