Prof Chris Lintott

Abstract:

Modern radio astronomy instruments generate vast amounts of data, and the increasingly challenging radio frequency interference (RFI) environment necessitates ever-more sophisticated RFI rejection algorithms. The ‘needle in a haystack’ nature of searches for transients and technosignatures requires us to develop methods that can determine whether a signal of interest has unique properties, or is a part of some larger set of pernicious RFI. In the past, this vetting has required onerous manual inspection of very large numbers of signals. In this paper, we present a fast and modular deep learning algorithm to search for lookalike signals of interest in radio spectrogram data. First, we trained a β-variational autoencoder on signals returned by an energy detection algorithm. We then adapted a positional embedding layer from classical transformer architecture to a embed additional metadata, which we demonstrate using a frequency-based embedding. Next we used the encoder component of the β-variational autoencoder to extract features from small (∼715 Hz, with a resolution of 2.79 Hz per frequency bin) windows in the radio spectrogram. We used our algorithm to conduct a search for a given query (encoded signal of interest) on a set of signals (encoded features of searched items) to produce the top candidates with similar features. We successfully demonstrate that the algorithm retrieves signals with similar appearance, given only the original radio spectrogram data. This algorithm can be used to improve the efficiency of vetting signals of interest in technosignature searches, but could also be applied to a wider variety of searches for ‘lookalike’ signals in large astronomical data sets.

Abstract:

The Milky Way is thought to host a huge population of interstellar objects (ISOs), numbering approximately $10^{15}\mathrm{pc}^{-3}$ around the Sun, which are formed and shaped by a diverse set of processes ranging from planet formation to galactic dynamics. We define a novel framework: firstly to predict the properties of this Galactic ISO population by combining models of processes across planetary and galactic scales, and secondly to make inferences about the processes modelled, by comparing the predicted population to what is observed. We predict the spatial and compositional distribution of the Galaxy's population of ISOs by modelling the Galactic stellar population with data from the APOGEE survey and combining this with a protoplanetary disk chemistry model. Selecting ISO water mass fraction as an example observable quantity, we evaluate its distribution both at the position of the Sun and averaged over the Galactic disk; our prediction for the Solar neighbourhood is compatible with the inferred water mass fraction of 2I/Borisov. We show that the well-studied Galactic stellar metallicity gradient has a corresponding ISO compositional gradient. We also demonstrate the inference part of the framework by using the current observed ISO composition distribution to constrain the parent star metallicity dependence of the ISO production rate. This constraint, and other inferences made with this framework, will improve dramatically as the Vera C. Rubin Observatory Legacy Survey of Space and Time (LSST) progresses and more ISOs are observed. Finally, we explore generalisations of this framework to other Galactic populations, such as that of exoplanets.Comment: Accepted to A