Matteo Bachetti
Matteo Bachetti
As discussed in #910, tstart should be defined from gtis, not the `time` attribute, but for the rest I agree that this should be fixed!
👍 for this feature.
@spranav1205 thanks! A few comments: edge cases should not be plainly 0.5, but account for the exact percentage of zeros and one in a given bin. Also, I suggest implementing...
Hi @pupperemeritus, it's very interesting. Isn't it a duplicate of the astropy implementation though? What would we add to it?
@pupperemeritus I certainly like experimenting with new algorithms that promise to be a lot more accurate; for a matter of fairness, I will accept a new implementation only if it's...
@pupperemeritus another route might be asking the original author to contribute their implementation to Stingray, and improve on their code through the various optimizations you propose.
Thanks for the discussion, I'm learning some new concepts here. One nitpick and one request for clarification: > > As for the performance you can pretty easily replace trig functions...
@kkrutkowski thanks! To reduce the black magic feeling, I think that if you open a PR with your current best implementation as a bunch of functions in `fourier.py` (where we...
@Sauravroy34 interesting, thanks. Do you have measurements for the 1D transform we use as well?
@Sauravroy34 would you mind re-running the benchmark inverting the order of the calls to numpy and scipy? Just as a sanity check