Clustering of emission line galaxies with IllustrisTNG I.: fundamental properties and halo occupation distribution

TL;DR

This study uses hydrodynamical simulations to analyze the clustering and halo occupation distribution of emission line galaxies, highlighting the importance of mock catalogues for accurate modeling of galaxy-halo connections.

Contribution

It provides a detailed analysis of ELG clustering and HOD using IllustrisTNG simulations, emphasizing the need for mock catalogues to improve HOD parameter inference.

Findings

HOD of ELGs shows a distinct population of low-mass infalling halos.

Inferred HODs from projected correlation functions can significantly deviate from simulation measurements.

Mock ELG catalogues are essential for calibrating HOD models and improving parameter constraints.

Abstract

Upcoming spectroscopic redshift surveys use emission line galaxies (ELGs) to trace the three-dimensional matter distributions with wider area coverage in the deeper Universe. Since the halos hosting ELGs are young and undergo infall towards more massive halos along filamentary structures, contrary to a widely employed luminous red galaxy sample, the dynamics specific to ELGs should be taken into account to refine the theoretical modelling at non-linear scales. In this paper, we scrutinise the halo occupation distribution (HOD) and clustering properties of ELGs by utilising IllustrisTNG galaxy formation hydrodynamical simulations. Leveraging stellar population synthesis technique coupled with the photo-ionization model, we compute line intensities of simulated galaxies and construct mock H$\alpha$ and [OII] ELG catalogues. The line luminosity functions and the relation between the star formation rate and line intensity are well consistent with observational estimates. Next, we measure the HOD and demonstrate that there is a distinct population for the central HOD, which corresponds to low-mass infalling halos. We then perform the statistical inference of HOD parameters from the projected correlation function. Our analysis indicates that the inferred HODs significantly deviate from the HOD measured directly from simulations although the best-fit model yields a good fit to the projected correlation function. It implies that the information content of the projected correlation function is not adequate to constrain HOD models correctly and thus, it is important to employ mock ELG catalogues to calibrate the functional form of HOD models and add prior information on HOD parameters to robustly determine the HOD.