Towards precise galaxy evolution: a comparison between spectral indices of $z\sim1$ galaxies in the IllustrisTNG simulation and the LEGA-C survey

TL;DR

This study compares observed high-redshift galaxy spectra with simulated spectra from the IllustrisTNG simulation, finding broad agreement but also highlighting small differences that inform galaxy evolution models.

Contribution

First direct comparison of observed and simulated galaxy spectral indices at z~1, assessing the realism of galaxy age distributions in simulations.

Findings

TNG100 broadly reproduces observed spectral index distributions.

Small discrepancies suggest differences in stellar populations or model uncertainties.

Results help refine galaxy formation and stellar population models.

Abstract

We present the first comparison of observed stellar continuum spectra of high-redshift galaxies and mock galaxy spectra generated from hydrodynamical simulations. The mock spectra are produced from the IllustrisTNG TNG100 simulation combined with stellar population models and take into account dust attenuation and realistic observational effects (aperture effects and noise). We compare the simulated $D_n4000$ and EW(H$\delta$) of galaxies with $10.5 \leq \log(M_\ast/M_\odot) \leq 11.5$ at $0.6 \leq z \leq 1.0$ to the observed distributions from the LEGA-C survey. TNG100 globally reproduces the observed distributions of spectral indices, implying that the age distribution of galaxies in TNG100 is generally realistic. Yet there are small but significant differences. For old galaxies, TNG100 shows small $D_n4000$ when compared to LEGA-C, while LEGA-C galaxies have larger EW(H$\delta$) at fixed $D_n4000$. There are several possible explanations: 1) LEGA-C galaxies have overall older ages combined with small contributions (a few percent in mass) from younger ($<1$~Gyr) stars, while TNG100 galaxies may not have such young sub-populations; 2) the spectral mismatch could be due to systematic uncertainties in the stellar population models used to convert stellar ages and metallicities to observables. In conclusion, the latest cosmological galaxy formation simulations broadly reproduce the global age distribution of galaxies at $z\sim1$ and, at the same time, the high quality of the latest observed and simulated datasets help constrain stellar population synthesis models as well as the physical models underlying the simulations.