Probing the Bright End of the Rest-frame Ultraviolet Luminosity Function at z = 8–10 with Hubble Pure-parallel Imaging

Abstract

Looking for bright galaxies born in the early universe is fundamental to investigating the Epoch of Reionization, the era when the first stars and galaxies ionized the intergalactic medium. We utilize Hubble Space Telescope pure parallel imaging to select galaxy candidates at a time 500 to 650 million years after the Big Bang, which corresponds to redshifts z ∼ 8−10. These data come from the Brightest of Reionizing Galaxies Survey (BoRG) Cycle 22 dataset, which consists of pure-parallel imaging in ∼ 90 different lines of sight that sum up to an area of ∼ 420 arcmin 2 . This survey uses five filters and has the advantage (compared to the Cycle 21 BoRG program) of including imaging in the JH 140 band, covering continuous wavelengths from the visible to near-infrared (λ = 0.35µm -1.7µm). This allows us to perform reliable selection of galaxies at z ≥ 8 using the photometric redshift technique. We use these galaxy candidates to constrain the bright end of the rest-frame ultraviolet luminosity function in this epoch. These candidates are excellent targets for follow-up observations, particularly with the James Webb Space Telescope.

A quantitative assessment of completeness correction methods and public release of a versatile simulation code

“…For example, e.g. Finkelstein et al (2015); Rojas-Ruiz et al (2020) define the completeness 𝑃(𝑀 in ) as the fraction of recovered galaxies, regardless of their output magnitudes, in an input magnitude 𝑀 in bin:…”

“…The completeness defined by method 1 is the summation along the column (highlighted in purple) divided by the total number of injected galaxies in the injected magnitude 𝑀 in . We note that the completeness by Finkelstein et al (2015) is also a function of two additional parameters: half-light radius and UV spectral slope, but these are not considered by Rojas-Ruiz et al (2020). With this completeness definition, the effective comoving volume in an observed image is also a function of 𝑀 in : where 𝑉 is the comoving volume associated to the observed image.…”

“…One aspect of this method is that the number of observed galaxies with intrinsic 𝑀 in magnitudes (𝑁 obs (𝑀 in )) may be challenging to quantify. Finkelstein et al (2015) and Rojas-Ruiz et al (2020) found that the recovered fluxes of the galaxies in the injection-recovery simulation were typically fainter than the input fluxes. Therefore, they apply the typical offset found in the simulations to the fluxes measured for the observed galaxies to derive the best estimate of their intrinsic fluxes:…”

“…Oesch et al (2007) and Bouwens et al (2015) inject artificial sources drawn from an underlying flat magnitude distribution. Finkelstein et al (2015) and Rojas-Ruiz et al (2020) inject galaxies with a faint-heavy distribution. Bowler et al (2015) inject galaxies that either follow a Schechter distribution or a double-power law distribution.…”

“…Finkelstein et al (2015) found a similar conclusion regarding the evolution of the Schechter parameters with redshift but their UVLFs at 𝑧 > 6 show some mild tension with those of Bouwens et al at 1-2 𝜎 level. On the other hand, Bowler et al (2015), Calvi et al (2016), Rojas-Ruiz et al (2020) and Bowler et al (2020) found that the UVLFs at high redshifts either comparatively show an excess at the bright end (although they are all still consistent with each other within a few sigmas and therefore the differences are not conclusively significant) or are better fit with a double power-law function with little evolution at the bright end. They also detected a significant change in 𝑀 * with redshift.…”