Predictive monitoring of aperiodic sources

点击翻译按钮获取中文摘要

Many studies of objects in the sky that vary in brightness look only at periodic variation, with a regular and repeating change with time. However, most variable astronomical sources are aperiodic, showing erratic and irregular changes. They remain poorly studied in comparison to periodic sources, even though they can play a key role in our understanding of complex dynamic physical environments, from stars, to gas and dust flows, to galaxies. This work studies aperiodic variability using the data from two NSF-supported projects, the Catalina Real-time Transient Survey (CRTS), and the newly begun Zwicky Transient Facility (ZTF). This is the first large-scale systematic study of these phenomena, and it will be the definitive study, well into the next decade. It is an excellent case study for data-intensive science, applying state-of-the-art machine learning techniques to real data. It will expose students to cutting-edge data science, and its products will contribute to projects that will train the next generation in how to handle big astronomical data.Even though the majority of variable sources are aperiodic, they are poorly understood, and for well-known examples like quasars and young stellar objects many fundamental questions remain about the physical mechanisms behind their optical variability. New sky surveys are enabling systematic studies of variability and discovering many new phenomena, including sub-parsec separated quasar binaries, multi-year long flares attributable to microlensing of explosive activity in the accretion disk, and changing-state sources indicative of variable accretion rates. These extreme behaviors should be easily discoverable with modern robust statistical methods. This work will create generative data-derived models with novel non-parametric discriminating features from CRTS data, and use the models to predict the future behavior of aperiodic sources, which can then be monitored in real-time using ZTF and other synoptic facilities. Prior work focused on robust statistical characterization and identifying extreme variable classes. This will be extended to more general aperiodic sources and more sophisticated non-parametric generative models. The unprecedented sky coverage of ZTF joins with the unequalled archival coverage of CRTS to make this the definitive study. In keeping with the CRTS Open Data and ZTF alert policies, all transitioning objects identified will be released to the community.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.