AstroAI Lunch Talks - April 1, 2024 - Tomas Ahumada & Aizhan Akhmetzhanova
01 Apr 2024 - Joshua Wing
The video can be found here: https://www.youtube.com/watch?v=kzZcMPeZuw0
Speaker: Tomas Ahumada (Caltech)
Title: Searching for gravitational wave optical counterparts with the Zwicky Transient Facility
Abstract: During the first half of the International Gravitational Wave Network (IGWN)’s fourth observing run (O4a), the Zwicky Transient Facility (ZTF) conducted a comprehensive search for kilonova (KN) counterparts to binary neutron star (BNS) and neutron star–black hole (NSBH) merger candidates. I will discuss a comprehensive study of the five high-significance BNS and NSBH candidates in O4a. Based on the ZTF observations of the high-significance events in O4a, we used a Bayesian approach to quantify the posterior probability of KN model parameters that are consistent with our non-detections. We find that KNe with initial absolute magnitudes fainter than -16 mag are favored, while the joint posterior probability of having a GW170817-like KN associated with all our O4a follow-ups was 0.5.
Additionally, we used the 5 confirmed events in O3 (1 BNS and 4 NSBH events), along with our O4a follow-ups, to assess the efficacy of our GW searches during the combined O3 and O4a period. Our combined filtered efficiency to detect a GW170817-like KN is 36%. We derived joint constraints on the underlying KN luminosity function based on our O3 and O4a follow-ups, determining that no more than 70% of KNe fading at 1 mag per day can peak at a magnitude brighter than -18 mag. We look forward to the second half of O4, with Virgo joining, as it will lend us further opportunities to better understand KN populations.
Speaker: Aizhan Akhmetzhanova (Harvard University)
Title: Self-Supervised Learning for Astrophysics and Cosmology
Abstract: The influx of massive amounts of data from current and upcoming cosmological surveys necessitates compression schemes that can efficiently summarize the data with minimal loss of information. Self-supervised machine learning has recently emerged as a powerful framework for learning meaningful representations of data across a variety of data modalities.
In my talk, I will first give a brief overview of the self-supervised machine learning framework and discuss some of its applications to astrophysical datasets. I will then present a method we developed that leverages the paradigm of self-supervised machine learning to construct representative summaries of massive datasets using simulation-based augmentations. Deploying the method on hydrodynamical cosmological simulations, I will show how this method can deliver highly informative summaries, which could be used for a variety of downstream tasks, including precise and accurate parameter inference. I will also discuss how this paradigm can be used to construct summary representations that are insensitive to prescribed systematic effects, such as the influence of baryonic physics.
Watch the talks below!