From panchromatic galaxy modeling to machine learning: how modern tools can help us learn more about galaxies
Dr. Médéric Boquien
Universidad de Antofagasta
The physical processes driving the transformation of baryonic matter in galaxies leave a profound imprint on their electromagnetic emission. Extracting the information that is deeply intertwined in this emission in order to estimate the physical properties of galaxies has become essential in our effort to solve the mysteries of galaxy formation and evolution. In this endeavor, astrophysicists have developed increasingly complex tools over the past decades. Starting from simple models of the optical emission of passive galaxies, modern panchromatic models of galaxy emission now cover multiple orders of magnitude in wavelength and can be routinely used to estimate a broad range of physical properties of galaxies (age, star formation rate, stellar mass, dust luminosity, attenuation laws, etc.). In the first part of this talk, I will describe the approach we have taken to develop the new panchromatic modeling code CIGALE and I will present a number of recent studies it has made possible from the characterization of great samples of galaxies to constraints on the interplay between light and dust. Panchromatic models are however not the only tools at our disposal. Another type of technique that has been seeing great developments over the past few years is machine learning. I will show a new effort to apply machine learning techniques to large spectroscopic surveys in order to find the most unusual (and hopefully the most astrophysically interesting) objects.