Speaker
Description
Even though the gas around and between galaxies (the circumgalactic and intergalactic media, CGM and IGM) harbour about 90% of the baryons in the Universe, they remain significantly less understood than the matter associated with starlight from galaxies. In this talk, I will introduce the Baryon Cycle (ByCycle) project, a large high-resolution (R~20,000) 4MOST/VISTA spectroscopic survey that aims to bring these hidden baryons to light. By looking for MgII absorbers in the spectra of background quasars, we will trace cool (10^4 K) CGM gas found in the line of sight. With the help of a Variational Autoencoder that can reconstruct the pure quasar continua, we are able to effectively identify MgII absorbers via contextual anomaly detection without the need for any explicit spectral fitting. Thanks to a Random Forest, we are then able to reject a majority of false detections and achieve high purity (>~80%). Our framework makes it possible to process the hundreds of thousands of spectra our survey expects. I will present our machine learning methodology, its ongoing extension to more absorbing species such as CIV and FeII, and our promising preliminary results on mock spectra and the first batches of 4MOST spectra. I will finish with an overview of the novel science that ByCycle will make possible.