Unevenly-sampled signals: a general formalism of the Lomb-Scargle periodogram

TL;DR

This paper introduces a comprehensive matrix algebra framework for analyzing the statistical properties of the Lomb-Scargle periodogram, applicable to unevenly sampled signals and various noise types, enhancing detection reliability.

Contribution

It develops a general formalism that extends the Lomb-Scargle periodogram to handle diverse noise characteristics and sampling patterns, beyond the traditional white noise assumption.

Findings

Provides a matrix algebra-based analysis method

Applicable to colored and non-stationary noise

Enhances detection threshold determination

Abstract

The periodogram is a popular tool that tests whether a signal consists only of noise or if it also includes other components. The main issue of this method is to define a critical detection threshold that allows identification of a component other than noise, when a peak in the periodogram exceeds it. In the case of signals sampled on a regular time grid, determination of such a threshold is relatively simple. When the sampling is uneven, however, things are more complicated. The most popular solution in this case is to use the "Lomb-Scargle" periodogram, but this method can be used only when the noise is the realization of a zero-mean, white (i.e. flat-spectrum) random process. In this paper, we present a general formalism based on matrix algebra, which permits analysis of the statistical properties of a periodogram independently of the characteristics of noise (e.g. colored and/or non-stationary), as well as the characteristics of sampling.