Comparison of numerical simulations
and observations

Different models for galaxy evolution: why numerical simulations?

We develop and analyze different theoretical models to tackle challenges in galaxy formation and evolution. Galaxies are complex systems, driving by process on a wide range of scales: from star formation and stellar evolution on small scales to cosmological gas accretion and the cosmic web on large scales. Interesting insights into galaxy evolution can already been gained from simple, analytical models (see here for an example of early galaxy evolution). As physicist, we like ab-initio models, where we describe the system of interest with mathematical equations, which can then be solved. However, in the world of galaxies, we have to approximate the physical laws with numerical models (discrete equations), which then need to be implemented as galaxy simulations. The dynamic range in galaxy evolution is huge and is therefore a big challenge in galaxy simulations: there is always a resolution limitation, these scales are unfortunately important (for example supernovae explosions). It is therefore important to ensure that the scientific questions can be answered with the given resolution of the simulations.

We focus on the following aspects of theoretical modeling:

  • interpretation of our observations: constraining the underlying physics and modeling of biases

  • predictions and planning for future surveys (JWST, Euclid, ELT, SKA, …)

  • exploring the plausibility of theoretical ideas

  • link galaxy properties to dark matter halos (➜ cosmology)

Predictions from IllustrisTNG

We collaborate with members of the IllustrisTNG team to make predictions for the up-coming James Webb Space Telescope (see right). We have projected the simulation quantities into the observational space, which allows robust predictions and detailed "apple-to-apple" comparisons. In particular, we have explored:

  • the dust modeling and the predicted luminosity functions (Vogelsberger et al. 2020);

  • galaxy line and continuum spectral indices and dust attenuation curves (Shen et al. 2020);

  • infrared luminosity functions, obscured star formation and dust temperature of high-redshift galaxies (Shen et al. 2020).

In addition, we have used similar approaches to model lower-redshift galaxies and do careful comparisons of the morphology of local galaxies (Rodriguez-Gomez et al. 2019).

The H-alpha emission as a star-formation rate tracer

We are currently working on a framework to predict emission lines (such as the Balmer emission line H-alpha; Smith, Kannan, Tacchella et al. 2022) from numerical simulations. After developing and applying this framework on isolated disk simulations (see below), we will apply it to high-redshift simulations such as THESAN to make predictions and study observational biases of star formation and metal content of high redshift galaxies.

In Tacchella et al. (2022), we focus on the Halpha emission line and study how well this emission line traces the instantaneous star-formation rate of and within galaxies. We find that the dust absorption of Lyman continuum photons is important on the 30% level, while the escape of Lyman continuum photons and emission from collisionally ionized gas contribute only on the 5% level and roughly cancel each other out. Furthermore, the Halpha-based star-formation rate tracer typically traces the star-formation rate on a timescale of 5-10 Myr, depending on the star-formation burstiness as well as amount of attenuation.