pulse shaper
The device takes a light pulse, splits it by spectral component, then offset different part of the spectrum in time, before recombining it.
A basic question we might ask is "given a ℂ-spectrum" and chosen time offset per spectral band, what is the output like. The immediate answer would be "just shift the phase per color channel". If the spectral decomposition we manipulated was really discretized and each band only contains one single color, that would sound reasonable... however, we manipulate a pulse here, so I think this is conceptually not the best description : the pulse will necessarily contain a lot of different frequency, or else it cannot be a pulse.
here is an interative tool to visualize the issue, as well as alternative ways to model it.
Instead of describing the phenomenon like that, we could assume that we will influence the spectrum in a smooth way. I don't have data to guarantee that what I am doing is right, but this is what I'm suggested to do.
here is an interative tool to visualize this interpretation.
Now let's look at the distribution of the peak of the signal in time domain