On attempting to automate the identification of mixed dipole modes for subgiant stars

TL;DR

This paper presents an automated method to identify mixed dipole modes in subgiant stars using power spectra from Kepler data, improving accuracy and efficiency over manual techniques.

Contribution

It introduces a genetic algorithm-based approach to automatically locate dipole modes and estimate stellar parameters, reducing systematic errors in mode identification.

Findings

Successfully retrieved over 80% of modes within 3 μHz for most stars

Achieved smaller error bars for parameters compared to previous methods

Confirmed period spacing and coupling factors align with prior measurements

Abstract

The existence of mixed modes in stars is a marker of stellar evolution. Their detection serves for a better determination of stellar age. The goal of this paper is to identify the dipole modes in an automatic manner without human intervention. I use the power spectra obtained by the Kepler mission for the application of the method. I compute asymptotic dipole mode frequencies as a function of coupling factor and dipole period spacing, and other parameters. For each star, I collapse the power in an echelle diagramme aligned onto the monopole and dipole mixed modes. The power at the null frequency is used as a figure of merit. Using a genetic algorithm, I then optimise the figure of merit by adjusting the location of the dipole frequencies in the power spectrum}. Using published frequencies, I compare the asymptotic dipole mode frequencies with published frequencies. I also used published frequencies for deriving coupling factor and dipole period spacing using a non-linear least squares fit. I use Monte-Carlo simulations of the non-linear least square fit for deriving error bars for each parameters. From the 44 subgiants studied, the automatic identification allows to retrieve within 3 $\mu$Hz at least 80\% of the modes for 32 stars, and within 6 $\mu$Hz at least 90% of the modes for 37 stars. The optimised and fitted gravity-mode period spacing and coupling factor agree with previous measurements. Random errors for the mixed-mode parameters deduced from Monte-Carlo simulation are about 30-50 times smaller than previously determined errors, which are in fact systematic errors. The period spacing and coupling factors of mixed modes in subgiants are confirmed. The current automated procedure will need to be improved using a more accurate asymptotic model and/or proper statistical tests.