COSMOS2020: Manifold Learning to Estimate Physical Parameters in Large Galaxy Surveys

We present a novel method to estimate galaxy physical properties from spectral energy distributions (SEDs), alternate to template fitting techniques and based on self-organizing maps (SOM) to learn the high-dimensional manifold of a photometric galaxy catalog. The method has been previously tested with hydrodynamical simulations in Davidzon et al. (2019) while here is applied to real data for the first time. It is crucial for its implementation to build the SOM with a high quality, panchromatic data set, which we elect to be the "COSMOS2020" galaxy catalog. After the training and calibration steps with COSMOS2020, other galaxies can be processed through SOM to obtain an estimate of their stellar mass and star formation rate (SFR). Both quantities result to be in good agreement with independent measurements derived from more extended photometric baseline, and also their combination (i.e., the SFR vs. stellar mass diagram) shows a main sequence of star forming galaxies consistent with previous studies. We discuss the advantages of this method compared to traditional SED fitting, highlighting the impact of having, instead of the usual synthetic templates, a collection of empirical SEDs built by the SOM in a "data-driven" way. Such an approach also allows, even for extremely large data sets, an efficient visual inspection to identify photometric errors or peculiar galaxy types. Considering in addition the computational speed of this new estimator, we argue that it will play a valuable role in the analysis of oncoming large-area surveys like Euclid or the Legacy Survey of Space and Time at the Vera Cooper Rubin Telescope.