Analytic characterization of the cumulant probability distribution

Derive a full analytic description of the cumulant probability density function, including its non-Gaussian tail behavior, to establish reliable statistical significance estimates for cumulant-based searches with large trials factors.

Background

The sensitivity analysis estimates the fourth-cumulant fluctuations for a natural Gaussian source and finds an approximately Gaussian distribution. However, the paper notes that the full cumulant estimator may have non-Gaussian tails, which are difficult to study numerically because simulations must begin with electric-field realizations and a sufficiently large number of realizations is computationally challenging.

A rigorous analytic probability density function would be needed to assess detection significance reliably, particularly when many sky positions and frequency channels generate large trials factors. The paper leaves this analytic characterization unresolved.

References

I do not have a full analytic description of the cumulant pdf, and since the simulations must start with the electric-field it is difficult to create enough realizations to study the tails of the pdf (they might be slightly heavy?). Looking at the components of $\kappa_4$ (Equation~\ref{eq:k4}), the distribution of the first term $\left< (EE*)2 \right>$ is a Weibull distribution with shape parameter $k=1/2$ and scale parameter $\lambda = s_N2$, leading to a standard deviation of $\sigma = \sqrt{20}\, s_N2/\sqrt{Bt}$. The second term $2 \left<EE^* \right>2$ has a non-central chi-squared distribution with $\sigma = 4 s_N2/\sqrt{Bt}$. However, the fluctuations of these terms are strongly correlated and the variance of the cumulant $\kappa_4$ is significantly narrower than the two terms. (Equation~\ref{eq:ksig} can be analytically confirmed using Wick's theorem, Matthew McQuinn personal communication.) Understanding the full non-Gaussian statistical significance (particularly when accounting for large trials factors) will necessitate an analytic understanding of the cumulant pdf and is left for future work.

— Identifying Casual SETI Transmissions and Ultra-Faint RFI with Cumulant Imaging  (2609.04472 - Morales, 3 Sep 2026) in Section 4, “Sensitivity”