It would help us to avoid errors if we stated explicitly what our dimensions are (like epochs, frequencies, etc.) in the Spectrum objects (and probably other future ones). It would also help to have the corresponding indices for these dimensions, like the list of actual frequencies used.
Handling these could be simplified by using xarray under the hood but, as @dominikwelke rightly pointed out, this would also bring certain dependencies like python 3.7+ and pandas. Let's stick to numpy for the time being and evaluate the potential benefit of using xarray later.
It would help us to avoid errors if we stated explicitly what our dimensions are (like
epochs,frequencies, etc.) in theSpectrumobjects (and probably other future ones). It would also help to have the corresponding indices for these dimensions, like the list of actual frequencies used.Handling these could be simplified by using
xarrayunder the hood but, as @dominikwelke rightly pointed out, this would also bring certain dependencies like python 3.7+ andpandas. Let's stick tonumpyfor the time being and evaluate the potential benefit of usingxarraylater.