Min S. YunMauro GiavaliscoHoujun MoMichael LavineGim, HansungGim, HansungGim, Hansung2024-04-262024-04-262018-092018-0910.7275/12622596https://hdl.handle.net/20.500.14394/17655One of the most important questions in modern astrophysics is how galaxies form and evolve. There are numerous processes involved in galaxy evolution, but the stellar mass buildup and supermassive black hole growth are two main drivers in galaxy evolution. Those activities are heavily obscured by dust, so we need another tracer without dust attenuation: low-frequency radio continuum observation. We understand the galaxy evolution through the deep radio continuum observations on the Great Observatories Origins Deep Survey (GOODS)-North, -South, and the COSMOS HI Large Extragalactic Survey (CHILES) fields. Exploiting the multi-wavelength dataset, we define the radio populations such as star-formation (SF) dominated, active galactic nuclei (AGN)-dominated, and passive galaxies by applying new selection criteria. Populations are defined according to the relative contributions of SF and AGN luminosities to the bolometric luminosities of host galaxies and star formation properties. We explore the physical properties of radio populations, interpret their properties in the context of galaxy evolution, and show that our results of radio populations are consistent with the galaxy evolution scenarios. These results are the same for the GOODS and CHILES fields. Our investigation of the radio spectral index and radio-FIR correlation for the GOODS fields show that the radio spectral index is characteristic of each population and the radio-FIR correlation is a robust tracer of star formation activity. The analysis of radio spectral index shows the prominence of steep spectrum sources at faint flux density first as expected in the Euclidean normalized number counts. We also show that the radio spectral index should be calculated by matching the beam sizes of images. The evolution of the radio-FIR correlation in SF-dominated galaxies is not found significantly with our data. On the other hand, the analysis of radio spectral index for the CHILES is not consistent with the result of the GOODS due to the larger difference of beam area. This inconsistency gives important implications for the future works.galaxy evolutionradio continuum observationfaint radio sourcesradio spectral indexradio-FIR correlationGOODS fieldsCHILESstar formationactive galactic nucleiAstrophysics and AstronomyExternal GalaxiesPhysical Sciences and Mathematicsdissertation