Josh Borrow, University of Durham
Inter-Lagrangian Transfer: The story from SIMBA and EAGLE
We use the inter-Lagrangian framework from Borrow+ (2020) to analyse the SIMBA and EAGLE cosmological simulations. By classifying the origin of particles in haloes at z=0 based on their Lagrangian regions, we show that the different feedback models present in both simulations lead to a significant change in baryonic history of each halo. The explicit AGN jet mode feedback present in the SIMBA simulation acts preventatively to reduce the baryon fraction of haloes with M_H~1e12 msun by around 50%, primarily affecting baryons that would assemble from outside of the region that contains the dark matter in the halo. In EAGLE this preventative feedback mode increases strongly with halo mass, and in the most massive haloes ensures that they are unable to accrete baryons from outside of their own Lagrangian region as defined by the dark matter. This tension between the origin of baryonic matter in haloes in these two models provides a key insight into how their numerical implementations affects the growth of galaxies without relying on comparison to calibrated quantities.
Host: David Barnes
Daniel Muthukrishna, University of Cambridge
Real-Time Classification of Explosive Transients Using Deep Learning
Astronomers are facing an unprecedented era of big data, observing more phenomena than humans can possibly visually examine alone. Upcoming large-scale surveys such as the Legacy Survey of Space and Time (LSST) will observe millions of transient alerts each night: two orders of magnitude more than any survey to date. To meet this challenge, we have developed a novel time-series classification tool, RAPID (Real-time Automated Photometric IDentification), capable of quickly classifying multi-channel, sparse, time-series datasets into several astrophysical types. Using a deep recurrent neural network, we present the first method specifically designed to provide early classifications of astronomical transients, identifying transients from within a day of the initial alert, to the full lifetime of a light curve. We have begun running RAPID on the real-time Zwicky Transient Facility (ZTF) survey, and have successfully classified several transients well before peak luminosity. In this talk, I will explain the main parts of our deep learning architecture and describe our approach’s performance on simulated and real data streams.
Bio: I am a PhD candidate at the Institute of Astronomy, University of Cambridge, UK. I have previously worked as a graduate research student at the Australian National University, the Gemini South Observatory in La Serena, Chile, and at the University of California, Santa Cruz. Before this, I worked as an Electrical Engineer in Brisbane, Australia. I received two undergraduate degrees in a Bachelor of Engineering (Electrical & Aerospace) and a Bachelor of Science (Physics) from the University of Queensland in Australia. My research focuses on transient astrophysics and I am particularly interested in deep learning and Bayesian statistics applied to time-domain astronomy. I have worked with large-scale survey collaborations such as DES, LSST DESC, and 4MOST and develop novel neural network methods to analyse supernovae and other time-varying phenomena.
Host: George Ricker