SNAD researchers have identified 11 previously undetected spatial anomalies using an artificial intelligence (AI) method.
The team examined digital images of the northern sky obtained using a kD tree in 2018 to detect spatial anomalies using the “nearest neighbor” method. The research then used machine learning algorithms to automate the research.
The study is published in New astronomy.
Survey the sky with AI
Astronomical discoveries have increased significantly in recent years due to large-scale astronomical studies. The Zwicky Transient Facility, for example, uses a wide-field camera to monitor the northern sky, generating around 1.4 TB of data each observation night with its catalog containing billions of objects.
However, manually processing these colossal amounts of data is extremely expensive and time-consuming. To overcome this, the SNAD team, made up of researchers from Russia, France and the United States, collaborated to design an automated process.
When analyzing astronomical objects, scientists observe their light curves, which demonstrate how an object’s brightness varies with time. Scientists first identify a flash of light in the sky, then track its progress to see if it gets brighter, dimmer or fades.
In their study, the researchers analyzed one million real lightcurves from the ZTF’s 2018 catalog and seven simulated livecurve models of the studied object types. They tracked a total of 40 parameters, including an object’s brightness amplitude and delay.
Konstantin Malanchev, co-author of the paper and postdoctoral fellow at the University of Illinois at Urbana-Champaign, commented: “We described the properties of our simulations using a set of features believed to be observed in astronomical bodies real. In the dataset of approximately one million objects, we were looking for super-powerful supernovae, type Ia supernovae, type II supernovae, and tidal disturbance events. We refer to these classes of objects as spatial anomalies. They are either very rare, with little known properties, or seem interesting enough to merit further study.
Subsequently, the team compared the light curve data of real objects to simulations using the kD tree algorithm, which is a geometric data structure for dividing space into parts. smaller by cutting it with hyperplanes, planes, lines or points. The algorithm was used to narrow the search range when searching for real objects with similar properties to these in all seven simulations.
Discover 11 new spatial anomalies
The researchers identified 15 nearest neighbors (real objects from the ZTF database) for each simulation – 105 matches in total, which were then visually examined for spatial anomalies. The manual verification process confirmed 11 spatial anomalies – seven were supernova candidates and four were active galactic nuclei candidates where tidal disturbance events could occur.
Maria Pruzhinskaya, co-author of the paper and researcher at the Sternberg Astronomical Institute, commented: “This is a very good result. In addition to the rare objects already discovered, we were able to detect several new ones previously missed by astronomers. This means that existing search algorithms can be improved to avoid missing such objects.
The study demonstrates that the method is very effective and easy to apply. Moreover, the method is universal and can be used to discover any astronomical object, not just rare types of supernovae.
Matvey Kornilov, Associate Professor in the Faculty of Physics at HSE University, concluded: “Astronomical and astrophysical phenomena that have not yet been discovered are, in fact, anomalies. Their observed manifestations are expected to differ from the properties of known objects. In the future, we will try to use our method to discover new classes of objects.