The FLAMINGO project: cosmological hydrodynamical simulations for large-scale structure and galaxy cluster surveys

TL;DR

FLAMINGO addresses the challenge of incorporating baryonic physics into predictions for large-scale structure by building a large-volume, multi-resolution suite of cosmological hydrodynamical simulations calibrated to low-redshift SMF and cluster gas fractions via Gaussian process emulators. The project combines Swift-based gravity+hydrodynamics, a delta-f neutrino treatment, and sophisticated subgrid models, with on-the-fly lightcones and multiple cosmologies, to produce converged predictions for galaxy and cluster observables as well as baryonic effects on the halo mass function and matter power spectrum. Key contributions include the emulator-driven calibration framework, the exploration of model variations (including jet-like AGN feedback and neutrino masses), and demonstrations that baryonic physics can suppress the halo mass function and matter power spectrum by significant margins, with implications for cosmological parameter inference from upcoming surveys. Overall, FLAMINGO provides a powerful, publicly relevant resource for advancing precision cosmology and the interpretation of LSS data in the presence of complex baryonic processes.

Abstract

We introduce the Virgo Consortium's FLAMINGO suite of hydrodynamical simulations for cosmology and galaxy cluster physics. To ensure the simulations are sufficiently realistic for studies of large-scale structure, the subgrid prescriptions for stellar and AGN feedback are calibrated to the observed low-redshift galaxy stellar mass function and cluster gas fractions. The calibration is performed using machine learning, separately for three resolutions. This approach enables specification of the model by the observables to which they are calibrated. The calibration accounts for a number of potential observational biases and for random errors in the observed stellar masses. The two most demanding simulations have box sizes of 1.0 and 2.8 Gpc and baryonic particle masses of $1\times10^8$ and $1\times10^9 \text{M}_\odot$, respectively. For the latter resolution the suite includes 12 model variations in a 1 Gpc box. There are 8 variations at fixed cosmology, including shifts in the stellar mass function and/or the cluster gas fractions to which we calibrate, and two alternative implementations of AGN feedback (thermal or jets). The remaining 4 variations use the unmodified calibration data but different cosmologies, including different neutrino masses. The 2.8 Gpc simulation follows $3\times10^{11}$ particles, making it the largest ever hydrodynamical simulation run to $z=0$. Lightcone output is produced on-the-fly for up to 8 different observers. We investigate numerical convergence, show that the simulations reproduce the calibration data, and compare with a number of galaxy, cluster, and large-scale structure observations, finding very good agreement with the data for converged predictions. Finally, by comparing hydrodynamical and `dark-matter-only' simulations, we confirm that baryonic effects can suppress the halo mass function and the matter power spectrum by up to $\approx20$ per cent.

Figures (23)

• Figure 1: A projection through a 40 Mpc thick slice through the fiducial, intermediate-resolution simulation with box side length 2.8 Gpc (run L2p8_m9 in Table \ref{['tab:simulations']}) at $z=0$. The luminosity of the background image gives the CDM surface density whilst the colour encodes the surface density of massive neutrinos, both on a logarithmic scale (see Fig. \ref{['fig:cdm_nu_slice']} for a side-by-side comparison of images of the CDM and neutrino surface densities with color bars for each). The insets show three consecutive zooms centred on the most massive halo (total mass $M_\text{200m} = 6.7\times 10^{15}\,\text{M}_\odot$). First inset: a projection of a $200\times 200 \times 40~\text{Mpc}^3$ sub-volume containing the cluster, showing the mass-weighted gas temperature along the line of sight. Second inset: CDM surface density in a $40\times 40\times 20~\text{Mpc}^3$ region. Final inset: X-ray surface brightness in the 0.5-2 keV band in a $20\times 20\times 20~\text{Mpc}^3$ region, computed from the $z=0$ snapshot but placing the cluster at $z=0.025$.
• Figure 2: Comparison of the resolutions (baryonic particle mass or target cell mass; resolution increases along the y-axis) and box sizes of the FLAMINGO runs (filled red circles) with cosmological, hydrodynamical simulations from the literature that include radiative cooling, use a box size of at least $(100~\text{Mpc})^3$, and were run down to $z=0$. The grey diagonal lines indicate the total number of baryonic resolution elements. The simulations from the literature shown are BAHAMAS McCarthy2017, cosmo-OWLS LeBrun2014, EAGLE Schaye2015, Horizon-AGN Dubois2014, IllustrisTNG Springel2018, Magneticum Dolag2016, MassiveBlack-II Khandai2015, MillenniumTNG Pakmor2022, OWLS Schaye2010, SIMBA Dave2019, and SLOW Dolag2023.
• Figure 3: The temperature of gas at the cosmic mean density as a function of redshift. The peaks at $z\approx 7$ and 3 are due to H and He reionization, respectively. The thermal evolution is in good agreement with observations of the Ly$\alpha$ forest. Data points are based on measurements of absorption line widths as a function of strength Schaye2000Rorai2018Hiss2018Telikova2019Gaikwad2020, on the small-scale cut-off in the flux power spectrum Boera2019Walther2019, or on both types of methods Gaikwad2021.
• Figure 4: Comparison of the CDM (left panel) and neutrino (right panel) surface density in a 20 Mpc thick slice through the $z=0$ snapshot of the L2p8_m9 simulation. Note that the dynamic range covered by the color bar is much smaller in the right panel. On scales of $\lesssim 10^2$ Mpc the neutrino distribution is much smoother than that of the CDM.
• Figure 5: Comparison of the gas (left panel) and CDM (right panel) surface density in a 50x50x20 Mpc slice through the $z=0$ snapshot of the L1_m8 simulation. The insets zoom in on a halo of total mass $M_\text{200c} = 1.26\times 10^{14}\,\text{M}_\odot$. Note that the color scale is identical for the two panels. On scales of $\lesssim 1$ Mpc the gas distribution is much smoother than that of the CDM.