Impact of redshift distribution uncertainties on Lyman-break galaxy cosmological parameter inference

TL;DR

The paper tackles the challenge of inferring cosmological parameters from Lyman-break galaxies (LBGs) when spectroscopic redshifts are unavailable for the full sample. It introduces a forward-model framework that combines Stellar Population Synthesis (SPS) with a flexible, GP-calibrated galaxy-population prior to generate redshift distributions $N(z)$ and their uncertainties, which are then marginalized in a Fisher forecast for an LSST-like survey. By analytically marginalizing over $N(z)$ using a linearized response matrix and a PCA-based redshift-distribution parameterization, the authors quantify how population-model uncertainties propagate into constraints on $\sigma_{8}$, $\Omega_{m}$, and galaxy biases, finding Planck-like precision for $\sigma_{8}$ under certain dust-model assumptions. The study also reveals that the treatment of dust attenuation in the galaxy population is a dominant systematic, significantly affecting interloper fractions and the resulting cosmological inferences. Overall, the work demonstrates the viability of photometric-only cosmology with LBGs while highlighting key model dependencies that guide future improvements in dust modelling and SPS realism.

Abstract

A significant number of Lyman-break galaxies (LBGs) with redshifts 3 < z < 5 are expected to be observed by the upcoming Vera C. Rubin Observatory Legacy Survey of Space and Time (LSST). This will enable us to probe the universe at higher redshifts than is currently possible with cosmological galaxy clustering and weak lensing surveys. However, accurate inference of cosmological parameters requires precise knowledge of the redshift distributions of selected galaxies, where the number of faint objects expected from LSST alone will make spectroscopic based methods of determining these distributions extremely challenging. To overcome this difficulty, it may be possible to leverage the information in the large volume of photometric data alone to precisely infer these distributions. This could be facilitated using forward models, where in this paper we use stellar population synthesis (SPS) to estimate uncertainties on LBG redshift distributions for a 10 year LSST (LSSTY10) survey. We characterise some of the modelling uncertainties inherent to SPS by introducing a flexible parameterisation of the galaxy population prior, informed by observations of the galaxy stellar mass function (GSMF) and cosmic star formation density (CSFRD). These uncertainties are subsequently marginalised over and propagated to cosmological constraints in a Fisher forecast. Assuming a known dust attenuation model for LBGs, we forecast constraints on the sigma8 parameter comparable to Planck cosmic microwave background (CMB) constraints.

Figures (14)

• Figure 1: Gaussian process models of the redshift dependence of the Schechter function parameters using measurements from santini2022 and nc2024. The errors on these data have been made symmetric by taking the average of the upper and lower error bound. Where parameters have been fixed in nc2024, highlighted here in red, we have estimated the uncertainties by using the uncertainties quoted for the nearest redshift bin to avoid over-fitting. The solid line shows the Gaussian process mean, with the shaded area showing two standard deviations above and below the mean.
• Figure 2: Galaxy stellar mass functions sampled by the Gaussian process model inside a number of redshift bins. The solid lines show the mean, with shaded regions representing the 16-84th percentiles. These are compared with Schechter function fits in santini2022 (dashed lines), weaver2023 (dotted lines) and nc2024 (dashed-dotted lines).
• Figure 3: Distribution of CSFRD models sampled from the Gaussian process fit to observational measurements of the CSFRD compiled by behroozi2019. The grey solid line show the Gaussian process mean to data shown by the black circles, while the purple solid line is the Gaussian process mean and after correcting for systematics. The shaded areas showing two standard deviations above and below the mean. The dashed lines show the inferred fits from behroozi2019. The error bars have been made symmetric by taking the average of the upper and lower bound in behroozi2019.
• Figure 4: Prior on $u$ (purple), $g$ (black) and $r$ (red) normalised LBG dropout redshift distributions. The solid lines indicate the mean, with the shaded regions (dark to light) are the 16-84th, 2.5-97.5th and 0.3-99.7th percentiles respectively, showing the variation in the redshift distribution caused by samples of galaxies draw from different realisations of the galaxy population model.
• Figure 5: Colours of $u$, $g$ and $r$ dropout LBGs (top to bottom) selected by the cuts defined in Section \ref{['sec:selec']}, from an initial sample of 4,000,000 galaxies drawn from the mean of our population model. The dashed black lines show the LBG colour cuts described in the text.