Gaussian auto-correlation for unevenly sampled data

This is a straighforward implementation using Gaussian RBF kernels of the auto-correlation function for a signal that is nonuniformly sampled. Usage:

import numpy as np
from gautocorr import gautocorr
t = 10 * np.linspace(0, 1, 100)**2  # (uneven) time points
x = np.cos(t)  # signal
tt = np.linspace(0, 3, 20)  # time lags to evaluate auto-correlation at
correlation = gautocorr(tt, t, x)

For a more involved example see gautocorr.py.

Requires numpy and numba.