S21B0070

High-z void galaxies, whose evolution has been driven almost completely by in-situ processes, are ideal targets to provide valuable insights into the role of external processes in driving galaxy evolution. We develop a new deep-learning based tool, VFnet, which extracts characteristics of 3D spatial distribution of point clouds to calculate the probability (P_UD) of being an under-density galaxy. The VFnet achieves 90% precision to find promising (recall=0.1%) galaxy candidates in underdense regions based on the sky distributions and (g−r) colors of surrounding galaxies. Under-density galaxies identified in the cosmological simulation show systematically less massive both in halo mass and stellar mass, lower SFR, and lower metallicity than other galaxies at z∼4. We apply the VFnet to the g-dropout galaxy catalog of the HSC-SSP Deep layer, and find five under-density galaxy candidates with high probability with P_UD>68%. We propose deep DEIMOS spectroscopy for these five void galaxy candidates to unveil their physical natures by detecting multiple rest-UV emission lines such as C iii] and C iv. We will, for the first time, shed light on the very rare high-z galaxy population in void, which can only be identified by the wide and deep HSC-SSP data set.

This document was translated from L^AT_EX by H^EV^EA.