Complex amplitudes in models

EDIT: Workaround at https://github.com/astropy/astropy/issues/8833#issuecomment-595057891

I have a complex problem that I’m hoping someone can help me fix.

I have the first N complex terms of the Fourier transform of a 1D data series, which can be used to reconstruct an approximation of the original data series. I would like to represent the approximation using astropy models, which can be achieved using Euler’s formula and a combination of Sine1D models. However, I need to construct a combination of Sine1D models which have complex amplitudes, like this:

import numpy as np
from astropy.modeling import models

fft = np.array([1323.05768443  +0.j        , -473.6786725 -120.72642364j,
       -199.59351462-147.75512434j,  -11.8952061  -41.34435797j,
         26.37171375 +56.57267775j,  -10.34686085 +61.77508734j,
        -26.32020494 +18.12691033j,    8.11154341  -6.89937425j,
         18.11907541  -6.4711258j ,    9.37345637  +9.30265208j])

N = 100 
n = np.arange(len(fft))
k = np.arange(N)[:, np.newaxis] 

m = models.Sine1D(frequency=k/N, phase=1/4) + models.Sine1D(amplitude=1j, frequency=k/N, phase=0)

Presently this raises an issue because the amplitude is expected to be convertable to a float. Is there any way around that?

[bmorris3]

@perrygreenfield I found a solution to my problem by recasting it as a real-valued-only problem.

[perrygreenfield]

Excellent! Thanks for working around this.