Jan Swierczek-Jereczek

Is now part of the code but needs to be made more performant. In particular, the FFTs have to be performed in place.

Hi @Datseris, I tried to optimize the segmented analysis. Improving things like changing `argmin(abs.(t .- t1))` and using views, as you suggested, is easy. But really, the bulk of the...

> Actually I don't understand what you say. We only generate surrogates once for the sliding window. Can you explain to me what part of the code you are referring...

> I am confused about what you do in teh segmented analysis. I thought yuou generate the surrogates _once_ and then you take the same segmented windows as the original...

Sure. Our whole approach of estimating statistics needs timeseries (or segments of it) that are long enough - that's not even specific to the surrogates. That's also why we require...

From a design perspective I agree with you. This would be the easiest. However, one might want to avoid interpolation in some cases, as it biases e.g. the AR1-regression coefficient...

> We can rely on Interpolations.jl to access the unevenly sampled version via interpolating. I am referring to this sentence. I understood that you (always) want to access unevenly-spaced time...