A comparison of Bayesian and Fourier methods for frequency determination in asteroseismology

TL;DR

This paper compares Bayesian and Fourier methods for frequency determination in asteroseismology, finding both effective but Bayesian methods offer additional insights into potential misidentifications.

Contribution

It provides a direct comparison of Bayesian MCMC and Fourier methods for oscillation frequency detection using simulated data in asteroseismology.

Findings

Both methods perform similarly in frequency accuracy.

Bayesian methods can identify potential aliasing issues.

Fourier method is less computationally intensive.

Abstract

Bayesian methods are becoming more widely used in asteroseismic analysis. In particular, they are being used to determine oscillation frequencies, which are also commonly found by Fourier analysis. It is important to establish whether the Bayesian methods provide an improvement on Fourier methods. We compare, using simulated data, the standard iterative sine-wave fitting method against a Markov Chain Monte Carlo (MCMC) code that has been introduced to infer purely the frequencies of oscillation modes (Brewer et al. 2007). A uniform prior probability distribution function is used for the MCMC method. We find the methods do equally well at determining the correct oscillation frequencies, although the Bayesian method is able to highlight the possibility of a misidentification due to aliasing, which can be useful. In general, we suggest that the least computationally intensive method is preferable.