Abstract
Galaxy number density as a function of intrinsic brightness, i.e., the galactic luminosity functions (LF), is one of the most fundamental observables for testing different cosmological models, including but not exclusive to different dark matter models. The abundance of faint galaxies at high redshifts is particularly informative in comparing two main-stream cold dark matter (CDM) models: dark matter as heavy particle, well-reproducing the large scale structure but over-predicting subhalos; and dark matter as ultralight particles, forming waves and condensates on astronomical scales and suppressing small scale structure formation with their wavy nature. To better detect faint galaxies at high redshifts, observations have increasingly turned to massive galaxy clusters for use as strong gravitational lenses, reaching 4-5 magnitudes fainter than would otherwise be possible. Nonetheless, reliable faint-end LFs remain hard to obtain in strong lensing fields, with accurately determining galaxy redshifts being the most fundamental challenge. Recently, Shipley et al, 2018 (S18) reported an excess of z~4 galaxies in the Hubble Frontier Fields (HFF) clusters relative to their accompanying blank non-lensing fields, for which no clear explanation has been offered. In this talk, I will demonstrate that the observed excess is the result of strong contamination by the misidentified low-redshift galaxies sharing similar spectral energy distributions as their high-z counterparts. I will also provide a quantitative estimate on the contamination level, and demonstrate that the presence of such strong contamination leads to unphysical turn-ups in the faint-end of UV LFs, with some seen examples including the LFs measured by Livermore et al, 2017. I will next comment on the prospect of individually mitigating the contaminants to achieve a more reliable test on the faint-end of UV LFs in the HFF cluster fields, highlighting the power of combining deep observations from the James Webb Space Telescope (JWST) with existing HFF observations. For HFF fields that currently do not have JWST data available, I will also demonstrate that deep learning methods work efficiently at learning the intricate differences between low-z interlopers and high-z galaxies, thereby extending the individual mitigating power of interlopers to those HFF fields. With the sample of machine learning cleaned high-z galaxies, I demonstrate that no evidence of faint-end turnovers in LFs was found from the observed surface number densities in differently magnified regions. And I will discuss the implications of this result on the wave dark matter models, in particular, I will demonstrate that the mass of the wave dark matter particle is constrained to be above 2.97E-22eV at 95% confidence level.
Anyone interested is welcome to attend.