Interpolation of Sagittarius A* multiwavelength data using a transformer based machine learning model
Presenter: Gabriel Sasseville
Title: Interpolation of Sagittarius A* multiwavelength data using a transformer based machine learning model
Date/Time: Friday, June 21st, 12:10 PM
Abstract: Understanding the dynamics of the supermassive black hole, Sagittarius A* (Sgr A), at the center of our galaxy is crucial for unraveling the fundamental mysteries of the Universe. This supermassive black hole displays variability across various wavelengths, suggesting potential correlations that could help understand the underlying physical processes governing its dynamics. However, observations of Sgr A across multiple wavelengths are often irregular, asynchronous, and incomplete, posing significant challenges for correlation analysis. To address this, our project aims to adapt and enhance a transformer based machine learning algorithm, the TripletFormer, for analyzing multi-wavelength data of Sgr A. Originally developed for asynchronous time series interpolation in healthcare, TripletFormer offers a promising base framework for addressing the complexities of astrophysical data. Our objective, therefore, is to develop a model capable of interpolating Sgr A’s data and to subsequently conduct cross-correlation analysis. Through this interdisciplinary approach, we aim to unveil deeper insights into the dynamics of Sgr A* and advance our understanding of supermassive black holes, while simultaneously providing a generalized model that can be used with other astrophysical data.