Dr. LI Minyu and Prof. ZHU Liying from the Yunnan Observatories, Chinese Academy of Sciences, have proposed a neural network-based automated search approach for heartbeat stars. Their findings have been published in The Astronomical Journal. This method significantly improves the efficiency and accuracy of detecting these rare binary systems, while requiring  relatively low computational learning costs.

Heartbeat stars are eccentric-orbit binary systems that exhibit periodic brightening events, producing light curves similar to electrocardiograms. These systems often show tidally excited oscillations (TEOs), providing a unique opportunity to study tidal interactions, stellar internal structure, and binary evolution. However, the morphological diversity of their light curves makes manual identification challenging, especially in large-scale astronomical surveys.

To address this, the research team proposed a new method using orbital harmonics from Fourier spectra as input features to train a neural network classifier. This approach achieved an identification accuracy of 86% for known heartbeat stars and remained effective for systems with TEOs. The method has led to a patent application titled "A Neural Network-Based Classification Method for Periodic Variable Star Light Curves".

Using photometric data from the Kepler Space Telescope, the team analyzed 153 heartbeat systems and established the largest parameter database of Kepler heartbeat stars to date. They discovered 21 new systems exhibiting TEOs and developed a tool to automatically distinguish between harmonic and non-harmonic oscillations. They identified the phase and mode for 14 samples with TEOs, revealing internal structural features based on pulsation theory.

These research achievements will provide crucial methodological support for future studies using data from the TESS mission and China’s upcoming CSST and ET projects. The study received funding including: the Yunnan Fundamental Research Projects, the International Partnership Program of Chinese Academy of Sciences, the China Manned Space Program with grant, the CAS “Light of West China” Program, the Basic Research Project of Yunnan Province, the Yunnan Revitalization Talent Support Program.

Figure 1: Light curves of four heartbeat stars exhibiting tidally excited oscillations. Image by LI.

Figure 2: Test results of the neural network to recognize the known OGLE, Kepler, and TESS heartbeat stars as well as eccentric-orbit binaries. Red dots denote heartbeat stars exhibiting tidally excited oscillations, while black dots represent those without such oscillations. Image by LI.

Figure 3: Analytic procedure of tidally excited oscillations and their pulsation phase and mode identification for KIC 11403032. Image by LI.

Contact:
LI Minyu
Yunnan Observatories, CAS
E-mail:liminyu@ynao.ac.cn