SampEn and MSE frequently report "inf" values

[hasibagen]

Hello,

I've been using the EntropyHub package to compute Sample Entropy (SampEn) and Multiscale Sample Entropy (MSE) for EEG and fMRI data. However, I've encountered frequent occurrences where these functions return "inf" values.

Sample details:

I generated synthetic data to test this:

np.random.seed(42)
n_samples = 2000
time = np.linspace(-30, 60, n_samples)

hbo_signal = (
    0.2 * np.sin(0.2 * time)
    + 0.1 * np.sin(0.05 * time)
    + 0.03 * np.random.normal(size=n_samples)
)
hbr_signal = -0.1 * np.sin(0.2 * time) + 0.05 * np.random.normal(size=n_samples)

Mobj = EH.MSobject("SampEn")
MSx, CI = EH.MSEn(hbo_signal, Mobj, Scales=5, Methodx="modified")

While trying to compute SampEn, the result often comes back as "inf" for certain scales.
Question: Why are SampEn and MSE reporting "inf" values so frequently, and is there a way to prevent this issue from occurring? I've observed this both with real and synthetic data, and the issue persists across different packages.

Your help or guidance would be greatly appreciated!