Matteo Bachetti

Hi @matteolucchini1, good catch! Between the two methods, I would absolutely vote for using the machinery in `avg_pds_from_timeseries`, as it is the better tested on a variety of input data....

@matteolucchini1 I don't agree with this though. The current implementation in `fourier.py` works around a lot of corner cases related to floating point rounding. Plain `np.int` is subject to a...

`avg_pds_from_timeseries` works with binned and unbinned data, and does the inference on the size of the timeseries internally thanks to `fix_segment_size_to_integer_samples`. We could use the same method outside, or move...

@kashish2210 thanks for your PR. However, pandoc is among the required dependencies to build the documentation, so this should never happen, and I don't think `setup.py` is the correct place...

@spranav1205 I guess one could just use `min` to avoid that the "ones" array has no elements

> Yes, that would certainly work. But I don't understand one thing. Say `start = 1`, `width = 2`, and `self.dt = 4`. The impulse should not appear while sampling,...

@andresgur in all our analysis, what counts is the _good_ light curve bins, those inside GTIs. AveragedPowerspectrum etc will always ignore data outside GTIs. Therefore, tstart and tseg are better...

@andresgur I thought a little about this, and I don't know how to solve this without breaking the API. tseg is used sparsely and always assumed to be from a...

@andresgur Thanks for your PR. I made a slight modification to avoid breaking the API.

> thanks @matteobachetti ! Is there anything else for me to do? Sorry still learning this whole PR thing No worries! I'm taking care of solving the API change issues....