Maarten Baes

Simulating the multi-wavelength �Tully-Fisher relation for cosmological simulations

PHISCC 2025 meeting, Cagliari, 22-26 September 2025

Vogelsberger et al. 2019

Cosmological galaxy formation simulations

Comparison of simulated and observed galaxies

Numerical simulations

• predict the 3D distribution of particles/cells with associated properties
• easy to calculate physical galaxy properties (stellar mass, SFR…)

Astronomical observations

• measure 2D distribution of light on plane of the sky
• yield fluxes, images, SEDs, polarization maps…

Should we compare simulated and observed galaxies in the physical or the observational domain ?

Forward modelling

Comparison of observed and simulated galaxies in the observational domain. �Requires the production of realistic mock observations

​

Can in principle account for

• any star-dust geometry (fully 3D)
• any star formation history
• any metallicity distribution
• …

​

Allows for apples-to-apples comparison, �including morphology, polarization, �velocity structure, etc.

​

Gebek et al. 2024

Baes et al. 2019

HI as a test of galaxy formation models

HI is an excellent probe:

• independent: not used for the calibration of the simulation sub-grid physics
• plenty of new data available, even for z > 0

Pan et al. 2023

Rajohnson et al. 2022

Multi-wavelength TFR

Verheijen et al. 2001

Most simulation studies look at the stellar mass or baryonic TFR

• easier: stellar and gas masses are readily available for simulations
• multi-wavelength TFR is interesting: different slopes at different wavelengths !

Kourkchi et al. 2020

Simulating the multi-wavelength TFR

Requires calculation of

• optical/NIR fluxes (including dust attenuation)
• HI profiles (including HI-H₂ partitioning, and measurement of line width)

www.skirt.ugent.be

Baes et al. 2003, 2011�Camps & Baes 2015, 2020�Gebek et al. 2023

Sample selection

Large sample of simulated TNG50 galaxies

• disc-dominated, star-forming galaxies, inclination > 45 deg
• properties agree well with observational TFR samples

Optical images and HI profiles

TNG50 multiwavelength TFR

TFR slope and tightness

TFR becomes steeper and tighter from UV to MIR

​

Agrees well with observational TFR studies

​

Value of slope and tightness depend sensitively on details of sample selection…

TFR slope and tightness

Search for a second parameter

Scatter in the TFR

• is it purely intrinsic ?
• can it be reduced by second parameter (and which one) ?

Best option for second parameter: sSFR or u – r colour

Eliminates the scatter of TFR almost completely, except for a wavelength-independent intrinsic level

Conclusions and future work

The solid agreement boosts the confidence in the TNG50 galaxy formation model and the SKIRT postprocessing methodology

Avenues to make comparison to observations more stringent:

• determine velocity scale from HI data cube modelling (v_flat)
• determine inclination from optical images
• better match sample selection

Future work

• test dependence on simulation resolution and physical ingredients
• apply similar methodology to other galaxy formation models

Schaye et al. subnitted

Maarten Baes

Simulating the multi-wavelength �Tully-Fisher relation for cosmological simulations

PHISCC 2025 meeting, Cagliari, 22-26 September 2025