Time domain methods for X-ray and gamma-ray astronomy

TL;DR

This paper reviews and compares various statistical methods for detecting and characterizing variability in low count rate X-ray and gamma-ray sources, highlighting new multidimensional approaches and software tools in R.

Contribution

It introduces new multidimensional variability detection methods and discusses the application of both nonparametric and parametric tools in high energy astronomy.

Findings

Multiple statistical tests can detect variability effectively.

New multidimensional methods expand analysis capabilities.

Most methods are underutilized in high energy astronomy.

Abstract

A variety of statistical methods for understanding variability in the time domain for low count rate X-ray and gamma-ray sources are explored. Variability can be detected using nonparametric (Anderson-Darling and overdispersion tests) and parametric (sequential likelihood-based tests) tools. Once detected, variability can be characterized by nonparametric (autocorrelation function, structure function,wavelet analysis) and parametric (multiple change point model such as Bayesian Blocks, integer autoregressive models, C-statistic and Poisson regression) methods. New multidimensional variability detection approaches are outlined. Software packages designed for high energy data analysis are deficient but tools are available in the R statistical software environment. Most of the methods presented here are not commonly used in high energy astronomy.