Protoclusters, progenitors of clusters of galaxies, are ideal targets to study the environmental effect. However, the extremely low spatial density of protoclusters makes it difficult to detect protoclusters from low- and high-mass at z>3, which prevent us from understanding for the general properties of galaxies in overdense environments. To tackle the issue, a new deep-learning method to detect protoclusters is developed: called as PCFNet, which can significantly increase the number of high-z protocluster samples and extend to a low-mass regime. PCFNet is applied to the g-dropout galaxies in the HSC-SSP Deep layer and finds over ∼120 protocluster candidates at z∼4. We propose MOIRCS multi-object spectroscopy of eight protocluster candidates over a wide mass range at 3.4<z<3.8 detected by PCFNet. Our science goals are (i) to measure the accurate redshift and its internal structure with [Oiii] emission line (ii) to investigate the diversity of protoclusters with different mass from SED modeling with multi-band photometry from u-band to K-band and the [Oiii]/Hβ emission line ratio (iii) to constrain the contribution of protoclusters to the cosmic star formation rate density.
This document was translated from LATEX by HEVEA.