# Multiscale entropy analysis of astronomical time series. Discovering subclusters of hybrid pulsators

**Authors:** Jeroen Audenaert, Andrew Tkachenko  
**arXiv:** [2206.13529](https://arxiv.org/abs/2206.13529)  
**Date:** 2022-06-27  
**Summary page:** https://www.summarizepaper.com/en/paper/2206.13529v1/

## Key points

- The paper explores the use of multiscale entropy (MSE) analysis in characterizing stellar variability patterns and identifying subclusters of hybrid pulsators.
- MSE assesses the complexity of a signal across different timescales and has been successfully used in biomedical research.
- MSE is a powerful tool for capturing short- and long-term variability patterns in stellar light curves, providing insights into the pulsation structure of a star based on time-domain information only.
- Longer data sets have a positive effect on robustness but can still yield good results from 27-day light curves.
- The authors use MSE as the basis for an interpretable clustering framework to distinguish hybrid pulsators with both p- and g-modes from stars with only p-mode or g-mode pulsations without requiring previously labeled data.
- This unsupervised clustering framework is unbiased by previous knowledge.
- The authors demonstrate the effectiveness of their clustering framework by successfully identifying subclusters of hybrid pulsators in sets of δ Sct and γ Dor stars respectively.
- MSE is correlated to the frequency content of a stellar signal, particularly to near-core rotation rates of g-mode pulsators.
- Overall, this paper highlights the potential usefulness of MSE analysis in studying stellar variability patterns and identifying subclusters within populations of variable stars.

## Summary

This paper explores the use of multiscale entropy (MSE) analysis in characterizing stellar variability patterns and identifying subclusters of hybrid pulsators. MSE assesses the complexity of a signal across different timescales and has been successfully used in biomedical research. The authors study the algorithmic properties of MSE in a stellar variability context and link it with traditional astronomical time series analysis methods. They find that MSE is a powerful tool for capturing short- and long-term variability patterns in stellar light curves, providing insights into the pulsation structure of a star based on time-domain information only. The authors also investigate the effect of longer time series on MSE values, finding that longer data sets have a positive effect on robustness but can still yield good results from 27-day light curves. Subsequently, they use MSE as the basis for an interpretable clustering framework to distinguish hybrid pulsators with both p- and g-modes from stars with only p-mode or g-mode pulsations without requiring previously labeled data. This unsupervised clustering framework is unbiased by previous knowledge. The authors demonstrate the effectiveness of their clustering framework by successfully identifying subclusters of hybrid pulsators in sets of δ Sct and γ Dor stars respectively. Additionally, they show that MSE is correlated to the frequency content of a stellar signal, particularly to near-core rotation rates of g-mode pulsators. Overall, this paper highlights the potential usefulness of MSE analysis in studying stellar variability patterns and identifying subclusters within populations of variable stars. It provides valuable insights into how short- and long-term variability interact with each other based on time domain information only which can aid our understanding of stellar physics and evolution.