Illustrating galaxy-halo connection in the DESI era with IllustrisTNG

TL;DR

This paper uses the IllustrisTNG simulation to analyze the galaxy-halo connection for DESI target samples, revealing complexities like galaxy assembly bias and informing future modeling and mock galaxy creation.

Contribution

It provides new insights into the galaxy-halo connection, highlighting deviations from common assumptions and emphasizing the importance of secondary biases for DESI analyses.

Findings

HOD of LRG and ELG matches baseline models but shows deviations.

Strong evidence for galaxy assembly bias affecting both samples.

Clustering analysis indicates a 10-15% excess due to secondary biases.

Abstract

We employ the hydrodynamical simulation IllustrisTNG to inform the galaxy-halo connection of the Luminous Red Galaxy (LRG) and Emission Line Galaxy (ELG) samples of the Dark Energy Spectroscopic Instrument (DESI) survey at redshift z ~ 0.8. Specifically, we model the galaxy colors of IllustrisTNG and apply sliding DESI color-magnitude cuts, matching the DESI target densities. We study the halo occupation distribution model (HOD) of the selected samples by matching them to their corresponding dark matter halos in the IllustrisTNG dark matter run. We find the HOD of both the LRG and ELG samples to be consistent with their respective baseline models, but also we find important deviations from common assumptions about the satellite distribution, velocity bias, and galaxy secondary biases. We identify strong evidence for concentration-based and environment-based occupational variance in both samples, an effect known as "galaxy assembly bias". The central and satellite galaxies have distinct dependencies on secondary halo properties, showing that centrals and satellites have distinct evolutionary trajectories and should be modelled separately. These results serve to inform the necessary complexities in modeling galaxy-halo connection for DESI analyses and also prepare for building high-fidelity mock galaxies. Finally, we present a shuffling-based clustering analysis that reveals a 10-15% excess in the LRG clustering of modest statistical significance due to secondary galaxy biases. We also find a similar excess signature for the ELGs, but with much lower statistical significance. When a larger hydrodynamical simulation volume becomes available, we expect our analysis pipeline to pinpoint the exact sources of such excess clustering signatures.