Detecting Type II-P Supernova Precursor Emission via Machine Learning
Open Access
- Author:
- Tartaglia, Anna
- Area of Honors:
- Astronomy and Astrophysics
- Degree:
- Bachelor of Science
- Document Type:
- Thesis
- Thesis Supervisors:
- Donald P Schneider, Thesis Supervisor
Cindy Yuexing Gulis, Thesis Honors Advisor - Keywords:
- Transient Astronomy
Supernovae
Machine Learning - Abstract:
- The final years of Red Supergiant (RSG) stars present a challenging observational window. However, insights gleaned from photometric and spectroscopic measurements of Type II-P supernovae reveal evidence of eruptive mass loss events in these stars, which precede supernova shock breakout (SBO) by months to years. With limited observation and modeling capabilities within this critical timeframe, the driving mechanisms behind these eruptions, their manifestations, and subsequent impacts on stellar evolution remain elusive. In this paper, we construct and train a Convolutional Neural Network to detect precursor eruptions within light curves. The network is designed in preparation for the forthcoming Vera C. Rubin Legacy Survey of Space and Time (LSST), a 10-year Southern sky survey anticipated to revolutionize transient observations. We develop models of eruptions and wield them to simulate realistic light curves in the depth and cadence of LSST for network training. Our study estimates a predicted detection rate of $\sim11 - 50$ precursor events per year, depending on different parameter distributions. The network exhibits strong performance for nearby events, with reliable detections achievable up to $z \lesssim 0.03$. In cases where precursors remain undetected, our network can still offer significant insights by delineating the limits of potential precursor events. These insights will contribute to a deeper understanding of the nuanced processes underlying stellar evolution and transient phenomena.