Prompt identification of rapidly evolving astrophysical transients in the Rubin Legacy Survey of Space and Time
Date
2021
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
University of Delaware
Abstract
Astronomical transients are one of the major subjects of research in astrophysics. Their studies give us a better understanding of the compositions and evolutions of stars, the formation of heavy elements, and even the evolutions of the universe. ☐ In the past decades, some new kinds of astronomical events have been discovered. In some cases these transients are short-lived. Unlike transients that have been known and studied for decades and that have lifetimes of months up to more than one year, e.g. supernovae, the lifetimes of these “fast transients” are as short as a couple of weeks, or, even, days or hours. These objects are so far poorly studied due to the limitation of present astrophysical surveys. ☐ Vera C. Rubin Observatory is a synoptic survey telescope under construction in Chile planning to start a 10-year survey photometric survey in 2024. The telescope will scan the night sky with an unprecedented depth (magnitude ~24) and more than a thousand astronomical transients will be discovered every night. One key mission of the Rubin Observatory is to discover and characterize transients. However, in order to characterize and study these fast transients, prompt follow-up observations (for example spectra or space-based observations at other wavelengths) are necessary within hours or days after discovery. This means the fast transients need to be rapidly identified with confidence since the follow-up resources are extremely scarce considering the volume of discoveries that Rubin Observatory will produce. ☐ To address this problem, the author built a probabilistic classifier that can be used to identify the fast and unusual transients from the slower-evolving ones with three observations taken in two photometric filters within two nights, measuring the transients’ color and rate of brightness change. This thesis describes the use of simulations of well-known astrophysical transient and variable phenomena to build a probability hypercube of observational characteristics. Given the large number of required calculations, this work delivers a small-size prototype of the hypercube along with code that enables the construction of the full-scale classifier leveraging High-Performance Computers in highly parallelized mode.
Description
Keywords
Fast transients, LSST, Observing strategy, Probability classifier